diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,162624 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 27099,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4600246002460026e-08,
+      "loss": 1.2306,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.920049200492005e-08,
+      "loss": 1.3239,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.380073800738008e-08,
+      "loss": 1.3582,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.84009840098401e-08,
+      "loss": 1.3686,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2300123001230013e-07,
+      "loss": 1.2511,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4760147601476016e-07,
+      "loss": 1.3655,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7220172201722017e-07,
+      "loss": 1.2636,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.968019680196802e-07,
+      "loss": 1.309,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2140221402214022e-07,
+      "loss": 1.2773,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4600246002460025e-07,
+      "loss": 1.2235,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.706027060270603e-07,
+      "loss": 1.2245,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9520295202952033e-07,
+      "loss": 1.278,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.198031980319803e-07,
+      "loss": 1.3163,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4440344403444035e-07,
+      "loss": 1.2859,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.690036900369004e-07,
+      "loss": 1.2609,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.936039360393604e-07,
+      "loss": 1.2717,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1820418204182045e-07,
+      "loss": 1.2438,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4280442804428044e-07,
+      "loss": 1.234,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6740467404674047e-07,
+      "loss": 1.1862,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.920049200492005e-07,
+      "loss": 1.2415,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.166051660516606e-07,
+      "loss": 1.2013,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.412054120541206e-07,
+      "loss": 1.2113,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.658056580565806e-07,
+      "loss": 1.2147,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.904059040590407e-07,
+      "loss": 1.2321,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.150061500615006e-07,
+      "loss": 1.1848,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.396063960639606e-07,
+      "loss": 1.1822,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.642066420664207e-07,
+      "loss": 1.106,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.888068880688807e-07,
+      "loss": 1.0863,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.134071340713408e-07,
+      "loss": 1.0782,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.380073800738008e-07,
+      "loss": 1.0254,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.626076260762607e-07,
+      "loss": 1.076,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.872078720787208e-07,
+      "loss": 1.0635,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.118081180811808e-07,
+      "loss": 1.0512,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.364083640836409e-07,
+      "loss": 1.0325,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.610086100861009e-07,
+      "loss": 1.1096,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.856088560885609e-07,
+      "loss": 1.0348,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.10209102091021e-07,
+      "loss": 0.937,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.348093480934809e-07,
+      "loss": 0.9716,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.59409594095941e-07,
+      "loss": 0.8902,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.84009840098401e-07,
+      "loss": 0.8521,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.008610086100861e-06,
+      "loss": 0.8676,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0332103321033212e-06,
+      "loss": 0.8388,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0578105781057812e-06,
+      "loss": 0.856,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0824108241082412e-06,
+      "loss": 0.8354,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1070110701107011e-06,
+      "loss": 0.8294,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1316113161131611e-06,
+      "loss": 0.8612,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1562115621156213e-06,
+      "loss": 0.8157,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1808118081180813e-06,
+      "loss": 0.8109,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2054120541205413e-06,
+      "loss": 0.7981,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2300123001230013e-06,
+      "loss": 0.8023,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2546125461254613e-06,
+      "loss": 0.7919,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2792127921279212e-06,
+      "loss": 0.7119,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3038130381303814e-06,
+      "loss": 0.7596,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3284132841328414e-06,
+      "loss": 0.749,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3530135301353014e-06,
+      "loss": 0.7392,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3776137761377614e-06,
+      "loss": 0.7359,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4022140221402214e-06,
+      "loss": 0.6906,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4268142681426816e-06,
+      "loss": 0.6638,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4514145141451415e-06,
+      "loss": 0.6622,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4760147601476015e-06,
+      "loss": 0.7077,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5006150061500615e-06,
+      "loss": 0.6379,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5252152521525215e-06,
+      "loss": 0.6816,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5498154981549817e-06,
+      "loss": 0.6622,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5744157441574417e-06,
+      "loss": 0.646,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5990159901599017e-06,
+      "loss": 0.635,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6236162361623616e-06,
+      "loss": 0.6518,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6482164821648216e-06,
+      "loss": 0.6564,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6728167281672818e-06,
+      "loss": 0.6466,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6974169741697418e-06,
+      "loss": 0.6606,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7220172201722018e-06,
+      "loss": 0.582,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7466174661746618e-06,
+      "loss": 0.6045,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7712177121771217e-06,
+      "loss": 0.629,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.795817958179582e-06,
+      "loss": 0.5979,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.820418204182042e-06,
+      "loss": 0.5896,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.845018450184502e-06,
+      "loss": 0.6459,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8696186961869619e-06,
+      "loss": 0.6061,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8942189421894219e-06,
+      "loss": 0.526,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.918819188191882e-06,
+      "loss": 0.5718,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9434194341943423e-06,
+      "loss": 0.6389,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.968019680196802e-06,
+      "loss": 0.6073,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9926199261992622e-06,
+      "loss": 0.5713,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.017220172201722e-06,
+      "loss": 0.5851,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.041820418204182e-06,
+      "loss": 0.5664,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0664206642066424e-06,
+      "loss": 0.595,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.091020910209102e-06,
+      "loss": 0.5759,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1156211562115624e-06,
+      "loss": 0.5711,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.140221402214022e-06,
+      "loss": 0.5311,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1648216482164823e-06,
+      "loss": 0.5582,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1894218942189425e-06,
+      "loss": 0.507,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2140221402214023e-06,
+      "loss": 0.5155,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2386223862238625e-06,
+      "loss": 0.5251,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2632226322263223e-06,
+      "loss": 0.5452,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2878228782287825e-06,
+      "loss": 0.5674,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3124231242312426e-06,
+      "loss": 0.4962,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3370233702337024e-06,
+      "loss": 0.4697,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3616236162361626e-06,
+      "loss": 0.569,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3862238622386224e-06,
+      "loss": 0.5201,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4108241082410826e-06,
+      "loss": 0.5201,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4354243542435428e-06,
+      "loss": 0.5559,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4600246002460025e-06,
+      "loss": 0.5133,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4846248462484627e-06,
+      "loss": 0.4754,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5092250922509225e-06,
+      "loss": 0.525,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5338253382533827e-06,
+      "loss": 0.4881,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5584255842558425e-06,
+      "loss": 0.4657,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5830258302583027e-06,
+      "loss": 0.5025,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.607626076260763e-06,
+      "loss": 0.5238,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6322263222632226e-06,
+      "loss": 0.5296,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.656826568265683e-06,
+      "loss": 0.5089,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6814268142681426e-06,
+      "loss": 0.4726,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.706027060270603e-06,
+      "loss": 0.5432,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.730627306273063e-06,
+      "loss": 0.5175,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7552275522755228e-06,
+      "loss": 0.4737,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.779827798277983e-06,
+      "loss": 0.4432,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8044280442804427e-06,
+      "loss": 0.5154,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.829028290282903e-06,
+      "loss": 0.4767,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.853628536285363e-06,
+      "loss": 0.533,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.878228782287823e-06,
+      "loss": 0.4753,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.902829028290283e-06,
+      "loss": 0.4903,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.927429274292743e-06,
+      "loss": 0.5203,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.952029520295203e-06,
+      "loss": 0.4797,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9766297662976633e-06,
+      "loss": 0.4687,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.001230012300123e-06,
+      "loss": 0.4877,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0258302583025832e-06,
+      "loss": 0.5155,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.050430504305043e-06,
+      "loss": 0.4732,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.075030750307503e-06,
+      "loss": 0.4544,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0996309963099634e-06,
+      "loss": 0.5009,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.124231242312423e-06,
+      "loss": 0.4766,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1488314883148833e-06,
+      "loss": 0.4976,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.173431734317343e-06,
+      "loss": 0.4424,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1980319803198033e-06,
+      "loss": 0.4841,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2226322263222635e-06,
+      "loss": 0.4804,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2472324723247233e-06,
+      "loss": 0.4678,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2718327183271835e-06,
+      "loss": 0.4728,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2964329643296432e-06,
+      "loss": 0.4458,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3210332103321034e-06,
+      "loss": 0.4488,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3456334563345636e-06,
+      "loss": 0.4551,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3702337023370234e-06,
+      "loss": 0.449,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3948339483394836e-06,
+      "loss": 0.464,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4194341943419434e-06,
+      "loss": 0.4689,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4440344403444036e-06,
+      "loss": 0.4898,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4686346863468638e-06,
+      "loss": 0.4481,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4932349323493235e-06,
+      "loss": 0.4383,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5178351783517837e-06,
+      "loss": 0.5045,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5424354243542435e-06,
+      "loss": 0.4698,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5670356703567037e-06,
+      "loss": 0.5015,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.591635916359164e-06,
+      "loss": 0.4645,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6162361623616237e-06,
+      "loss": 0.4819,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.640836408364084e-06,
+      "loss": 0.4622,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6654366543665436e-06,
+      "loss": 0.4577,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.690036900369004e-06,
+      "loss": 0.3974,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.714637146371464e-06,
+      "loss": 0.4854,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7392373923739238e-06,
+      "loss": 0.4804,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.763837638376384e-06,
+      "loss": 0.5034,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7884378843788438e-06,
+      "loss": 0.4663,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.813038130381304e-06,
+      "loss": 0.4783,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.837638376383764e-06,
+      "loss": 0.4269,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.862238622386224e-06,
+      "loss": 0.4198,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8868388683886845e-06,
+      "loss": 0.4756,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911439114391144e-06,
+      "loss": 0.4692,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.936039360393604e-06,
+      "loss": 0.4584,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.960639606396065e-06,
+      "loss": 0.4732,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9852398523985245e-06,
+      "loss": 0.422,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.009840098400984e-06,
+      "loss": 0.4607,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.034440344403444e-06,
+      "loss": 0.4338,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.059040590405905e-06,
+      "loss": 0.4865,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.083640836408364e-06,
+      "loss": 0.4431,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.108241082410824e-06,
+      "loss": 0.4638,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.132841328413285e-06,
+      "loss": 0.4559,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1574415744157446e-06,
+      "loss": 0.4817,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.182041820418204e-06,
+      "loss": 0.46,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.206642066420665e-06,
+      "loss": 0.4383,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.231242312423125e-06,
+      "loss": 0.4848,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2558425584255845e-06,
+      "loss": 0.4562,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.280442804428044e-06,
+      "loss": 0.4471,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.305043050430505e-06,
+      "loss": 0.428,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.329643296432965e-06,
+      "loss": 0.4192,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.354243542435424e-06,
+      "loss": 0.4501,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.378843788437885e-06,
+      "loss": 0.432,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.403444034440345e-06,
+      "loss": 0.4391,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.428044280442805e-06,
+      "loss": 0.4999,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.452644526445265e-06,
+      "loss": 0.457,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.477244772447725e-06,
+      "loss": 0.4688,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.501845018450185e-06,
+      "loss": 0.4521,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5264452644526445e-06,
+      "loss": 0.4545,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.551045510455105e-06,
+      "loss": 0.5063,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.575645756457565e-06,
+      "loss": 0.4629,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600246002460025e-06,
+      "loss": 0.4368,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.624846248462485e-06,
+      "loss": 0.3931,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.649446494464945e-06,
+      "loss": 0.386,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674046740467405e-06,
+      "loss": 0.4501,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6986469864698655e-06,
+      "loss": 0.417,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.723247232472325e-06,
+      "loss": 0.473,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.747847478474785e-06,
+      "loss": 0.4481,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.772447724477245e-06,
+      "loss": 0.4501,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.797047970479705e-06,
+      "loss": 0.4247,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.821648216482165e-06,
+      "loss": 0.4022,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.846248462484625e-06,
+      "loss": 0.4439,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8708487084870856e-06,
+      "loss": 0.4482,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.895448954489545e-06,
+      "loss": 0.4405,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920049200492005e-06,
+      "loss": 0.435,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.944649446494466e-06,
+      "loss": 0.3982,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9692496924969255e-06,
+      "loss": 0.4049,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993849938499385e-06,
+      "loss": 0.4548,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.018450184501845e-06,
+      "loss": 0.4398,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.043050430504306e-06,
+      "loss": 0.3968,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.067650676506765e-06,
+      "loss": 0.4483,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.092250922509225e-06,
+      "loss": 0.4895,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.116851168511685e-06,
+      "loss": 0.4269,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.141451414514146e-06,
+      "loss": 0.392,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.166051660516605e-06,
+      "loss": 0.4382,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.190651906519065e-06,
+      "loss": 0.4104,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.215252152521526e-06,
+      "loss": 0.4257,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2398523985239855e-06,
+      "loss": 0.4343,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.264452644526445e-06,
+      "loss": 0.4443,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.289052890528906e-06,
+      "loss": 0.4026,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.313653136531366e-06,
+      "loss": 0.4521,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3382533825338254e-06,
+      "loss": 0.4107,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.362853628536285e-06,
+      "loss": 0.4331,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.387453874538746e-06,
+      "loss": 0.4148,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.412054120541206e-06,
+      "loss": 0.4753,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.436654366543665e-06,
+      "loss": 0.3961,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.461254612546126e-06,
+      "loss": 0.4331,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.485854858548586e-06,
+      "loss": 0.4537,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5104551045510455e-06,
+      "loss": 0.4036,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.535055350553506e-06,
+      "loss": 0.4308,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.559655596555966e-06,
+      "loss": 0.455,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.584255842558426e-06,
+      "loss": 0.4245,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6088560885608855e-06,
+      "loss": 0.4506,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.633456334563346e-06,
+      "loss": 0.4418,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.658056580565806e-06,
+      "loss": 0.431,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.682656826568266e-06,
+      "loss": 0.4559,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.707257072570726e-06,
+      "loss": 0.4259,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.731857318573186e-06,
+      "loss": 0.435,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.756457564575646e-06,
+      "loss": 0.3814,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.781057810578106e-06,
+      "loss": 0.4385,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.805658056580566e-06,
+      "loss": 0.3895,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.830258302583026e-06,
+      "loss": 0.4161,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.854858548585486e-06,
+      "loss": 0.4249,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.879458794587946e-06,
+      "loss": 0.4595,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.904059040590406e-06,
+      "loss": 0.4284,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.928659286592866e-06,
+      "loss": 0.4591,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9532595325953265e-06,
+      "loss": 0.4242,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.977859778597786e-06,
+      "loss": 0.4206,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.002460024600246e-06,
+      "loss": 0.4212,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.027060270602707e-06,
+      "loss": 0.4275,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0516605166051664e-06,
+      "loss": 0.465,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.076260762607626e-06,
+      "loss": 0.4014,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.100861008610086e-06,
+      "loss": 0.3779,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.125461254612547e-06,
+      "loss": 0.4361,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.150061500615006e-06,
+      "loss": 0.429,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.174661746617466e-06,
+      "loss": 0.4011,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.199261992619927e-06,
+      "loss": 0.4064,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.2238622386223865e-06,
+      "loss": 0.4337,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.248462484624846e-06,
+      "loss": 0.3876,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.273062730627307e-06,
+      "loss": 0.3949,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.297662976629767e-06,
+      "loss": 0.4376,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3222632226322265e-06,
+      "loss": 0.4396,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.346863468634686e-06,
+      "loss": 0.42,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.371463714637147e-06,
+      "loss": 0.3708,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.396063960639607e-06,
+      "loss": 0.4022,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.420664206642066e-06,
+      "loss": 0.4055,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.445264452644527e-06,
+      "loss": 0.3895,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.469864698646987e-06,
+      "loss": 0.4015,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4944649446494466e-06,
+      "loss": 0.4555,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.519065190651907e-06,
+      "loss": 0.4054,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.543665436654367e-06,
+      "loss": 0.3868,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.568265682656827e-06,
+      "loss": 0.3927,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5928659286592865e-06,
+      "loss": 0.4334,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.617466174661747e-06,
+      "loss": 0.424,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.642066420664207e-06,
+      "loss": 0.4415,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.4107,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.691266912669127e-06,
+      "loss": 0.4342,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.715867158671587e-06,
+      "loss": 0.412,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.740467404674047e-06,
+      "loss": 0.3979,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7650676506765074e-06,
+      "loss": 0.425,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.789667896678967e-06,
+      "loss": 0.4144,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.814268142681427e-06,
+      "loss": 0.4159,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.838868388683887e-06,
+      "loss": 0.4427,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.863468634686347e-06,
+      "loss": 0.391,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.888068880688807e-06,
+      "loss": 0.4295,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.912669126691267e-06,
+      "loss": 0.4611,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9372693726937275e-06,
+      "loss": 0.3997,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.961869618696187e-06,
+      "loss": 0.4036,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.986469864698647e-06,
+      "loss": 0.418,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.011070110701108e-06,
+      "loss": 0.4283,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.0356703567035675e-06,
+      "loss": 0.46,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.060270602706027e-06,
+      "loss": 0.3563,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.084870848708487e-06,
+      "loss": 0.4031,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.109471094710948e-06,
+      "loss": 0.3575,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.134071340713407e-06,
+      "loss": 0.4264,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.158671586715867e-06,
+      "loss": 0.4158,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.183271832718328e-06,
+      "loss": 0.4051,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2078720787207875e-06,
+      "loss": 0.4095,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.232472324723247e-06,
+      "loss": 0.4124,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.257072570725709e-06,
+      "loss": 0.3904,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.281672816728168e-06,
+      "loss": 0.3783,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3062730627306275e-06,
+      "loss": 0.3972,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.330873308733087e-06,
+      "loss": 0.3922,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.355473554735548e-06,
+      "loss": 0.4521,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.380073800738008e-06,
+      "loss": 0.4136,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.404674046740467e-06,
+      "loss": 0.4224,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.429274292742928e-06,
+      "loss": 0.4592,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.453874538745388e-06,
+      "loss": 0.3995,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4784747847478476e-06,
+      "loss": 0.4099,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.503075030750309e-06,
+      "loss": 0.4477,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.527675276752768e-06,
+      "loss": 0.4162,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.552275522755228e-06,
+      "loss": 0.4421,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5768757687576875e-06,
+      "loss": 0.3993,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.601476014760148e-06,
+      "loss": 0.4153,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.626076260762608e-06,
+      "loss": 0.4075,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.650676506765068e-06,
+      "loss": 0.4409,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.675276752767528e-06,
+      "loss": 0.4174,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.699876998769989e-06,
+      "loss": 0.4211,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.724477244772448e-06,
+      "loss": 0.4128,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.749077490774908e-06,
+      "loss": 0.4033,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.773677736777369e-06,
+      "loss": 0.4184,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.798277982779828e-06,
+      "loss": 0.4351,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.822878228782289e-06,
+      "loss": 0.3827,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.84747847478475e-06,
+      "loss": 0.4014,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.872078720787208e-06,
+      "loss": 0.4087,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.896678966789669e-06,
+      "loss": 0.4053,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.92127921279213e-06,
+      "loss": 0.395,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.945879458794588e-06,
+      "loss": 0.39,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.970479704797049e-06,
+      "loss": 0.4145,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.99507995079951e-06,
+      "loss": 0.4128,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.019680196801968e-06,
+      "loss": 0.4242,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.044280442804429e-06,
+      "loss": 0.424,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.068880688806888e-06,
+      "loss": 0.4139,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.093480934809349e-06,
+      "loss": 0.4016,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.11808118081181e-06,
+      "loss": 0.4348,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.142681426814268e-06,
+      "loss": 0.3718,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.167281672816729e-06,
+      "loss": 0.4019,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.19188191881919e-06,
+      "loss": 0.4501,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.216482164821648e-06,
+      "loss": 0.4055,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.241082410824109e-06,
+      "loss": 0.4385,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.26568265682657e-06,
+      "loss": 0.3937,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.290282902829028e-06,
+      "loss": 0.4223,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.314883148831489e-06,
+      "loss": 0.3858,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.33948339483395e-06,
+      "loss": 0.4094,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.364083640836409e-06,
+      "loss": 0.3703,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.38868388683887e-06,
+      "loss": 0.4119,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.41328413284133e-06,
+      "loss": 0.4018,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.437884378843789e-06,
+      "loss": 0.422,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.46248462484625e-06,
+      "loss": 0.4249,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.48708487084871e-06,
+      "loss": 0.398,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.511685116851169e-06,
+      "loss": 0.4335,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.53628536285363e-06,
+      "loss": 0.4043,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.560885608856089e-06,
+      "loss": 0.4065,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.585485854858549e-06,
+      "loss": 0.3897,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.61008610086101e-06,
+      "loss": 0.3943,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.634686346863469e-06,
+      "loss": 0.4581,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.65928659286593e-06,
+      "loss": 0.422,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.68388683886839e-06,
+      "loss": 0.4616,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.708487084870849e-06,
+      "loss": 0.3956,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.73308733087331e-06,
+      "loss": 0.4023,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.75768757687577e-06,
+      "loss": 0.3903,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.782287822878229e-06,
+      "loss": 0.4387,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.80688806888069e-06,
+      "loss": 0.377,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.83148831488315e-06,
+      "loss": 0.4311,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.85608856088561e-06,
+      "loss": 0.3961,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.88068880688807e-06,
+      "loss": 0.4065,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.90528905289053e-06,
+      "loss": 0.3987,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92988929889299e-06,
+      "loss": 0.3789,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.95448954489545e-06,
+      "loss": 0.3887,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.97908979089791e-06,
+      "loss": 0.4126,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.00369003690037e-06,
+      "loss": 0.4213,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.02829028290283e-06,
+      "loss": 0.4235,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.052890528905289e-06,
+      "loss": 0.4311,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.07749077490775e-06,
+      "loss": 0.393,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.10209102091021e-06,
+      "loss": 0.4057,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.12669126691267e-06,
+      "loss": 0.4104,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.15129151291513e-06,
+      "loss": 0.4072,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.17589175891759e-06,
+      "loss": 0.3858,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.20049200492005e-06,
+      "loss": 0.3701,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.22509225092251e-06,
+      "loss": 0.3773,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.24969249692497e-06,
+      "loss": 0.3793,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.27429274292743e-06,
+      "loss": 0.4328,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.29889298892989e-06,
+      "loss": 0.4059,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.32349323493235e-06,
+      "loss": 0.4264,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.34809348093481e-06,
+      "loss": 0.364,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.37269372693727e-06,
+      "loss": 0.4027,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.397293972939731e-06,
+      "loss": 0.4265,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.42189421894219e-06,
+      "loss": 0.375,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.44649446494465e-06,
+      "loss": 0.4365,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.471094710947111e-06,
+      "loss": 0.4006,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.49569495694957e-06,
+      "loss": 0.4384,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.52029520295203e-06,
+      "loss": 0.3777,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.54489544895449e-06,
+      "loss": 0.3863,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.56949569495695e-06,
+      "loss": 0.4038,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.59409594095941e-06,
+      "loss": 0.4186,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.61869618696187e-06,
+      "loss": 0.3865,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.64329643296433e-06,
+      "loss": 0.4262,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.667896678966791e-06,
+      "loss": 0.3942,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.69249692496925e-06,
+      "loss": 0.3914,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.71709717097171e-06,
+      "loss": 0.3866,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.741697416974171e-06,
+      "loss": 0.3979,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.76629766297663e-06,
+      "loss": 0.4181,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.79089790897909e-06,
+      "loss": 0.3778,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.815498154981551e-06,
+      "loss": 0.4316,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.84009840098401e-06,
+      "loss": 0.4111,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86469864698647e-06,
+      "loss": 0.4169,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.889298892988931e-06,
+      "loss": 0.3723,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.91389913899139e-06,
+      "loss": 0.3848,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.938499384993851e-06,
+      "loss": 0.3834,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963099630996312e-06,
+      "loss": 0.4085,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98769987699877e-06,
+      "loss": 0.3902,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0012300123001231e-05,
+      "loss": 0.4324,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.003690036900369e-05,
+      "loss": 0.3882,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.006150061500615e-05,
+      "loss": 0.3905,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0086100861008611e-05,
+      "loss": 0.4115,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0110701107011072e-05,
+      "loss": 0.4465,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.013530135301353e-05,
+      "loss": 0.3924,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0159901599015991e-05,
+      "loss": 0.4131,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.018450184501845e-05,
+      "loss": 0.4062,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0209102091020911e-05,
+      "loss": 0.4062,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.023370233702337e-05,
+      "loss": 0.4056,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0258302583025832e-05,
+      "loss": 0.3706,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0282902829028291e-05,
+      "loss": 0.4184,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0307503075030752e-05,
+      "loss": 0.3872,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.033210332103321e-05,
+      "loss": 0.4046,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0356703567035671e-05,
+      "loss": 0.3739,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.038130381303813e-05,
+      "loss": 0.4044,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0405904059040591e-05,
+      "loss": 0.4166,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0430504305043051e-05,
+      "loss": 0.3567,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0455104551045512e-05,
+      "loss": 0.4111,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0479704797047971e-05,
+      "loss": 0.3784,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0504305043050432e-05,
+      "loss": 0.3859,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.052890528905289e-05,
+      "loss": 0.4398,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0553505535055351e-05,
+      "loss": 0.4118,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0578105781057812e-05,
+      "loss": 0.4146,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0602706027060272e-05,
+      "loss": 0.4124,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0627306273062731e-05,
+      "loss": 0.416,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0651906519065192e-05,
+      "loss": 0.4217,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0676506765067651e-05,
+      "loss": 0.4391,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0701107011070112e-05,
+      "loss": 0.3745,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.072570725707257e-05,
+      "loss": 0.4161,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0750307503075033e-05,
+      "loss": 0.3891,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0774907749077492e-05,
+      "loss": 0.3861,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0799507995079952e-05,
+      "loss": 0.4232,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0824108241082411e-05,
+      "loss": 0.412,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0848708487084872e-05,
+      "loss": 0.3915,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.087330873308733e-05,
+      "loss": 0.4087,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0897908979089791e-05,
+      "loss": 0.4252,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0922509225092252e-05,
+      "loss": 0.411,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0947109471094713e-05,
+      "loss": 0.4205,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0971709717097172e-05,
+      "loss": 0.4397,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0996309963099632e-05,
+      "loss": 0.3746,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1020910209102091e-05,
+      "loss": 0.4048,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1045510455104552e-05,
+      "loss": 0.3723,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1070110701107012e-05,
+      "loss": 0.3841,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1094710947109473e-05,
+      "loss": 0.4064,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1119311193111932e-05,
+      "loss": 0.36,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1143911439114392e-05,
+      "loss": 0.4066,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1168511685116851e-05,
+      "loss": 0.3641,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1193111931119312e-05,
+      "loss": 0.3998,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1217712177121771e-05,
+      "loss": 0.3782,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1242312423124233e-05,
+      "loss": 0.3698,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1266912669126692e-05,
+      "loss": 0.4181,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1291512915129153e-05,
+      "loss": 0.4155,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1316113161131612e-05,
+      "loss": 0.4007,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1340713407134072e-05,
+      "loss": 0.3743,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1365313653136531e-05,
+      "loss": 0.4406,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1389913899138992e-05,
+      "loss": 0.4274,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1414514145141452e-05,
+      "loss": 0.3844,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1439114391143913e-05,
+      "loss": 0.3868,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1463714637146372e-05,
+      "loss": 0.3959,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1488314883148833e-05,
+      "loss": 0.3724,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1512915129151292e-05,
+      "loss": 0.3842,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1537515375153752e-05,
+      "loss": 0.3673,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1562115621156213e-05,
+      "loss": 0.3777,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1586715867158673e-05,
+      "loss": 0.398,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1611316113161132e-05,
+      "loss": 0.3785,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1635916359163593e-05,
+      "loss": 0.393,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1660516605166052e-05,
+      "loss": 0.3931,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1685116851168513e-05,
+      "loss": 0.4114,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1709717097170971e-05,
+      "loss": 0.3633,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1734317343173434e-05,
+      "loss": 0.4117,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1758917589175893e-05,
+      "loss": 0.4119,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1783517835178353e-05,
+      "loss": 0.392,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1808118081180812e-05,
+      "loss": 0.3771,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1832718327183273e-05,
+      "loss": 0.4088,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1857318573185732e-05,
+      "loss": 0.3505,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1881918819188192e-05,
+      "loss": 0.3913,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1906519065190653e-05,
+      "loss": 0.3909,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1931119311193114e-05,
+      "loss": 0.3809,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1955719557195573e-05,
+      "loss": 0.4134,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1980319803198033e-05,
+      "loss": 0.3384,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2004920049200492e-05,
+      "loss": 0.3751,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2029520295202953e-05,
+      "loss": 0.3957,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2054120541205413e-05,
+      "loss": 0.3591,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2078720787207874e-05,
+      "loss": 0.3816,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2103321033210333e-05,
+      "loss": 0.3876,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2127921279212793e-05,
+      "loss": 0.3829,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2152521525215252e-05,
+      "loss": 0.378,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2177121771217713e-05,
+      "loss": 0.4089,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2201722017220172e-05,
+      "loss": 0.3913,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2226322263222634e-05,
+      "loss": 0.4004,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2250922509225093e-05,
+      "loss": 0.3638,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2275522755227554e-05,
+      "loss": 0.4301,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2300123001230013e-05,
+      "loss": 0.3839,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2324723247232473e-05,
+      "loss": 0.3696,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2349323493234932e-05,
+      "loss": 0.3823,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2373923739237393e-05,
+      "loss": 0.4228,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2398523985239854e-05,
+      "loss": 0.3976,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2423124231242314e-05,
+      "loss": 0.3958,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2447724477244773e-05,
+      "loss": 0.3874,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2472324723247234e-05,
+      "loss": 0.4579,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2496924969249693e-05,
+      "loss": 0.3636,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2521525215252153e-05,
+      "loss": 0.4093,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2546125461254614e-05,
+      "loss": 0.3927,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2570725707257074e-05,
+      "loss": 0.4595,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2595325953259533e-05,
+      "loss": 0.4163,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2619926199261994e-05,
+      "loss": 0.3798,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2644526445264453e-05,
+      "loss": 0.3723,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2669126691266914e-05,
+      "loss": 0.4445,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2693726937269372e-05,
+      "loss": 0.4135,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2718327183271835e-05,
+      "loss": 0.3946,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2742927429274294e-05,
+      "loss": 0.3972,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2767527675276754e-05,
+      "loss": 0.3735,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2792127921279213e-05,
+      "loss": 0.3784,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2816728167281674e-05,
+      "loss": 0.3593,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2841328413284133e-05,
+      "loss": 0.4083,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2865928659286593e-05,
+      "loss": 0.403,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2890528905289054e-05,
+      "loss": 0.4442,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2915129151291515e-05,
+      "loss": 0.3933,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2939729397293974e-05,
+      "loss": 0.3941,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2964329643296434e-05,
+      "loss": 0.3603,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2988929889298893e-05,
+      "loss": 0.4017,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3013530135301354e-05,
+      "loss": 0.3728,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3038130381303814e-05,
+      "loss": 0.3762,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3062730627306275e-05,
+      "loss": 0.411,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3087330873308734e-05,
+      "loss": 0.381,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3111931119311195e-05,
+      "loss": 0.4324,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3136531365313653e-05,
+      "loss": 0.4047,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3161131611316114e-05,
+      "loss": 0.3966,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3185731857318573e-05,
+      "loss": 0.3787,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3210332103321035e-05,
+      "loss": 0.4103,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3234932349323494e-05,
+      "loss": 0.3772,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3259532595325955e-05,
+      "loss": 0.4079,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3284132841328414e-05,
+      "loss": 0.3993,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3308733087330874e-05,
+      "loss": 0.3756,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.3828,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3357933579335794e-05,
+      "loss": 0.3361,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3382533825338255e-05,
+      "loss": 0.3562,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3407134071340715e-05,
+      "loss": 0.4162,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3431734317343174e-05,
+      "loss": 0.3927,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3456334563345635e-05,
+      "loss": 0.3587,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3480934809348094e-05,
+      "loss": 0.3882,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3505535055350554e-05,
+      "loss": 0.3814,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3530135301353015e-05,
+      "loss": 0.3699,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3554735547355475e-05,
+      "loss": 0.4044,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3579335793357934e-05,
+      "loss": 0.3662,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3603936039360395e-05,
+      "loss": 0.3714,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3628536285362854e-05,
+      "loss": 0.4033,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3653136531365315e-05,
+      "loss": 0.3514,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3677736777367773e-05,
+      "loss": 0.3677,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3702337023370236e-05,
+      "loss": 0.4178,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3726937269372695e-05,
+      "loss": 0.392,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3751537515375155e-05,
+      "loss": 0.4263,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3776137761377614e-05,
+      "loss": 0.4126,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3800738007380075e-05,
+      "loss": 0.388,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3825338253382534e-05,
+      "loss": 0.4015,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3849938499384994e-05,
+      "loss": 0.3839,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3874538745387455e-05,
+      "loss": 0.4052,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3899138991389916e-05,
+      "loss": 0.381,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3923739237392375e-05,
+      "loss": 0.4263,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3948339483394835e-05,
+      "loss": 0.3685,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3972939729397294e-05,
+      "loss": 0.4495,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3997539975399755e-05,
+      "loss": 0.3987,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4022140221402215e-05,
+      "loss": 0.3785,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4046740467404676e-05,
+      "loss": 0.4038,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4071340713407135e-05,
+      "loss": 0.4103,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4095940959409596e-05,
+      "loss": 0.3527,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4120541205412054e-05,
+      "loss": 0.3792,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4145141451414515e-05,
+      "loss": 0.4177,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4169741697416974e-05,
+      "loss": 0.3986,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4194341943419436e-05,
+      "loss": 0.4028,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4218942189421895e-05,
+      "loss": 0.361,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4243542435424356e-05,
+      "loss": 0.3878,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4268142681426815e-05,
+      "loss": 0.3834,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4292742927429275e-05,
+      "loss": 0.3699,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4317343173431734e-05,
+      "loss": 0.3781,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4341943419434195e-05,
+      "loss": 0.3988,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4366543665436656e-05,
+      "loss": 0.3346,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4391143911439116e-05,
+      "loss": 0.4069,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4415744157441575e-05,
+      "loss": 0.382,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4440344403444036e-05,
+      "loss": 0.4306,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4464944649446495e-05,
+      "loss": 0.3949,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4489544895448955e-05,
+      "loss": 0.3754,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4514145141451418e-05,
+      "loss": 0.4092,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4538745387453876e-05,
+      "loss": 0.3842,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4563345633456335e-05,
+      "loss": 0.3809,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4587945879458796e-05,
+      "loss": 0.397,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4612546125461255e-05,
+      "loss": 0.3637,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4637146371463716e-05,
+      "loss": 0.373,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4661746617466175e-05,
+      "loss": 0.3787,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4686346863468637e-05,
+      "loss": 0.4255,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4710947109471096e-05,
+      "loss": 0.3797,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4735547355473556e-05,
+      "loss": 0.3754,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4760147601476015e-05,
+      "loss": 0.3944,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4784747847478476e-05,
+      "loss": 0.4043,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4809348093480935e-05,
+      "loss": 0.4206,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4833948339483395e-05,
+      "loss": 0.4174,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4858548585485856e-05,
+      "loss": 0.3778,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4883148831488317e-05,
+      "loss": 0.3993,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4907749077490776e-05,
+      "loss": 0.3966,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4932349323493236e-05,
+      "loss": 0.4652,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4956949569495695e-05,
+      "loss": 0.3988,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4981549815498156e-05,
+      "loss": 0.4127,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5006150061500618e-05,
+      "loss": 0.4096,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5030750307503077e-05,
+      "loss": 0.3863,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5055350553505536e-05,
+      "loss": 0.3863,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5079950799507997e-05,
+      "loss": 0.3849,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5104551045510455e-05,
+      "loss": 0.3696,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5129151291512916e-05,
+      "loss": 0.3784,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5153751537515375e-05,
+      "loss": 0.3936,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5178351783517837e-05,
+      "loss": 0.3676,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5202952029520296e-05,
+      "loss": 0.3898,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5227552275522757e-05,
+      "loss": 0.3942,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5252152521525216e-05,
+      "loss": 0.3695,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5276752767527675e-05,
+      "loss": 0.4,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5301353013530135e-05,
+      "loss": 0.383,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5325953259532596e-05,
+      "loss": 0.3296,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5350553505535057e-05,
+      "loss": 0.3991,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5375153751537517e-05,
+      "loss": 0.391,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5399753997539978e-05,
+      "loss": 0.3708,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5424354243542435e-05,
+      "loss": 0.3763,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5448954489544896e-05,
+      "loss": 0.3727,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5473554735547356e-05,
+      "loss": 0.3872,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5498154981549817e-05,
+      "loss": 0.3828,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5522755227552278e-05,
+      "loss": 0.4063,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5547355473554738e-05,
+      "loss": 0.3866,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5571955719557195e-05,
+      "loss": 0.385,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5596555965559656e-05,
+      "loss": 0.3771,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5621156211562117e-05,
+      "loss": 0.3703,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5645756457564577e-05,
+      "loss": 0.3396,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5670356703567038e-05,
+      "loss": 0.4397,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.56949569495695e-05,
+      "loss": 0.381,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5719557195571956e-05,
+      "loss": 0.3931,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5744157441574416e-05,
+      "loss": 0.4033,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5768757687576877e-05,
+      "loss": 0.396,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5793357933579338e-05,
+      "loss": 0.3843,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5817958179581795e-05,
+      "loss": 0.4006,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.584255842558426e-05,
+      "loss": 0.359,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5867158671586716e-05,
+      "loss": 0.3999,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5891758917589177e-05,
+      "loss": 0.3767,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5916359163591637e-05,
+      "loss": 0.3455,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5940959409594098e-05,
+      "loss": 0.3722,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5965559655596555e-05,
+      "loss": 0.4276,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.599015990159902e-05,
+      "loss": 0.376,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6014760147601476e-05,
+      "loss": 0.3633,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6039360393603937e-05,
+      "loss": 0.4282,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6063960639606398e-05,
+      "loss": 0.3635,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6088560885608858e-05,
+      "loss": 0.3851,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6113161131611315e-05,
+      "loss": 0.3462,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6137761377613776e-05,
+      "loss": 0.3727,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.616236162361624e-05,
+      "loss": 0.3766,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6186961869618697e-05,
+      "loss": 0.3842,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6211562115621158e-05,
+      "loss": 0.3805,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.623616236162362e-05,
+      "loss": 0.4083,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6260762607626076e-05,
+      "loss": 0.377,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6285362853628536e-05,
+      "loss": 0.3873,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6309963099630997e-05,
+      "loss": 0.3928,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6334563345633458e-05,
+      "loss": 0.3965,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6359163591635918e-05,
+      "loss": 0.3807,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.638376383763838e-05,
+      "loss": 0.3836,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6408364083640836e-05,
+      "loss": 0.4023,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6432964329643297e-05,
+      "loss": 0.363,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6457564575645757e-05,
+      "loss": 0.3721,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6482164821648218e-05,
+      "loss": 0.3809,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.650676506765068e-05,
+      "loss": 0.3438,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.653136531365314e-05,
+      "loss": 0.356,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6555965559655596e-05,
+      "loss": 0.3862,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6580565805658057e-05,
+      "loss": 0.3801,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6605166051660518e-05,
+      "loss": 0.374,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6629766297662978e-05,
+      "loss": 0.3598,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.665436654366544e-05,
+      "loss": 0.4084,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.66789667896679e-05,
+      "loss": 0.3679,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6703567035670357e-05,
+      "loss": 0.3419,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6728167281672817e-05,
+      "loss": 0.4069,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6752767527675278e-05,
+      "loss": 0.4066,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.677736777367774e-05,
+      "loss": 0.3597,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6801968019680196e-05,
+      "loss": 0.367,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.682656826568266e-05,
+      "loss": 0.3849,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6851168511685117e-05,
+      "loss": 0.3725,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6875768757687578e-05,
+      "loss": 0.3929,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6900369003690038e-05,
+      "loss": 0.3884,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.69249692496925e-05,
+      "loss": 0.3865,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6949569495694956e-05,
+      "loss": 0.3749,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.697416974169742e-05,
+      "loss": 0.3807,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.699876998769988e-05,
+      "loss": 0.3441,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7023370233702338e-05,
+      "loss": 0.3501,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.70479704797048e-05,
+      "loss": 0.3749,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.707257072570726e-05,
+      "loss": 0.3836,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7097170971709716e-05,
+      "loss": 0.3901,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7121771217712177e-05,
+      "loss": 0.37,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.714637146371464e-05,
+      "loss": 0.3898,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7170971709717098e-05,
+      "loss": 0.3768,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.719557195571956e-05,
+      "loss": 0.3705,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.722017220172202e-05,
+      "loss": 0.3503,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7244772447724477e-05,
+      "loss": 0.3972,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7269372693726937e-05,
+      "loss": 0.3931,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7293972939729398e-05,
+      "loss": 0.3488,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.731857318573186e-05,
+      "loss": 0.4058,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.734317343173432e-05,
+      "loss": 0.3953,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.736777367773678e-05,
+      "loss": 0.3591,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7392373923739237e-05,
+      "loss": 0.3207,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7416974169741698e-05,
+      "loss": 0.3836,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.744157441574416e-05,
+      "loss": 0.3597,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.746617466174662e-05,
+      "loss": 0.3758,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.749077490774908e-05,
+      "loss": 0.3703,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.751537515375154e-05,
+      "loss": 0.3817,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7539975399753997e-05,
+      "loss": 0.3771,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7564575645756458e-05,
+      "loss": 0.3761,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.758917589175892e-05,
+      "loss": 0.4025,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.761377613776138e-05,
+      "loss": 0.3811,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.763837638376384e-05,
+      "loss": 0.3977,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.76629766297663e-05,
+      "loss": 0.4035,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7687576875768758e-05,
+      "loss": 0.3945,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.771217712177122e-05,
+      "loss": 0.3463,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.773677736777368e-05,
+      "loss": 0.4033,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.776137761377614e-05,
+      "loss": 0.4097,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7785977859778597e-05,
+      "loss": 0.3964,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.781057810578106e-05,
+      "loss": 0.3698,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.783517835178352e-05,
+      "loss": 0.4023,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785977859778598e-05,
+      "loss": 0.3881,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.788437884378844e-05,
+      "loss": 0.3678,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.79089790897909e-05,
+      "loss": 0.3442,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7933579335793357e-05,
+      "loss": 0.3786,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.795817958179582e-05,
+      "loss": 0.3572,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7982779827798282e-05,
+      "loss": 0.3387,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.800738007380074e-05,
+      "loss": 0.3953,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.80319803198032e-05,
+      "loss": 0.3966,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.805658056580566e-05,
+      "loss": 0.405,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8081180811808117e-05,
+      "loss": 0.3644,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8105781057810578e-05,
+      "loss": 0.3768,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8130381303813042e-05,
+      "loss": 0.3777,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.81549815498155e-05,
+      "loss": 0.3819,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.817958179581796e-05,
+      "loss": 0.3575,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.820418204182042e-05,
+      "loss": 0.36,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8228782287822878e-05,
+      "loss": 0.3505,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.825338253382534e-05,
+      "loss": 0.3915,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.82779827798278e-05,
+      "loss": 0.3719,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.830258302583026e-05,
+      "loss": 0.3667,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.832718327183272e-05,
+      "loss": 0.3797,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.835178351783518e-05,
+      "loss": 0.3693,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8376383763837638e-05,
+      "loss": 0.3727,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.84009840098401e-05,
+      "loss": 0.3612,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842558425584256e-05,
+      "loss": 0.3594,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.845018450184502e-05,
+      "loss": 0.3757,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.847478474784748e-05,
+      "loss": 0.3511,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.849938499384994e-05,
+      "loss": 0.3553,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.85239852398524e-05,
+      "loss": 0.3626,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.854858548585486e-05,
+      "loss": 0.3813,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.857318573185732e-05,
+      "loss": 0.4025,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.859778597785978e-05,
+      "loss": 0.3755,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.862238622386224e-05,
+      "loss": 0.358,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.86469864698647e-05,
+      "loss": 0.3832,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.867158671586716e-05,
+      "loss": 0.3388,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.869618696186962e-05,
+      "loss": 0.3574,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.872078720787208e-05,
+      "loss": 0.4351,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.874538745387454e-05,
+      "loss": 0.3605,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8769987699876998e-05,
+      "loss": 0.3832,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8794587945879462e-05,
+      "loss": 0.3599,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8819188191881922e-05,
+      "loss": 0.3564,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.884378843788438e-05,
+      "loss": 0.3731,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.886838868388684e-05,
+      "loss": 0.3775,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.88929889298893e-05,
+      "loss": 0.3912,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8917589175891758e-05,
+      "loss": 0.3319,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8942189421894222e-05,
+      "loss": 0.3593,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8966789667896683e-05,
+      "loss": 0.366,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.899138991389914e-05,
+      "loss": 0.3666,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.90159901599016e-05,
+      "loss": 0.363,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.904059040590406e-05,
+      "loss": 0.4032,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.906519065190652e-05,
+      "loss": 0.3845,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.908979089790898e-05,
+      "loss": 0.3587,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9114391143911443e-05,
+      "loss": 0.3813,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.91389913899139e-05,
+      "loss": 0.368,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.916359163591636e-05,
+      "loss": 0.3563,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.918819188191882e-05,
+      "loss": 0.365,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.921279212792128e-05,
+      "loss": 0.3767,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.923739237392374e-05,
+      "loss": 0.3646,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.92619926199262e-05,
+      "loss": 0.3674,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.928659286592866e-05,
+      "loss": 0.3644,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.931119311193112e-05,
+      "loss": 0.379,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9335793357933582e-05,
+      "loss": 0.3844,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.936039360393604e-05,
+      "loss": 0.3642,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.93849938499385e-05,
+      "loss": 0.3834,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.940959409594096e-05,
+      "loss": 0.4054,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.943419434194342e-05,
+      "loss": 0.3764,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945879458794588e-05,
+      "loss": 0.3365,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9483394833948342e-05,
+      "loss": 0.3979,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95079950799508e-05,
+      "loss": 0.3729,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.953259532595326e-05,
+      "loss": 0.3874,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.955719557195572e-05,
+      "loss": 0.3945,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958179581795818e-05,
+      "loss": 0.3829,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9606396063960642e-05,
+      "loss": 0.3425,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630996309963103e-05,
+      "loss": 0.4078,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655596555965563e-05,
+      "loss": 0.413,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968019680196802e-05,
+      "loss": 0.3996,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970479704797048e-05,
+      "loss": 0.3632,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.972939729397294e-05,
+      "loss": 0.4282,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97539975399754e-05,
+      "loss": 0.388,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9778597785977863e-05,
+      "loss": 0.3376,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803198031980323e-05,
+      "loss": 0.3706,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982779827798278e-05,
+      "loss": 0.348,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985239852398524e-05,
+      "loss": 0.3911,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9876998769987702e-05,
+      "loss": 0.3846,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990159901599016e-05,
+      "loss": 0.3943,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9926199261992623e-05,
+      "loss": 0.3987,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9950799507995084e-05,
+      "loss": 0.3738,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997539975399754e-05,
+      "loss": 0.3997,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3736,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999928579854e-05,
+      "loss": 0.3803,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999714319414e-05,
+      "loss": 0.3219,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999935721868e-05,
+      "loss": 0.377,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999885727766e-05,
+      "loss": 0.3711,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999821449637e-05,
+      "loss": 0.3914,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997428874804e-05,
+      "loss": 0.3711,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999996500412978e-05,
+      "loss": 0.3733,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999542911091e-05,
+      "loss": 0.376,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994214968616e-05,
+      "loss": 0.3798,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999992857986106e-05,
+      "loss": 0.4124,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999991358163403e-05,
+      "loss": 0.4054,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999989715500526e-05,
+      "loss": 0.3951,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999987929997506e-05,
+      "loss": 0.3566,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999986001654364e-05,
+      "loss": 0.3774,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999983930471126e-05,
+      "loss": 0.4036,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998171644782e-05,
+      "loss": 0.3719,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999979359584484e-05,
+      "loss": 0.3451,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999976859881145e-05,
+      "loss": 0.3692,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999974217337847e-05,
+      "loss": 0.4123,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999971431954615e-05,
+      "loss": 0.3763,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999685037315e-05,
+      "loss": 0.3665,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999965432668542e-05,
+      "loss": 0.3463,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999962218765782e-05,
+      "loss": 0.4133,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999958862023265e-05,
+      "loss": 0.3549,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999955362441043e-05,
+      "loss": 0.3374,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999951720019162e-05,
+      "loss": 0.3735,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999947934757676e-05,
+      "loss": 0.3877,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999994400665664e-05,
+      "loss": 0.3757,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993993571611e-05,
+      "loss": 0.3553,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993572193614e-05,
+      "loss": 0.3717,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999931365316795e-05,
+      "loss": 0.3675,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999926865858135e-05,
+      "loss": 0.3879,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999922223560226e-05,
+      "loss": 0.3747,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999917438423133e-05,
+      "loss": 0.3784,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999912510446924e-05,
+      "loss": 0.404,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999907439631672e-05,
+      "loss": 0.3778,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999902225977443e-05,
+      "loss": 0.3457,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999989686948432e-05,
+      "loss": 0.3986,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999891370152375e-05,
+      "loss": 0.387,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999885727981684e-05,
+      "loss": 0.3536,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999879942972335e-05,
+      "loss": 0.3897,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999874015124404e-05,
+      "loss": 0.3525,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999867944437973e-05,
+      "loss": 0.3592,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999986173091314e-05,
+      "loss": 0.3787,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999855374549985e-05,
+      "loss": 0.3561,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99998488753486e-05,
+      "loss": 0.3635,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999842233309082e-05,
+      "loss": 0.3522,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999835448431518e-05,
+      "loss": 0.3313,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999828520716013e-05,
+      "loss": 0.3838,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999821450162665e-05,
+      "loss": 0.4169,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999814236771566e-05,
+      "loss": 0.3656,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999980688054283e-05,
+      "loss": 0.367,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999979938147656e-05,
+      "loss": 0.3409,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999791739572855e-05,
+      "loss": 0.3546,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999783954831833e-05,
+      "loss": 0.3778,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999776027253602e-05,
+      "loss": 0.3319,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999767956838278e-05,
+      "loss": 0.3997,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999975974358597e-05,
+      "loss": 0.3927,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99997513874968e-05,
+      "loss": 0.3511,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999742888570887e-05,
+      "loss": 0.3601,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999734246808352e-05,
+      "loss": 0.3895,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999725462209317e-05,
+      "loss": 0.3605,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999971653477391e-05,
+      "loss": 0.3337,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999707464502257e-05,
+      "loss": 0.394,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999698251394492e-05,
+      "loss": 0.3694,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999688895450734e-05,
+      "loss": 0.3596,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999967939667113e-05,
+      "loss": 0.3696,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999966975505581e-05,
+      "loss": 0.3701,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999659970604912e-05,
+      "loss": 0.3817,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999650043318577e-05,
+      "loss": 0.3531,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999639973196947e-05,
+      "loss": 0.337,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999629760240165e-05,
+      "loss": 0.3918,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999619404448372e-05,
+      "loss": 0.3763,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999608905821726e-05,
+      "loss": 0.3568,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959826436037e-05,
+      "loss": 0.3927,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999587480064456e-05,
+      "loss": 0.3702,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999957655293414e-05,
+      "loss": 0.3274,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999956548296958e-05,
+      "loss": 0.353,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999554270170928e-05,
+      "loss": 0.3859,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999954291453835e-05,
+      "loss": 0.4002,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999531416072008e-05,
+      "loss": 0.3741,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999519774772068e-05,
+      "loss": 0.3437,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999507990638686e-05,
+      "loss": 0.4078,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999949606367204e-05,
+      "loss": 0.3443,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999483993872296e-05,
+      "loss": 0.3639,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999471781239627e-05,
+      "loss": 0.3624,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999945942577421e-05,
+      "loss": 0.3495,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999944692747622e-05,
+      "loss": 0.3569,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999434286345837e-05,
+      "loss": 0.367,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999421502383235e-05,
+      "loss": 0.3548,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999408575588607e-05,
+      "loss": 0.3941,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999939550596213e-05,
+      "loss": 0.3655,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999382293503992e-05,
+      "loss": 0.3927,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999368938214383e-05,
+      "loss": 0.3581,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999355440093492e-05,
+      "loss": 0.3564,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999341799141514e-05,
+      "loss": 0.3895,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999328015358642e-05,
+      "loss": 0.365,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999931408874508e-05,
+      "loss": 0.393,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999300019301015e-05,
+      "loss": 0.3569,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999285807026655e-05,
+      "loss": 0.358,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999271451922202e-05,
+      "loss": 0.3605,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999925695398786e-05,
+      "loss": 0.3646,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999242313223837e-05,
+      "loss": 0.3713,
+      "step": 916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999227529630343e-05,
+      "loss": 0.3604,
+      "step": 917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999212603207587e-05,
+      "loss": 0.3564,
+      "step": 918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999197533955785e-05,
+      "loss": 0.3681,
+      "step": 919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999918232187515e-05,
+      "loss": 0.335,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99991669669659e-05,
+      "loss": 0.3716,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999151469228252e-05,
+      "loss": 0.3476,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999913582866243e-05,
+      "loss": 0.3923,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999120045268664e-05,
+      "loss": 0.3477,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999104119047164e-05,
+      "loss": 0.3536,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999088049998173e-05,
+      "loss": 0.371,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999071838121907e-05,
+      "loss": 0.3484,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999055483418606e-05,
+      "loss": 0.3947,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999038985888505e-05,
+      "loss": 0.3747,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999022345531834e-05,
+      "loss": 0.3717,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999005562348836e-05,
+      "loss": 0.3012,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998988636339743e-05,
+      "loss": 0.3812,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998971567504804e-05,
+      "loss": 0.3667,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999895435584426e-05,
+      "loss": 0.3235,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998937001358357e-05,
+      "loss": 0.3619,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998919504047346e-05,
+      "loss": 0.3152,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999890186391147e-05,
+      "loss": 0.3665,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998884080950985e-05,
+      "loss": 0.3531,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999886615516615e-05,
+      "loss": 0.3554,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999884808655721e-05,
+      "loss": 0.3801,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999882987512443e-05,
+      "loss": 0.3282,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998811520868073e-05,
+      "loss": 0.36,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998793023788395e-05,
+      "loss": 0.3527,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998774383885663e-05,
+      "loss": 0.3812,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998755601160146e-05,
+      "loss": 0.3579,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998736675612104e-05,
+      "loss": 0.3746,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998717607241818e-05,
+      "loss": 0.3769,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999869839604955e-05,
+      "loss": 0.3487,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998679042035583e-05,
+      "loss": 0.3736,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998659545200187e-05,
+      "loss": 0.3595,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998639905543644e-05,
+      "loss": 0.3565,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998620123066237e-05,
+      "loss": 0.3385,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998600197768243e-05,
+      "loss": 0.3866,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998580129649947e-05,
+      "loss": 0.3548,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998559918711643e-05,
+      "loss": 0.3938,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999853956495361e-05,
+      "loss": 0.3504,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998519068376142e-05,
+      "loss": 0.3632,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998498428979538e-05,
+      "loss": 0.3957,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998477646764084e-05,
+      "loss": 0.3353,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999845672173008e-05,
+      "loss": 0.3818,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998435653877828e-05,
+      "loss": 0.3605,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998414443207625e-05,
+      "loss": 0.3554,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998393089719775e-05,
+      "loss": 0.3228,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998371593414585e-05,
+      "loss": 0.3794,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999834995429236e-05,
+      "loss": 0.3533,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998328172353408e-05,
+      "loss": 0.3775,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999830624759804e-05,
+      "loss": 0.3324,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998284180026578e-05,
+      "loss": 0.3456,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998261969639324e-05,
+      "loss": 0.3888,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998239616436604e-05,
+      "loss": 0.3644,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998217120418735e-05,
+      "loss": 0.3578,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998194481586035e-05,
+      "loss": 0.3541,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998171699938834e-05,
+      "loss": 0.3906,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999814877547745e-05,
+      "loss": 0.3911,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998125708202214e-05,
+      "loss": 0.3501,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999810249811346e-05,
+      "loss": 0.3712,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999807914521151e-05,
+      "loss": 0.376,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998055649496706e-05,
+      "loss": 0.3918,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999803201096938e-05,
+      "loss": 0.352,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999800822962987e-05,
+      "loss": 0.3723,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997984305478516e-05,
+      "loss": 0.4074,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997960238515656e-05,
+      "loss": 0.3493,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997936028741643e-05,
+      "loss": 0.3628,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997911676156814e-05,
+      "loss": 0.3618,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997887180761518e-05,
+      "loss": 0.3506,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999786254255611e-05,
+      "loss": 0.348,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997837761540938e-05,
+      "loss": 0.3388,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997812837716357e-05,
+      "loss": 0.3904,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997787771082723e-05,
+      "loss": 0.3654,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999776256164039e-05,
+      "loss": 0.3082,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997737209389724e-05,
+      "loss": 0.3412,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997711714331087e-05,
+      "loss": 0.3543,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999768607646484e-05,
+      "loss": 0.4138,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999766029579135e-05,
+      "loss": 0.3526,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997634372310987e-05,
+      "loss": 0.3656,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999760830602412e-05,
+      "loss": 0.3407,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997582096931117e-05,
+      "loss": 0.361,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997555745032362e-05,
+      "loss": 0.379,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997529250328223e-05,
+      "loss": 0.3592,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997502612819083e-05,
+      "loss": 0.3458,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999747583250532e-05,
+      "loss": 0.3965,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999744890938732e-05,
+      "loss": 0.3465,
+      "step": 1002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999742184346546e-05,
+      "loss": 0.3462,
+      "step": 1003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997394634740134e-05,
+      "loss": 0.3815,
+      "step": 1004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999736728321173e-05,
+      "loss": 0.3392,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997339788880638e-05,
+      "loss": 0.3756,
+      "step": 1006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997312151747245e-05,
+      "loss": 0.4038,
+      "step": 1007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997284371811955e-05,
+      "loss": 0.3611,
+      "step": 1008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997256449075156e-05,
+      "loss": 0.3373,
+      "step": 1009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997228383537256e-05,
+      "loss": 0.3485,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997200175198647e-05,
+      "loss": 0.3378,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999717182405974e-05,
+      "loss": 0.3576,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997143330120933e-05,
+      "loss": 0.373,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997114693382638e-05,
+      "loss": 0.3421,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999708591384526e-05,
+      "loss": 0.3835,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997056991509214e-05,
+      "loss": 0.3416,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999702792637491e-05,
+      "loss": 0.3864,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996998718442764e-05,
+      "loss": 0.33,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996969367713195e-05,
+      "loss": 0.3678,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999693987418662e-05,
+      "loss": 0.3622,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999691023786346e-05,
+      "loss": 0.3463,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999688045874414e-05,
+      "loss": 0.3118,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999685053682909e-05,
+      "loss": 0.3525,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996820472118725e-05,
+      "loss": 0.3662,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996790264613483e-05,
+      "loss": 0.3563,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996759914313797e-05,
+      "loss": 0.3687,
+      "step": 1026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996729421220097e-05,
+      "loss": 0.3525,
+      "step": 1027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996698785332817e-05,
+      "loss": 0.3783,
+      "step": 1028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99966680066524e-05,
+      "loss": 0.3643,
+      "step": 1029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999663708517928e-05,
+      "loss": 0.3245,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99966060209139e-05,
+      "loss": 0.3743,
+      "step": 1031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999657481385671e-05,
+      "loss": 0.3706,
+      "step": 1032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999654346400815e-05,
+      "loss": 0.3194,
+      "step": 1033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996511971368667e-05,
+      "loss": 0.3866,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999648033593871e-05,
+      "loss": 0.372,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996448557718736e-05,
+      "loss": 0.3826,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996416636709196e-05,
+      "loss": 0.3676,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996384572910544e-05,
+      "loss": 0.3509,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996352366323243e-05,
+      "loss": 0.3478,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999632001694775e-05,
+      "loss": 0.3382,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996287524784522e-05,
+      "loss": 0.3648,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999625488983403e-05,
+      "loss": 0.327,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996222112096744e-05,
+      "loss": 0.3506,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999618919157312e-05,
+      "loss": 0.4015,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999615612826364e-05,
+      "loss": 0.3687,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999612292216877e-05,
+      "loss": 0.3593,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996089573288985e-05,
+      "loss": 0.3459,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996056081624763e-05,
+      "loss": 0.3298,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996022447176577e-05,
+      "loss": 0.3596,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995988669944918e-05,
+      "loss": 0.3484,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995954749930256e-05,
+      "loss": 0.3592,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995920687133087e-05,
+      "loss": 0.3437,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995886481553888e-05,
+      "loss": 0.3953,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995852133193154e-05,
+      "loss": 0.3305,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995817642051373e-05,
+      "loss": 0.3468,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995783008129036e-05,
+      "loss": 0.3542,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995748231426644e-05,
+      "loss": 0.3448,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999571331194469e-05,
+      "loss": 0.3448,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995678249683668e-05,
+      "loss": 0.3553,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995643044644088e-05,
+      "loss": 0.3663,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995607696826445e-05,
+      "loss": 0.3237,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995572206231247e-05,
+      "loss": 0.3466,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995536572859e-05,
+      "loss": 0.3605,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995500796710217e-05,
+      "loss": 0.3322,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995464877785403e-05,
+      "loss": 0.3703,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995428816085076e-05,
+      "loss": 0.3703,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995392611609748e-05,
+      "loss": 0.3256,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999535626435994e-05,
+      "loss": 0.3373,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995319774336165e-05,
+      "loss": 0.3574,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999528314153895e-05,
+      "loss": 0.3936,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995246365968815e-05,
+      "loss": 0.3811,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995209447626286e-05,
+      "loss": 0.321,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995172386511893e-05,
+      "loss": 0.3314,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995135182626158e-05,
+      "loss": 0.3655,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995097835969622e-05,
+      "loss": 0.3702,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995060346542814e-05,
+      "loss": 0.3822,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995022714346267e-05,
+      "loss": 0.3404,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999498493938052e-05,
+      "loss": 0.3562,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999494702164612e-05,
+      "loss": 0.337,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994908961143595e-05,
+      "loss": 0.3527,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994870757873495e-05,
+      "loss": 0.3611,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994832411836375e-05,
+      "loss": 0.3066,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994793923032766e-05,
+      "loss": 0.362,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994755291463228e-05,
+      "loss": 0.3233,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994716517128313e-05,
+      "loss": 0.3733,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999467760002857e-05,
+      "loss": 0.3424,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999463854016456e-05,
+      "loss": 0.3384,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994599337536836e-05,
+      "loss": 0.3592,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994559992145962e-05,
+      "loss": 0.3492,
+      "step": 1089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99945205039925e-05,
+      "loss": 0.343,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999448087307701e-05,
+      "loss": 0.3034,
+      "step": 1091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999444109940006e-05,
+      "loss": 0.395,
+      "step": 1092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994401182962222e-05,
+      "loss": 0.3477,
+      "step": 1093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999436112376406e-05,
+      "loss": 0.3342,
+      "step": 1094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999432092180615e-05,
+      "loss": 0.3486,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994280577089064e-05,
+      "loss": 0.3358,
+      "step": 1096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994240089613384e-05,
+      "loss": 0.3322,
+      "step": 1097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999419945937968e-05,
+      "loss": 0.3219,
+      "step": 1098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994158686388535e-05,
+      "loss": 0.3299,
+      "step": 1099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994117770640535e-05,
+      "loss": 0.3839,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994076712136262e-05,
+      "loss": 0.3751,
+      "step": 1101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994035510876305e-05,
+      "loss": 0.3162,
+      "step": 1102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993994166861248e-05,
+      "loss": 0.354,
+      "step": 1103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993952680091682e-05,
+      "loss": 0.3372,
+      "step": 1104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993911050568204e-05,
+      "loss": 0.3655,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993869278291408e-05,
+      "loss": 0.3249,
+      "step": 1106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993827363261882e-05,
+      "loss": 0.3398,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993785305480234e-05,
+      "loss": 0.3604,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993743104947066e-05,
+      "loss": 0.3535,
+      "step": 1109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993700761662975e-05,
+      "loss": 0.3764,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993658275628566e-05,
+      "loss": 0.3778,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993615646844447e-05,
+      "loss": 0.3357,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993572875311228e-05,
+      "loss": 0.3457,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999352996102952e-05,
+      "loss": 0.3893,
+      "step": 1114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993486903999937e-05,
+      "loss": 0.3923,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993443704223092e-05,
+      "loss": 0.3588,
+      "step": 1116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993400361699603e-05,
+      "loss": 0.382,
+      "step": 1117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993356876430087e-05,
+      "loss": 0.3868,
+      "step": 1118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993313248415166e-05,
+      "loss": 0.342,
+      "step": 1119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993269477655464e-05,
+      "loss": 0.3521,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999322556415161e-05,
+      "loss": 0.3407,
+      "step": 1121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993181507904223e-05,
+      "loss": 0.3574,
+      "step": 1122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999313730891394e-05,
+      "loss": 0.3308,
+      "step": 1123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993092967181388e-05,
+      "loss": 0.3576,
+      "step": 1124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99930484827072e-05,
+      "loss": 0.3659,
+      "step": 1125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993003855492016e-05,
+      "loss": 0.349,
+      "step": 1126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992959085536468e-05,
+      "loss": 0.3639,
+      "step": 1127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992914172841198e-05,
+      "loss": 0.325,
+      "step": 1128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999286911740685e-05,
+      "loss": 0.398,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992823919234062e-05,
+      "loss": 0.3267,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999277857832348e-05,
+      "loss": 0.3331,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999273309467576e-05,
+      "loss": 0.3328,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992687468291544e-05,
+      "loss": 0.3069,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992641699171486e-05,
+      "loss": 0.3385,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999259578731624e-05,
+      "loss": 0.3439,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992549732726457e-05,
+      "loss": 0.3028,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992503535402804e-05,
+      "loss": 0.3441,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999245719534593e-05,
+      "loss": 0.3919,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999241071255651e-05,
+      "loss": 0.3334,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992364087035197e-05,
+      "loss": 0.3547,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999231731878266e-05,
+      "loss": 0.3485,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992270407799574e-05,
+      "loss": 0.3688,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992223354086593e-05,
+      "loss": 0.371,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992176157644408e-05,
+      "loss": 0.3433,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992128818473677e-05,
+      "loss": 0.3261,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999208133657509e-05,
+      "loss": 0.3498,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992033711949313e-05,
+      "loss": 0.3524,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991985944597038e-05,
+      "loss": 0.3364,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991938034518935e-05,
+      "loss": 0.3444,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991889981715696e-05,
+      "loss": 0.3478,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999184178618801e-05,
+      "loss": 0.3946,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991793447936557e-05,
+      "loss": 0.3825,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991744966962036e-05,
+      "loss": 0.3678,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999169634326513e-05,
+      "loss": 0.3212,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991647576846542e-05,
+      "loss": 0.3548,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991598667706966e-05,
+      "loss": 0.3547,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991549615847102e-05,
+      "loss": 0.3541,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991500421267645e-05,
+      "loss": 0.349,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991451083969303e-05,
+      "loss": 0.3362,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999140160395278e-05,
+      "loss": 0.3678,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991351981218778e-05,
+      "loss": 0.3548,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991302215768015e-05,
+      "loss": 0.3866,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991252307601192e-05,
+      "loss": 0.3278,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991202256719032e-05,
+      "loss": 0.3541,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991152063122238e-05,
+      "loss": 0.3535,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991101726811537e-05,
+      "loss": 0.3834,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991051247787646e-05,
+      "loss": 0.3331,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991000626051283e-05,
+      "loss": 0.3598,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999094986160317e-05,
+      "loss": 0.3488,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990898954444036e-05,
+      "loss": 0.3378,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999084790457461e-05,
+      "loss": 0.3298,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990796711995615e-05,
+      "loss": 0.3631,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990745376707784e-05,
+      "loss": 0.3667,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990693898711856e-05,
+      "loss": 0.3375,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999064227800856e-05,
+      "loss": 0.3534,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990590514598632e-05,
+      "loss": 0.3785,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990538608482816e-05,
+      "loss": 0.3422,
+      "step": 1177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990486559661854e-05,
+      "loss": 0.3867,
+      "step": 1178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990434368136488e-05,
+      "loss": 0.3558,
+      "step": 1179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999038203390746e-05,
+      "loss": 0.3781,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990329556975524e-05,
+      "loss": 0.3844,
+      "step": 1181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990276937341423e-05,
+      "loss": 0.3537,
+      "step": 1182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990224175005914e-05,
+      "loss": 0.3318,
+      "step": 1183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990171269969747e-05,
+      "loss": 0.3757,
+      "step": 1184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990118222233682e-05,
+      "loss": 0.3757,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999006503179847e-05,
+      "loss": 0.3362,
+      "step": 1186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990011698664874e-05,
+      "loss": 0.3475,
+      "step": 1187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998995822283366e-05,
+      "loss": 0.3366,
+      "step": 1188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989904604305584e-05,
+      "loss": 0.4006,
+      "step": 1189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989850843081423e-05,
+      "loss": 0.3192,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998979693916193e-05,
+      "loss": 0.3836,
+      "step": 1191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989742892547888e-05,
+      "loss": 0.3472,
+      "step": 1192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989688703240065e-05,
+      "loss": 0.3651,
+      "step": 1193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989634371239232e-05,
+      "loss": 0.3542,
+      "step": 1194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989579896546165e-05,
+      "loss": 0.3734,
+      "step": 1195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989525279161647e-05,
+      "loss": 0.369,
+      "step": 1196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989470519086456e-05,
+      "loss": 0.359,
+      "step": 1197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989415616321372e-05,
+      "loss": 0.3514,
+      "step": 1198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998936057086718e-05,
+      "loss": 0.3634,
+      "step": 1199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998930538272467e-05,
+      "loss": 0.3721,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989250051894627e-05,
+      "loss": 0.3182,
+      "step": 1201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989194578377842e-05,
+      "loss": 0.377,
+      "step": 1202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989138962175105e-05,
+      "loss": 0.3583,
+      "step": 1203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989083203287214e-05,
+      "loss": 0.376,
+      "step": 1204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989027301714964e-05,
+      "loss": 0.3556,
+      "step": 1205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988971257459153e-05,
+      "loss": 0.3164,
+      "step": 1206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988915070520584e-05,
+      "loss": 0.3307,
+      "step": 1207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988858740900054e-05,
+      "loss": 0.3774,
+      "step": 1208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988802268598376e-05,
+      "loss": 0.3394,
+      "step": 1209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998874565361635e-05,
+      "loss": 0.3454,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988688895954783e-05,
+      "loss": 0.3695,
+      "step": 1211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988631995614495e-05,
+      "loss": 0.3325,
+      "step": 1212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988574952596288e-05,
+      "loss": 0.3319,
+      "step": 1213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988517766900984e-05,
+      "loss": 0.3471,
+      "step": 1214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99884604385294e-05,
+      "loss": 0.3676,
+      "step": 1215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998840296748235e-05,
+      "loss": 0.324,
+      "step": 1216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988345353760657e-05,
+      "loss": 0.3488,
+      "step": 1217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998828759736515e-05,
+      "loss": 0.3616,
+      "step": 1218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988229698296644e-05,
+      "loss": 0.3231,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998817165655597e-05,
+      "loss": 0.3313,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998811347214396e-05,
+      "loss": 0.3798,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998805514506144e-05,
+      "loss": 0.4095,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998799667530925e-05,
+      "loss": 0.3876,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987938062888218e-05,
+      "loss": 0.3433,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987879307799184e-05,
+      "loss": 0.3372,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987820410042988e-05,
+      "loss": 0.3147,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998776136962047e-05,
+      "loss": 0.3542,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987702186532475e-05,
+      "loss": 0.3131,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998764286077985e-05,
+      "loss": 0.3449,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998758339236344e-05,
+      "loss": 0.372,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987523781284093e-05,
+      "loss": 0.3403,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987464027542658e-05,
+      "loss": 0.3694,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987404131139998e-05,
+      "loss": 0.3197,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987344092076962e-05,
+      "loss": 0.3515,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987283910354404e-05,
+      "loss": 0.3336,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987223585973195e-05,
+      "loss": 0.3362,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987163118934183e-05,
+      "loss": 0.3997,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987102509238243e-05,
+      "loss": 0.3762,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987041756886235e-05,
+      "loss": 0.3625,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986980861879028e-05,
+      "loss": 0.3482,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986919824217493e-05,
+      "loss": 0.2927,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986858643902502e-05,
+      "loss": 0.3187,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986797320934927e-05,
+      "loss": 0.3666,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986735855315644e-05,
+      "loss": 0.3824,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986674247045533e-05,
+      "loss": 0.3599,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998661249612547e-05,
+      "loss": 0.344,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986550602556338e-05,
+      "loss": 0.3661,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998648856633903e-05,
+      "loss": 0.3286,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998642638747442e-05,
+      "loss": 0.3673,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986364065963403e-05,
+      "loss": 0.3446,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986301601806865e-05,
+      "loss": 0.3325,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99862389950057e-05,
+      "loss": 0.3424,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986176245560804e-05,
+      "loss": 0.3617,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986113353473073e-05,
+      "loss": 0.3022,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986050318743406e-05,
+      "loss": 0.3606,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985987141372698e-05,
+      "loss": 0.3377,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998592382136186e-05,
+      "loss": 0.3433,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985860358711787e-05,
+      "loss": 0.3502,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985796753423393e-05,
+      "loss": 0.3545,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985733005497584e-05,
+      "loss": 0.3274,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998566911493527e-05,
+      "loss": 0.3294,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985605081737362e-05,
+      "loss": 0.3791,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998554090590478e-05,
+      "loss": 0.3382,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985476587438435e-05,
+      "loss": 0.3322,
+      "step": 1264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985412126339246e-05,
+      "loss": 0.3675,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998534752260814e-05,
+      "loss": 0.3318,
+      "step": 1266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985282776246032e-05,
+      "loss": 0.3226,
+      "step": 1267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985217887253856e-05,
+      "loss": 0.3604,
+      "step": 1268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998515285563253e-05,
+      "loss": 0.3406,
+      "step": 1269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985087681382987e-05,
+      "loss": 0.3146,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985022364506154e-05,
+      "loss": 0.3439,
+      "step": 1271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984956905002973e-05,
+      "loss": 0.3772,
+      "step": 1272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984891302874368e-05,
+      "loss": 0.3626,
+      "step": 1273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984825558121285e-05,
+      "loss": 0.3239,
+      "step": 1274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984759670744656e-05,
+      "loss": 0.3464,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984693640745428e-05,
+      "loss": 0.3377,
+      "step": 1276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998462746812454e-05,
+      "loss": 0.354,
+      "step": 1277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984561152882943e-05,
+      "loss": 0.3773,
+      "step": 1278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984494695021577e-05,
+      "loss": 0.3328,
+      "step": 1279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984428094541394e-05,
+      "loss": 0.373,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984361351443343e-05,
+      "loss": 0.3265,
+      "step": 1281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984294465728386e-05,
+      "loss": 0.3542,
+      "step": 1282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998422743739747e-05,
+      "loss": 0.366,
+      "step": 1283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984160266451555e-05,
+      "loss": 0.3385,
+      "step": 1284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99840929528916e-05,
+      "loss": 0.3496,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984025496718567e-05,
+      "loss": 0.3199,
+      "step": 1286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998395789793342e-05,
+      "loss": 0.376,
+      "step": 1287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983890156537126e-05,
+      "loss": 0.3397,
+      "step": 1288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998382227253065e-05,
+      "loss": 0.3958,
+      "step": 1289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998375424591496e-05,
+      "loss": 0.3816,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983686076691032e-05,
+      "loss": 0.3755,
+      "step": 1291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983617764859838e-05,
+      "loss": 0.3439,
+      "step": 1292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998354931042235e-05,
+      "loss": 0.3498,
+      "step": 1293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983480713379552e-05,
+      "loss": 0.3512,
+      "step": 1294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983411973732423e-05,
+      "loss": 0.2975,
+      "step": 1295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998334309148194e-05,
+      "loss": 0.3432,
+      "step": 1296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998327406662909e-05,
+      "loss": 0.3709,
+      "step": 1297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983204899174862e-05,
+      "loss": 0.3324,
+      "step": 1298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998313558912024e-05,
+      "loss": 0.341,
+      "step": 1299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983066136466212e-05,
+      "loss": 0.3249,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982996541213776e-05,
+      "loss": 0.3429,
+      "step": 1301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982926803363922e-05,
+      "loss": 0.3491,
+      "step": 1302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982856922917647e-05,
+      "loss": 0.3672,
+      "step": 1303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998278689987595e-05,
+      "loss": 0.3306,
+      "step": 1304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982716734239832e-05,
+      "loss": 0.3481,
+      "step": 1305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998264642601029e-05,
+      "loss": 0.3498,
+      "step": 1306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982575975188337e-05,
+      "loss": 0.3701,
+      "step": 1307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998250538177497e-05,
+      "loss": 0.3936,
+      "step": 1308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982434645771204e-05,
+      "loss": 0.3192,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998236376717805e-05,
+      "loss": 0.3036,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982292745996514e-05,
+      "loss": 0.3157,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982221582227616e-05,
+      "loss": 0.3376,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998215027587237e-05,
+      "loss": 0.3793,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982078826931794e-05,
+      "loss": 0.338,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982007235406914e-05,
+      "loss": 0.3468,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981935501298745e-05,
+      "loss": 0.3811,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981863624608314e-05,
+      "loss": 0.3421,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998179160533665e-05,
+      "loss": 0.3361,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981719443484778e-05,
+      "loss": 0.3168,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981647139053737e-05,
+      "loss": 0.3603,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998157469204455e-05,
+      "loss": 0.3081,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981502102458258e-05,
+      "loss": 0.3235,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981429370295893e-05,
+      "loss": 0.3564,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981356495558497e-05,
+      "loss": 0.3623,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981283478247113e-05,
+      "loss": 0.3414,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998121031836278e-05,
+      "loss": 0.3656,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981137015906546e-05,
+      "loss": 0.3417,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981063570879454e-05,
+      "loss": 0.3088,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980989983282558e-05,
+      "loss": 0.3628,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980916253116905e-05,
+      "loss": 0.407,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980842380383555e-05,
+      "loss": 0.3361,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980768365083553e-05,
+      "loss": 0.3326,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980694207217964e-05,
+      "loss": 0.3401,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980619906787844e-05,
+      "loss": 0.342,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998054546379425e-05,
+      "loss": 0.3223,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998047087823826e-05,
+      "loss": 0.3317,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980396150120925e-05,
+      "loss": 0.3243,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980321279443315e-05,
+      "loss": 0.3486,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980246266206505e-05,
+      "loss": 0.3517,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998017111041156e-05,
+      "loss": 0.3727,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998009581205956e-05,
+      "loss": 0.3764,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980020371151575e-05,
+      "loss": 0.3532,
+      "step": 1342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979944787688686e-05,
+      "loss": 0.3415,
+      "step": 1343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997986906167197e-05,
+      "loss": 0.3363,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997979319310251e-05,
+      "loss": 0.3716,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979717181981394e-05,
+      "loss": 0.3768,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99796410283097e-05,
+      "loss": 0.3542,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979564732088522e-05,
+      "loss": 0.4063,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979488293318946e-05,
+      "loss": 0.3358,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979411712002062e-05,
+      "loss": 0.3436,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997933498813897e-05,
+      "loss": 0.3272,
+      "step": 1351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979258121730763e-05,
+      "loss": 0.3153,
+      "step": 1352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979181112778535e-05,
+      "loss": 0.3089,
+      "step": 1353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979103961283395e-05,
+      "loss": 0.3229,
+      "step": 1354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979026667246437e-05,
+      "loss": 0.3598,
+      "step": 1355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997894923066877e-05,
+      "loss": 0.3093,
+      "step": 1356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978871651551496e-05,
+      "loss": 0.3429,
+      "step": 1357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978793929895724e-05,
+      "loss": 0.3049,
+      "step": 1358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978716065702566e-05,
+      "loss": 0.3351,
+      "step": 1359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978638058973137e-05,
+      "loss": 0.3089,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978559909708547e-05,
+      "loss": 0.3036,
+      "step": 1361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997848161790991e-05,
+      "loss": 0.3651,
+      "step": 1362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997840318357835e-05,
+      "loss": 0.3801,
+      "step": 1363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978324606714986e-05,
+      "loss": 0.3485,
+      "step": 1364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978245887320935e-05,
+      "loss": 0.3248,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997816702539733e-05,
+      "loss": 0.344,
+      "step": 1366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997808802094529e-05,
+      "loss": 0.3239,
+      "step": 1367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978008873965948e-05,
+      "loss": 0.3637,
+      "step": 1368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997792958446043e-05,
+      "loss": 0.3301,
+      "step": 1369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977850152429875e-05,
+      "loss": 0.3176,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977770577875413e-05,
+      "loss": 0.3592,
+      "step": 1371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977690860798184e-05,
+      "loss": 0.3517,
+      "step": 1372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997761100119932e-05,
+      "loss": 0.33,
+      "step": 1373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997753099907997e-05,
+      "loss": 0.3696,
+      "step": 1374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977450854441277e-05,
+      "loss": 0.3327,
+      "step": 1375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977370567284376e-05,
+      "loss": 0.3353,
+      "step": 1376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997729013761042e-05,
+      "loss": 0.3529,
+      "step": 1377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977209565420558e-05,
+      "loss": 0.3135,
+      "step": 1378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977128850715944e-05,
+      "loss": 0.3694,
+      "step": 1379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977047993497724e-05,
+      "loss": 0.3495,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976966993767056e-05,
+      "loss": 0.3313,
+      "step": 1381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99768858515251e-05,
+      "loss": 0.3293,
+      "step": 1382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997680456677301e-05,
+      "loss": 0.3265,
+      "step": 1383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997672313951195e-05,
+      "loss": 0.3397,
+      "step": 1384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997664156974308e-05,
+      "loss": 0.3064,
+      "step": 1385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976559857467572e-05,
+      "loss": 0.3641,
+      "step": 1386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997647800268659e-05,
+      "loss": 0.3463,
+      "step": 1387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99763960054013e-05,
+      "loss": 0.3525,
+      "step": 1388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976313865612872e-05,
+      "loss": 0.3249,
+      "step": 1389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997623158332249e-05,
+      "loss": 0.3506,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976149158531314e-05,
+      "loss": 0.3533,
+      "step": 1391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976066591240534e-05,
+      "loss": 0.3362,
+      "step": 1392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975983881451325e-05,
+      "loss": 0.318,
+      "step": 1393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975901029164865e-05,
+      "loss": 0.3276,
+      "step": 1394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997581803438234e-05,
+      "loss": 0.3658,
+      "step": 1395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975734897104943e-05,
+      "loss": 0.3079,
+      "step": 1396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997565161733385e-05,
+      "loss": 0.3343,
+      "step": 1397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975568195070253e-05,
+      "loss": 0.3281,
+      "step": 1398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997548463031535e-05,
+      "loss": 0.3151,
+      "step": 1399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975400923070328e-05,
+      "loss": 0.3349,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997531707333639e-05,
+      "loss": 0.3716,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975233081114723e-05,
+      "loss": 0.349,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975148946406537e-05,
+      "loss": 0.2893,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975064669213023e-05,
+      "loss": 0.3455,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974980249535398e-05,
+      "loss": 0.3206,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974895687374857e-05,
+      "loss": 0.3197,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974810982732614e-05,
+      "loss": 0.3445,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974726135609878e-05,
+      "loss": 0.3526,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974641146007856e-05,
+      "loss": 0.3461,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997455601392777e-05,
+      "loss": 0.3543,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997447073937083e-05,
+      "loss": 0.3483,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974385322338255e-05,
+      "loss": 0.3624,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974299762831266e-05,
+      "loss": 0.3089,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974214060851085e-05,
+      "loss": 0.3712,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997412821639894e-05,
+      "loss": 0.3721,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997404222947605e-05,
+      "loss": 0.3095,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973956100083646e-05,
+      "loss": 0.3366,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997386982822296e-05,
+      "loss": 0.3722,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973783413895225e-05,
+      "loss": 0.3316,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973696857101676e-05,
+      "loss": 0.3401,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997361015784354e-05,
+      "loss": 0.3079,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973523316122068e-05,
+      "loss": 0.3965,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973436331938493e-05,
+      "loss": 0.3148,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973349205294062e-05,
+      "loss": 0.3541,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973261936190013e-05,
+      "loss": 0.3785,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973174524627595e-05,
+      "loss": 0.3434,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973086970608063e-05,
+      "loss": 0.3413,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972999274132663e-05,
+      "loss": 0.3161,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972911435202645e-05,
+      "loss": 0.3354,
+      "step": 1429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972823453819266e-05,
+      "loss": 0.3355,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972735329983784e-05,
+      "loss": 0.3632,
+      "step": 1431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972647063697458e-05,
+      "loss": 0.3338,
+      "step": 1432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972558654961546e-05,
+      "loss": 0.331,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997247010377731e-05,
+      "loss": 0.3207,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972381410146018e-05,
+      "loss": 0.31,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972292574068936e-05,
+      "loss": 0.3685,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972203595547334e-05,
+      "loss": 0.3322,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972114474582484e-05,
+      "loss": 0.322,
+      "step": 1438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972025211175654e-05,
+      "loss": 0.329,
+      "step": 1439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997193580532812e-05,
+      "loss": 0.3519,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971846257041167e-05,
+      "loss": 0.3744,
+      "step": 1441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997175656631606e-05,
+      "loss": 0.3149,
+      "step": 1442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971666733154095e-05,
+      "loss": 0.3508,
+      "step": 1443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971576757556545e-05,
+      "loss": 0.3611,
+      "step": 1444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971486639524704e-05,
+      "loss": 0.3348,
+      "step": 1445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971396379059847e-05,
+      "loss": 0.3231,
+      "step": 1446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971305976163273e-05,
+      "loss": 0.317,
+      "step": 1447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971215430836272e-05,
+      "loss": 0.3732,
+      "step": 1448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971124743080132e-05,
+      "loss": 0.3146,
+      "step": 1449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971033912896155e-05,
+      "loss": 0.3414,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970942940285637e-05,
+      "loss": 0.3544,
+      "step": 1451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970851825249874e-05,
+      "loss": 0.3212,
+      "step": 1452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997076056779017e-05,
+      "loss": 0.3131,
+      "step": 1453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997066916790783e-05,
+      "loss": 0.3345,
+      "step": 1454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970577625604154e-05,
+      "loss": 0.335,
+      "step": 1455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970485940880454e-05,
+      "loss": 0.3235,
+      "step": 1456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997039411373804e-05,
+      "loss": 0.3892,
+      "step": 1457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970302144178224e-05,
+      "loss": 0.3497,
+      "step": 1458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970210032202314e-05,
+      "loss": 0.3266,
+      "step": 1459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970117777811636e-05,
+      "loss": 0.3063,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970025381007497e-05,
+      "loss": 0.3434,
+      "step": 1461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969932841791224e-05,
+      "loss": 0.3184,
+      "step": 1462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969840160164134e-05,
+      "loss": 0.3603,
+      "step": 1463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969747336127557e-05,
+      "loss": 0.3384,
+      "step": 1464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996965436968281e-05,
+      "loss": 0.3516,
+      "step": 1465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969561260831232e-05,
+      "loss": 0.3277,
+      "step": 1466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969468009574144e-05,
+      "loss": 0.3379,
+      "step": 1467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996937461591288e-05,
+      "loss": 0.3521,
+      "step": 1468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969281079848776e-05,
+      "loss": 0.3239,
+      "step": 1469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996918740138317e-05,
+      "loss": 0.3558,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969093580517397e-05,
+      "loss": 0.3031,
+      "step": 1471
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968999617252792e-05,
+      "loss": 0.3537,
+      "step": 1472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968905511590707e-05,
+      "loss": 0.305,
+      "step": 1473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968811263532483e-05,
+      "loss": 0.3155,
+      "step": 1474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968716873079462e-05,
+      "loss": 0.3521,
+      "step": 1475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968622340232993e-05,
+      "loss": 0.3619,
+      "step": 1476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996852766499443e-05,
+      "loss": 0.3262,
+      "step": 1477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968432847365127e-05,
+      "loss": 0.3236,
+      "step": 1478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968337887346435e-05,
+      "loss": 0.3206,
+      "step": 1479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968242784939714e-05,
+      "loss": 0.3307,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968147540146315e-05,
+      "loss": 0.3532,
+      "step": 1481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968052152967603e-05,
+      "loss": 0.3304,
+      "step": 1482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996795662340494e-05,
+      "loss": 0.3366,
+      "step": 1483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996786095145969e-05,
+      "loss": 0.3424,
+      "step": 1484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996776513713322e-05,
+      "loss": 0.3247,
+      "step": 1485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967669180426902e-05,
+      "loss": 0.3696,
+      "step": 1486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 0.3532,
+      "step": 1487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967476839880194e-05,
+      "loss": 0.3742,
+      "step": 1488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967380456042552e-05,
+      "loss": 0.3398,
+      "step": 1489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967283929830558e-05,
+      "loss": 0.334,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996718726124558e-05,
+      "loss": 0.3564,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996709045028901e-05,
+      "loss": 0.3325,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966993496962224e-05,
+      "loss": 0.3639,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996689640126661e-05,
+      "loss": 0.3119,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966799163203558e-05,
+      "loss": 0.344,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966701782774445e-05,
+      "loss": 0.3609,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966604259980677e-05,
+      "loss": 0.326,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966506594823635e-05,
+      "loss": 0.3232,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996640878730472e-05,
+      "loss": 0.3188,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996631083742533e-05,
+      "loss": 0.3577,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966212745186863e-05,
+      "loss": 0.3294,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996611451059072e-05,
+      "loss": 0.342,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966016133638298e-05,
+      "loss": 0.3202,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965917614331012e-05,
+      "loss": 0.3427,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965818952670265e-05,
+      "loss": 0.3384,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965720148657463e-05,
+      "loss": 0.3256,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965621202294024e-05,
+      "loss": 0.3471,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965522113581356e-05,
+      "loss": 0.3265,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996542288252088e-05,
+      "loss": 0.3199,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965323509114004e-05,
+      "loss": 0.3349,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996522399336216e-05,
+      "loss": 0.3629,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965124335266756e-05,
+      "loss": 0.3416,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965024534829225e-05,
+      "loss": 0.3018,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996492459205099e-05,
+      "loss": 0.3134,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996482450693348e-05,
+      "loss": 0.319,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964724279478122e-05,
+      "loss": 0.3653,
+      "step": 1516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964623909686346e-05,
+      "loss": 0.3556,
+      "step": 1517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996452339755959e-05,
+      "loss": 0.3063,
+      "step": 1518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964422743099292e-05,
+      "loss": 0.3475,
+      "step": 1519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996432194630688e-05,
+      "loss": 0.3289,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964221007183803e-05,
+      "loss": 0.3507,
+      "step": 1521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99641199257315e-05,
+      "loss": 0.3427,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964018701951413e-05,
+      "loss": 0.3117,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996391733584499e-05,
+      "loss": 0.323,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963815827413677e-05,
+      "loss": 0.3278,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963714176658924e-05,
+      "loss": 0.3369,
+      "step": 1526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963612383582186e-05,
+      "loss": 0.3227,
+      "step": 1527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963510448184913e-05,
+      "loss": 0.3156,
+      "step": 1528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963408370468563e-05,
+      "loss": 0.3423,
+      "step": 1529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99633061504346e-05,
+      "loss": 0.353,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963203788084472e-05,
+      "loss": 0.3017,
+      "step": 1531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963101283419645e-05,
+      "loss": 0.3498,
+      "step": 1532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996299863644159e-05,
+      "loss": 0.3317,
+      "step": 1533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996289584715177e-05,
+      "loss": 0.3345,
+      "step": 1534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996279291555165e-05,
+      "loss": 0.3193,
+      "step": 1535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962689841642705e-05,
+      "loss": 0.3504,
+      "step": 1536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99625866254264e-05,
+      "loss": 0.331,
+      "step": 1537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962483266904216e-05,
+      "loss": 0.3162,
+      "step": 1538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962379766077627e-05,
+      "loss": 0.3236,
+      "step": 1539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962276122948114e-05,
+      "loss": 0.3259,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962172337517154e-05,
+      "loss": 0.3266,
+      "step": 1541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962068409786228e-05,
+      "loss": 0.3314,
+      "step": 1542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996196433975683e-05,
+      "loss": 0.3384,
+      "step": 1543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961860127430432e-05,
+      "loss": 0.3216,
+      "step": 1544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961755772808534e-05,
+      "loss": 0.3689,
+      "step": 1545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961651275892626e-05,
+      "loss": 0.3323,
+      "step": 1546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996154663668419e-05,
+      "loss": 0.3461,
+      "step": 1547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961441855184732e-05,
+      "loss": 0.3631,
+      "step": 1548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961336931395745e-05,
+      "loss": 0.3413,
+      "step": 1549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961231865318728e-05,
+      "loss": 0.3492,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996112665695518e-05,
+      "loss": 0.329,
+      "step": 1551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961021306306607e-05,
+      "loss": 0.3747,
+      "step": 1552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996091581337451e-05,
+      "loss": 0.3188,
+      "step": 1553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99608101781604e-05,
+      "loss": 0.3382,
+      "step": 1554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960704400665786e-05,
+      "loss": 0.323,
+      "step": 1555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996059848089217e-05,
+      "loss": 0.3245,
+      "step": 1556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960492418841077e-05,
+      "loss": 0.3161,
+      "step": 1557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960386214514016e-05,
+      "loss": 0.3345,
+      "step": 1558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960279867912504e-05,
+      "loss": 0.3439,
+      "step": 1559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996017337903806e-05,
+      "loss": 0.3349,
+      "step": 1560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960066747892206e-05,
+      "loss": 0.3512,
+      "step": 1561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959959974476467e-05,
+      "loss": 0.3464,
+      "step": 1562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959853058792364e-05,
+      "loss": 0.344,
+      "step": 1563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959746000841426e-05,
+      "loss": 0.3404,
+      "step": 1564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959638800625183e-05,
+      "loss": 0.3659,
+      "step": 1565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995953145814517e-05,
+      "loss": 0.3469,
+      "step": 1566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995942397340291e-05,
+      "loss": 0.3291,
+      "step": 1567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959316346399946e-05,
+      "loss": 0.3229,
+      "step": 1568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959208577137813e-05,
+      "loss": 0.2862,
+      "step": 1569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959100665618053e-05,
+      "loss": 0.3136,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958992611842206e-05,
+      "loss": 0.3412,
+      "step": 1571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958884415811814e-05,
+      "loss": 0.335,
+      "step": 1572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958776077528425e-05,
+      "loss": 0.3282,
+      "step": 1573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995866759699358e-05,
+      "loss": 0.3224,
+      "step": 1574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995855897420884e-05,
+      "loss": 0.3486,
+      "step": 1575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958450209175747e-05,
+      "loss": 0.3678,
+      "step": 1576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958341301895858e-05,
+      "loss": 0.3189,
+      "step": 1577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958232252370728e-05,
+      "loss": 0.3245,
+      "step": 1578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958123060601915e-05,
+      "loss": 0.3376,
+      "step": 1579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958013726590978e-05,
+      "loss": 0.3535,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957904250339478e-05,
+      "loss": 0.3166,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957794631848985e-05,
+      "loss": 0.3091,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957684871121053e-05,
+      "loss": 0.3186,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957574968157262e-05,
+      "loss": 0.3065,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957464922959176e-05,
+      "loss": 0.3414,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995735473552837e-05,
+      "loss": 0.3612,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995724440586641e-05,
+      "loss": 0.3334,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957133933974878e-05,
+      "loss": 0.3098,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957023319855357e-05,
+      "loss": 0.3177,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956912563509415e-05,
+      "loss": 0.3711,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956801664938642e-05,
+      "loss": 0.3136,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956690624144627e-05,
+      "loss": 0.3299,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956579441128942e-05,
+      "loss": 0.3124,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956468115893183e-05,
+      "loss": 0.3251,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956356648438945e-05,
+      "loss": 0.3312,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995624503876781e-05,
+      "loss": 0.3329,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995613328688138e-05,
+      "loss": 0.3517,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956021392781247e-05,
+      "loss": 0.3742,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955909356469013e-05,
+      "loss": 0.3699,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955797177946275e-05,
+      "loss": 0.329,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955684857214638e-05,
+      "loss": 0.3383,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955572394275703e-05,
+      "loss": 0.3423,
+      "step": 1602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995545978913108e-05,
+      "loss": 0.3315,
+      "step": 1603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955347041782377e-05,
+      "loss": 0.318,
+      "step": 1604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99552341522312e-05,
+      "loss": 0.3337,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955121120479167e-05,
+      "loss": 0.3263,
+      "step": 1606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995500794652789e-05,
+      "loss": 0.3356,
+      "step": 1607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954894630378987e-05,
+      "loss": 0.3109,
+      "step": 1608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954781172034074e-05,
+      "loss": 0.3234,
+      "step": 1609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954667571494773e-05,
+      "loss": 0.3433,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995455382876271e-05,
+      "loss": 0.3146,
+      "step": 1611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954439943839506e-05,
+      "loss": 0.322,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954325916726786e-05,
+      "loss": 0.3518,
+      "step": 1613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954211747426182e-05,
+      "loss": 0.37,
+      "step": 1614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954097435939324e-05,
+      "loss": 0.3218,
+      "step": 1615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953982982267846e-05,
+      "loss": 0.3831,
+      "step": 1616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953868386413383e-05,
+      "loss": 0.3329,
+      "step": 1617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953753648377568e-05,
+      "loss": 0.2791,
+      "step": 1618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953638768162042e-05,
+      "loss": 0.3341,
+      "step": 1619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953523745768448e-05,
+      "loss": 0.325,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953408581198426e-05,
+      "loss": 0.332,
+      "step": 1621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953293274453625e-05,
+      "loss": 0.3347,
+      "step": 1622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995317782553569e-05,
+      "loss": 0.3469,
+      "step": 1623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953062234446265e-05,
+      "loss": 0.3192,
+      "step": 1624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995294650118701e-05,
+      "loss": 0.3365,
+      "step": 1625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952830625759568e-05,
+      "loss": 0.3222,
+      "step": 1626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952714608165605e-05,
+      "loss": 0.3336,
+      "step": 1627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995259844840677e-05,
+      "loss": 0.3316,
+      "step": 1628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995248214648473e-05,
+      "loss": 0.31,
+      "step": 1629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952365702401136e-05,
+      "loss": 0.3621,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995224911615766e-05,
+      "loss": 0.3016,
+      "step": 1631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952132387755965e-05,
+      "loss": 0.3226,
+      "step": 1632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952015517197717e-05,
+      "loss": 0.3597,
+      "step": 1633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951898504484586e-05,
+      "loss": 0.3622,
+      "step": 1634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951781349618242e-05,
+      "loss": 0.3577,
+      "step": 1635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951664052600362e-05,
+      "loss": 0.3431,
+      "step": 1636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995154661343262e-05,
+      "loss": 0.3468,
+      "step": 1637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995142903211669e-05,
+      "loss": 0.3491,
+      "step": 1638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951311308654256e-05,
+      "loss": 0.3867,
+      "step": 1639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951193443046997e-05,
+      "loss": 0.3617,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951075435296596e-05,
+      "loss": 0.3418,
+      "step": 1641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950957285404745e-05,
+      "loss": 0.3229,
+      "step": 1642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995083899337312e-05,
+      "loss": 0.354,
+      "step": 1643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950720559203423e-05,
+      "loss": 0.3698,
+      "step": 1644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950601982897338e-05,
+      "loss": 0.3533,
+      "step": 1645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950483264456563e-05,
+      "loss": 0.3316,
+      "step": 1646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995036440388279e-05,
+      "loss": 0.3264,
+      "step": 1647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995024540117772e-05,
+      "loss": 0.2863,
+      "step": 1648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995012625634305e-05,
+      "loss": 0.3603,
+      "step": 1649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950006969380484e-05,
+      "loss": 0.3248,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949887540291724e-05,
+      "loss": 0.3266,
+      "step": 1651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994976796907848e-05,
+      "loss": 0.3366,
+      "step": 1652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994964825574245e-05,
+      "loss": 0.3534,
+      "step": 1653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949528400285358e-05,
+      "loss": 0.3446,
+      "step": 1654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949408402708904e-05,
+      "loss": 0.3277,
+      "step": 1655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949288263014815e-05,
+      "loss": 0.2992,
+      "step": 1656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994916798120479e-05,
+      "loss": 0.3109,
+      "step": 1657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949047557280564e-05,
+      "loss": 0.3303,
+      "step": 1658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948926991243843e-05,
+      "loss": 0.34,
+      "step": 1659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994880628309636e-05,
+      "loss": 0.3026,
+      "step": 1660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948685432839833e-05,
+      "loss": 0.3319,
+      "step": 1661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994856444047599e-05,
+      "loss": 0.3288,
+      "step": 1662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994844330600656e-05,
+      "loss": 0.3665,
+      "step": 1663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994832202943327e-05,
+      "loss": 0.3275,
+      "step": 1664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948200610757857e-05,
+      "loss": 0.3162,
+      "step": 1665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948079049982058e-05,
+      "loss": 0.3638,
+      "step": 1666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99479573471076e-05,
+      "loss": 0.3072,
+      "step": 1667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947835502136223e-05,
+      "loss": 0.3634,
+      "step": 1668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947713515069674e-05,
+      "loss": 0.3569,
+      "step": 1669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947591385909692e-05,
+      "loss": 0.326,
+      "step": 1670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994746911465802e-05,
+      "loss": 0.3277,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947346701316406e-05,
+      "loss": 0.3355,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99472241458866e-05,
+      "loss": 0.3396,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994710144837035e-05,
+      "loss": 0.3385,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994697860876941e-05,
+      "loss": 0.351,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946855627085532e-05,
+      "loss": 0.3659,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994673250332048e-05,
+      "loss": 0.3189,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946609237476e-05,
+      "loss": 0.325,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994648582955387e-05,
+      "loss": 0.3455,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946362279555837e-05,
+      "loss": 0.3521,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946238587483674e-05,
+      "loss": 0.3214,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946114753339147e-05,
+      "loss": 0.3196,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994599077712402e-05,
+      "loss": 0.3172,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994586665884007e-05,
+      "loss": 0.3506,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994574239848907e-05,
+      "loss": 0.3158,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994561799607279e-05,
+      "loss": 0.3368,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945493451593005e-05,
+      "loss": 0.3411,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99453687650515e-05,
+      "loss": 0.3075,
+      "step": 1688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994524393645006e-05,
+      "loss": 0.3361,
+      "step": 1689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994511896579046e-05,
+      "loss": 0.3013,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944993853074485e-05,
+      "loss": 0.2781,
+      "step": 1691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944868598303926e-05,
+      "loss": 0.3039,
+      "step": 1692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944743201480573e-05,
+      "loss": 0.328,
+      "step": 1693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944617662606213e-05,
+      "loss": 0.3141,
+      "step": 1694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994449198168264e-05,
+      "loss": 0.304,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944366158711652e-05,
+      "loss": 0.297,
+      "step": 1696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944240193695043e-05,
+      "loss": 0.3045,
+      "step": 1697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944114086634617e-05,
+      "loss": 0.3263,
+      "step": 1698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994398783753217e-05,
+      "loss": 0.3206,
+      "step": 1699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943861446389508e-05,
+      "loss": 0.2782,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943734913208436e-05,
+      "loss": 0.314,
+      "step": 1701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994360823799076e-05,
+      "loss": 0.3192,
+      "step": 1702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943481420738295e-05,
+      "loss": 0.3069,
+      "step": 1703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943354461452844e-05,
+      "loss": 0.322,
+      "step": 1704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943227360136227e-05,
+      "loss": 0.3471,
+      "step": 1705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994310011679026e-05,
+      "loss": 0.3587,
+      "step": 1706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942972731416754e-05,
+      "loss": 0.3059,
+      "step": 1707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942845204017533e-05,
+      "loss": 0.3523,
+      "step": 1708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942717534594417e-05,
+      "loss": 0.3255,
+      "step": 1709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942589723149233e-05,
+      "loss": 0.2966,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99424617696838e-05,
+      "loss": 0.3142,
+      "step": 1711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942333674199954e-05,
+      "loss": 0.3455,
+      "step": 1712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994220543669952e-05,
+      "loss": 0.2979,
+      "step": 1713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994207705718433e-05,
+      "loss": 0.3173,
+      "step": 1714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941948535656217e-05,
+      "loss": 0.3734,
+      "step": 1715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994181987211702e-05,
+      "loss": 0.3387,
+      "step": 1716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941691066568575e-05,
+      "loss": 0.3321,
+      "step": 1717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941562119012722e-05,
+      "loss": 0.3131,
+      "step": 1718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941433029451304e-05,
+      "loss": 0.3004,
+      "step": 1719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994130379788616e-05,
+      "loss": 0.3129,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994117442431914e-05,
+      "loss": 0.3008,
+      "step": 1721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941044908752092e-05,
+      "loss": 0.3352,
+      "step": 1722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940915251186867e-05,
+      "loss": 0.3213,
+      "step": 1723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940785451625312e-05,
+      "loss": 0.3434,
+      "step": 1724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940655510069286e-05,
+      "loss": 0.3363,
+      "step": 1725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940525426520647e-05,
+      "loss": 0.3306,
+      "step": 1726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940395200981246e-05,
+      "loss": 0.3382,
+      "step": 1727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994026483345295e-05,
+      "loss": 0.3615,
+      "step": 1728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940134323937615e-05,
+      "loss": 0.3079,
+      "step": 1729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994000367243711e-05,
+      "loss": 0.3313,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99398728789533e-05,
+      "loss": 0.3457,
+      "step": 1731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939741943488052e-05,
+      "loss": 0.3432,
+      "step": 1732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993961086604324e-05,
+      "loss": 0.3351,
+      "step": 1733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939479646620727e-05,
+      "loss": 0.297,
+      "step": 1734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99393482852224e-05,
+      "loss": 0.3283,
+      "step": 1735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939216781850127e-05,
+      "loss": 0.3583,
+      "step": 1736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939085136505786e-05,
+      "loss": 0.2881,
+      "step": 1737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938953349191262e-05,
+      "loss": 0.3299,
+      "step": 1738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938821419908436e-05,
+      "loss": 0.3474,
+      "step": 1739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993868934865919e-05,
+      "loss": 0.3228,
+      "step": 1740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938557135445415e-05,
+      "loss": 0.3249,
+      "step": 1741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938424780268995e-05,
+      "loss": 0.3599,
+      "step": 1742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993829228313182e-05,
+      "loss": 0.3543,
+      "step": 1743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938159644035792e-05,
+      "loss": 0.3276,
+      "step": 1744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938026862982793e-05,
+      "loss": 0.3301,
+      "step": 1745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937893939974727e-05,
+      "loss": 0.3249,
+      "step": 1746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993776087501349e-05,
+      "loss": 0.3407,
+      "step": 1747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937627668100985e-05,
+      "loss": 0.3139,
+      "step": 1748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937494319239112e-05,
+      "loss": 0.3237,
+      "step": 1749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993736082842978e-05,
+      "loss": 0.3383,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937227195674894e-05,
+      "loss": 0.3166,
+      "step": 1751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937093420976357e-05,
+      "loss": 0.3464,
+      "step": 1752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936959504336085e-05,
+      "loss": 0.3126,
+      "step": 1753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936825445755995e-05,
+      "loss": 0.3278,
+      "step": 1754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936691245237997e-05,
+      "loss": 0.3042,
+      "step": 1755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936556902784003e-05,
+      "loss": 0.3324,
+      "step": 1756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993642241839594e-05,
+      "loss": 0.3082,
+      "step": 1757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936287792075727e-05,
+      "loss": 0.3503,
+      "step": 1758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936153023825287e-05,
+      "loss": 0.3151,
+      "step": 1759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936018113646542e-05,
+      "loss": 0.3564,
+      "step": 1760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935883061541426e-05,
+      "loss": 0.3486,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935747867511857e-05,
+      "loss": 0.3535,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935612531559777e-05,
+      "loss": 0.3388,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935477053687113e-05,
+      "loss": 0.3075,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935341433895803e-05,
+      "loss": 0.2957,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993520567218778e-05,
+      "loss": 0.3241,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993506976856499e-05,
+      "loss": 0.3276,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993493372302937e-05,
+      "loss": 0.3176,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934797535582862e-05,
+      "loss": 0.3182,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934661206227417e-05,
+      "loss": 0.3476,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993452473496498e-05,
+      "loss": 0.3315,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993438812179749e-05,
+      "loss": 0.3222,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993425136672691e-05,
+      "loss": 0.3267,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934114469755194e-05,
+      "loss": 0.319,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933977430884294e-05,
+      "loss": 0.3133,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933840250116166e-05,
+      "loss": 0.3193,
+      "step": 1776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993370292745277e-05,
+      "loss": 0.3305,
+      "step": 1777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993356546289607e-05,
+      "loss": 0.2962,
+      "step": 1778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993342785644803e-05,
+      "loss": 0.3157,
+      "step": 1779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933290108110614e-05,
+      "loss": 0.3323,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933152217885784e-05,
+      "loss": 0.3313,
+      "step": 1781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993301418577552e-05,
+      "loss": 0.3601,
+      "step": 1782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932876011781786e-05,
+      "loss": 0.3382,
+      "step": 1783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932737695906556e-05,
+      "loss": 0.3442,
+      "step": 1784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932599238151814e-05,
+      "loss": 0.2978,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932460638519528e-05,
+      "loss": 0.3649,
+      "step": 1786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932321897011682e-05,
+      "loss": 0.3104,
+      "step": 1787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932183013630257e-05,
+      "loss": 0.3464,
+      "step": 1788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932043988377236e-05,
+      "loss": 0.3478,
+      "step": 1789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993190482125461e-05,
+      "loss": 0.3545,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993176551226436e-05,
+      "loss": 0.3291,
+      "step": 1791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931626061408474e-05,
+      "loss": 0.3421,
+      "step": 1792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931486468688956e-05,
+      "loss": 0.3223,
+      "step": 1793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931346734107788e-05,
+      "loss": 0.3239,
+      "step": 1794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993120685766697e-05,
+      "loss": 0.3449,
+      "step": 1795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931066839368505e-05,
+      "loss": 0.3365,
+      "step": 1796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930926679214386e-05,
+      "loss": 0.3629,
+      "step": 1797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930786377206618e-05,
+      "loss": 0.2982,
+      "step": 1798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930645933347205e-05,
+      "loss": 0.3351,
+      "step": 1799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930505347638153e-05,
+      "loss": 0.3222,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930364620081467e-05,
+      "loss": 0.3413,
+      "step": 1801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930223750679162e-05,
+      "loss": 0.336,
+      "step": 1802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993008273943325e-05,
+      "loss": 0.3142,
+      "step": 1803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992994158634574e-05,
+      "loss": 0.3381,
+      "step": 1804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929800291418657e-05,
+      "loss": 0.3453,
+      "step": 1805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992965885465401e-05,
+      "loss": 0.3053,
+      "step": 1806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929517276053824e-05,
+      "loss": 0.3385,
+      "step": 1807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992937555562012e-05,
+      "loss": 0.3304,
+      "step": 1808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929233693354922e-05,
+      "loss": 0.3445,
+      "step": 1809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929091689260258e-05,
+      "loss": 0.3349,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928949543338153e-05,
+      "loss": 0.3588,
+      "step": 1811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928807255590644e-05,
+      "loss": 0.3219,
+      "step": 1812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992866482601976e-05,
+      "loss": 0.3425,
+      "step": 1813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928522254627534e-05,
+      "loss": 0.3135,
+      "step": 1814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928379541416002e-05,
+      "loss": 0.3728,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928236686387205e-05,
+      "loss": 0.313,
+      "step": 1816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992809368954318e-05,
+      "loss": 0.3396,
+      "step": 1817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927950550885976e-05,
+      "loss": 0.3525,
+      "step": 1818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992780727041763e-05,
+      "loss": 0.301,
+      "step": 1819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927663848140197e-05,
+      "loss": 0.3054,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927520284055718e-05,
+      "loss": 0.3561,
+      "step": 1821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927376578166248e-05,
+      "loss": 0.3441,
+      "step": 1822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927232730473836e-05,
+      "loss": 0.3344,
+      "step": 1823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 0.3383,
+      "step": 1824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926944609688418e-05,
+      "loss": 0.3116,
+      "step": 1825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926800336599523e-05,
+      "loss": 0.3425,
+      "step": 1826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926655921715924e-05,
+      "loss": 0.3512,
+      "step": 1827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926511365039674e-05,
+      "loss": 0.3226,
+      "step": 1828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992636666657285e-05,
+      "loss": 0.3307,
+      "step": 1829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926221826317503e-05,
+      "loss": 0.2984,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926076844275716e-05,
+      "loss": 0.3346,
+      "step": 1831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925931720449555e-05,
+      "loss": 0.344,
+      "step": 1832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992578645484109e-05,
+      "loss": 0.3189,
+      "step": 1833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99256410474524e-05,
+      "loss": 0.3086,
+      "step": 1834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992549549828556e-05,
+      "loss": 0.3182,
+      "step": 1835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925349807342655e-05,
+      "loss": 0.3627,
+      "step": 1836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925203974625754e-05,
+      "loss": 0.2962,
+      "step": 1837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992505800013695e-05,
+      "loss": 0.3563,
+      "step": 1838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924911883878324e-05,
+      "loss": 0.3084,
+      "step": 1839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924765625851964e-05,
+      "loss": 0.3213,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992461922605996e-05,
+      "loss": 0.3452,
+      "step": 1841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99244726845044e-05,
+      "loss": 0.3062,
+      "step": 1842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924326001187384e-05,
+      "loss": 0.3214,
+      "step": 1843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924179176111e-05,
+      "loss": 0.3739,
+      "step": 1844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924032209277347e-05,
+      "loss": 0.3398,
+      "step": 1845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992388510068853e-05,
+      "loss": 0.3326,
+      "step": 1846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923737850346643e-05,
+      "loss": 0.3215,
+      "step": 1847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923590458253792e-05,
+      "loss": 0.3135,
+      "step": 1848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992344292441208e-05,
+      "loss": 0.313,
+      "step": 1849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992329524882362e-05,
+      "loss": 0.3417,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923147431490517e-05,
+      "loss": 0.3524,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922999472414885e-05,
+      "loss": 0.3327,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922851371598834e-05,
+      "loss": 0.3427,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922703129044483e-05,
+      "loss": 0.3264,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922554744753948e-05,
+      "loss": 0.3599,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992240621872935e-05,
+      "loss": 0.3461,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922257550972804e-05,
+      "loss": 0.2867,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922108741486444e-05,
+      "loss": 0.3488,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921959790272386e-05,
+      "loss": 0.3723,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921810697332763e-05,
+      "loss": 0.3646,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921661462669704e-05,
+      "loss": 0.3396,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992151208628534e-05,
+      "loss": 0.3621,
+      "step": 1862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921362568181805e-05,
+      "loss": 0.3468,
+      "step": 1863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921212908361237e-05,
+      "loss": 0.3286,
+      "step": 1864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992106310682577e-05,
+      "loss": 0.345,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920913163577542e-05,
+      "loss": 0.3021,
+      "step": 1866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920763078618704e-05,
+      "loss": 0.3543,
+      "step": 1867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992061285195139e-05,
+      "loss": 0.2861,
+      "step": 1868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920462483577746e-05,
+      "loss": 0.344,
+      "step": 1869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992031197349993e-05,
+      "loss": 0.3554,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992016132172008e-05,
+      "loss": 0.3371,
+      "step": 1871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920010528240357e-05,
+      "loss": 0.3183,
+      "step": 1872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991985959306291e-05,
+      "loss": 0.3218,
+      "step": 1873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919708516189898e-05,
+      "loss": 0.2829,
+      "step": 1874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919557297623475e-05,
+      "loss": 0.336,
+      "step": 1875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919405937365804e-05,
+      "loss": 0.3286,
+      "step": 1876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919254435419043e-05,
+      "loss": 0.3092,
+      "step": 1877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991910279178536e-05,
+      "loss": 0.333,
+      "step": 1878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918951006466927e-05,
+      "loss": 0.3137,
+      "step": 1879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99187990794659e-05,
+      "loss": 0.3,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918647010784455e-05,
+      "loss": 0.3101,
+      "step": 1881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918494800424762e-05,
+      "loss": 0.3524,
+      "step": 1882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918342448389003e-05,
+      "loss": 0.2896,
+      "step": 1883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918189954679344e-05,
+      "loss": 0.3085,
+      "step": 1884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918037319297966e-05,
+      "loss": 0.328,
+      "step": 1885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991788454224705e-05,
+      "loss": 0.3433,
+      "step": 1886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991773162352878e-05,
+      "loss": 0.3057,
+      "step": 1887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917578563145342e-05,
+      "loss": 0.3017,
+      "step": 1888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991742536109892e-05,
+      "loss": 0.3174,
+      "step": 1889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917272017391697e-05,
+      "loss": 0.337,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991711853202587e-05,
+      "loss": 0.3259,
+      "step": 1891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916964905003632e-05,
+      "loss": 0.3249,
+      "step": 1892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916811136327173e-05,
+      "loss": 0.3417,
+      "step": 1893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991665722599869e-05,
+      "loss": 0.3318,
+      "step": 1894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916503174020384e-05,
+      "loss": 0.3093,
+      "step": 1895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916348980394455e-05,
+      "loss": 0.3625,
+      "step": 1896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916194645123106e-05,
+      "loss": 0.3279,
+      "step": 1897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916040168208536e-05,
+      "loss": 0.3141,
+      "step": 1898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991588554965296e-05,
+      "loss": 0.3262,
+      "step": 1899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991573078945858e-05,
+      "loss": 0.327,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915575887627612e-05,
+      "loss": 0.3134,
+      "step": 1901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915420844162265e-05,
+      "loss": 0.3405,
+      "step": 1902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915265659064752e-05,
+      "loss": 0.3473,
+      "step": 1903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915110332337293e-05,
+      "loss": 0.3664,
+      "step": 1904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914954863982106e-05,
+      "loss": 0.3184,
+      "step": 1905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914799254001415e-05,
+      "loss": 0.3302,
+      "step": 1906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991464350239743e-05,
+      "loss": 0.3268,
+      "step": 1907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914487609172396e-05,
+      "loss": 0.3373,
+      "step": 1908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914331574328523e-05,
+      "loss": 0.3304,
+      "step": 1909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914175397868046e-05,
+      "loss": 0.3494,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914019079793195e-05,
+      "loss": 0.3704,
+      "step": 1911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913862620106202e-05,
+      "loss": 0.3735,
+      "step": 1912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991370601880931e-05,
+      "loss": 0.3166,
+      "step": 1913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913549275904742e-05,
+      "loss": 0.3333,
+      "step": 1914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913392391394744e-05,
+      "loss": 0.33,
+      "step": 1915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991323536528156e-05,
+      "loss": 0.2905,
+      "step": 1916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991307819756743e-05,
+      "loss": 0.3287,
+      "step": 1917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912920888254597e-05,
+      "loss": 0.3484,
+      "step": 1918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991276343734531e-05,
+      "loss": 0.3182,
+      "step": 1919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912605844841815e-05,
+      "loss": 0.3108,
+      "step": 1920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991244811074637e-05,
+      "loss": 0.3222,
+      "step": 1921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912290235061226e-05,
+      "loss": 0.3216,
+      "step": 1922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991213221778863e-05,
+      "loss": 0.3368,
+      "step": 1923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991197405893085e-05,
+      "loss": 0.3203,
+      "step": 1924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991181575849014e-05,
+      "loss": 0.3145,
+      "step": 1925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911657316468758e-05,
+      "loss": 0.3202,
+      "step": 1926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991149873286897e-05,
+      "loss": 0.311,
+      "step": 1927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991134000769305e-05,
+      "loss": 0.3243,
+      "step": 1928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991118114094325e-05,
+      "loss": 0.3235,
+      "step": 1929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991102213262185e-05,
+      "loss": 0.3467,
+      "step": 1930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910862982731114e-05,
+      "loss": 0.3141,
+      "step": 1931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991070369127332e-05,
+      "loss": 0.3438,
+      "step": 1932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910544258250746e-05,
+      "loss": 0.3586,
+      "step": 1933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910384683665663e-05,
+      "loss": 0.3128,
+      "step": 1934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910224967520352e-05,
+      "loss": 0.2974,
+      "step": 1935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99100651098171e-05,
+      "loss": 0.2912,
+      "step": 1936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.990990511055818e-05,
+      "loss": 0.3635,
+      "step": 1937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909744969745887e-05,
+      "loss": 0.3239,
+      "step": 1938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909584687382503e-05,
+      "loss": 0.3249,
+      "step": 1939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909424263470322e-05,
+      "loss": 0.2961,
+      "step": 1940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909263698011627e-05,
+      "loss": 0.359,
+      "step": 1941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909102991008722e-05,
+      "loss": 0.3284,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908942142463898e-05,
+      "loss": 0.3282,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990878115237945e-05,
+      "loss": 0.3075,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990862002075768e-05,
+      "loss": 0.3315,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990845874760089e-05,
+      "loss": 0.3052,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908297332911384e-05,
+      "loss": 0.3325,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908135776691468e-05,
+      "loss": 0.3203,
+      "step": 1948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907974078943442e-05,
+      "loss": 0.3327,
+      "step": 1949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990781223966963e-05,
+      "loss": 0.3399,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907650258872336e-05,
+      "loss": 0.3213,
+      "step": 1951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907488136553867e-05,
+      "loss": 0.3414,
+      "step": 1952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907325872716548e-05,
+      "loss": 0.314,
+      "step": 1953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907163467362697e-05,
+      "loss": 0.3135,
+      "step": 1954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907000920494632e-05,
+      "loss": 0.348,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990683823211467e-05,
+      "loss": 0.3324,
+      "step": 1956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906675402225137e-05,
+      "loss": 0.3312,
+      "step": 1957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906512430828363e-05,
+      "loss": 0.3205,
+      "step": 1958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906349317926678e-05,
+      "loss": 0.329,
+      "step": 1959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99061860635224e-05,
+      "loss": 0.2885,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906022667617873e-05,
+      "loss": 0.3148,
+      "step": 1961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905859130215425e-05,
+      "loss": 0.3487,
+      "step": 1962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905695451317394e-05,
+      "loss": 0.3351,
+      "step": 1963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905531630926114e-05,
+      "loss": 0.3147,
+      "step": 1964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990536766904393e-05,
+      "loss": 0.3317,
+      "step": 1965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990520356567318e-05,
+      "loss": 0.3172,
+      "step": 1966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905039320816212e-05,
+      "loss": 0.3239,
+      "step": 1967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904874934475372e-05,
+      "loss": 0.3653,
+      "step": 1968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904710406653007e-05,
+      "loss": 0.3287,
+      "step": 1969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904545737351463e-05,
+      "loss": 0.31,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99043809265731e-05,
+      "loss": 0.3279,
+      "step": 1971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904215974320267e-05,
+      "loss": 0.3488,
+      "step": 1972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904050880595318e-05,
+      "loss": 0.3396,
+      "step": 1973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903885645400617e-05,
+      "loss": 0.3214,
+      "step": 1974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990372026873852e-05,
+      "loss": 0.3623,
+      "step": 1975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990355475061139e-05,
+      "loss": 0.3427,
+      "step": 1976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903389091021594e-05,
+      "loss": 0.3495,
+      "step": 1977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903223289971496e-05,
+      "loss": 0.3178,
+      "step": 1978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903057347463464e-05,
+      "loss": 0.3259,
+      "step": 1979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990289126349987e-05,
+      "loss": 0.3051,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902725038083083e-05,
+      "loss": 0.3307,
+      "step": 1981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990255867121548e-05,
+      "loss": 0.317,
+      "step": 1982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990239216289944e-05,
+      "loss": 0.2945,
+      "step": 1983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902225513137336e-05,
+      "loss": 0.3277,
+      "step": 1984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902058721931553e-05,
+      "loss": 0.3293,
+      "step": 1985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990189178928447e-05,
+      "loss": 0.3317,
+      "step": 1986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990172471519847e-05,
+      "loss": 0.3011,
+      "step": 1987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901557499675945e-05,
+      "loss": 0.2914,
+      "step": 1988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990139014271928e-05,
+      "loss": 0.299,
+      "step": 1989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990122264433087e-05,
+      "loss": 0.3233,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901055004513102e-05,
+      "loss": 0.3176,
+      "step": 1991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900887223268367e-05,
+      "loss": 0.366,
+      "step": 1992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900719300599073e-05,
+      "loss": 0.325,
+      "step": 1993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990055123650761e-05,
+      "loss": 0.3101,
+      "step": 1994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900383030996385e-05,
+      "loss": 0.2877,
+      "step": 1995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900214684067794e-05,
+      "loss": 0.3123,
+      "step": 1996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900046195724244e-05,
+      "loss": 0.3508,
+      "step": 1997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899877565968143e-05,
+      "loss": 0.3225,
+      "step": 1998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98997087948019e-05,
+      "loss": 0.3175,
+      "step": 1999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899539882227927e-05,
+      "loss": 0.3467,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989937082824863e-05,
+      "loss": 0.2832,
+      "step": 2001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899201632866427e-05,
+      "loss": 0.3258,
+      "step": 2002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989903229608374e-05,
+      "loss": 0.333,
+      "step": 2003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898862817902982e-05,
+      "loss": 0.3431,
+      "step": 2004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989869319832658e-05,
+      "loss": 0.3455,
+      "step": 2005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898523437356947e-05,
+      "loss": 0.3188,
+      "step": 2006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898353534996517e-05,
+      "loss": 0.3416,
+      "step": 2007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989818349124771e-05,
+      "loss": 0.312,
+      "step": 2008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989801330611296e-05,
+      "loss": 0.3282,
+      "step": 2009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897842979594697e-05,
+      "loss": 0.3139,
+      "step": 2010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897672511695352e-05,
+      "loss": 0.3346,
+      "step": 2011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897501902417362e-05,
+      "loss": 0.3077,
+      "step": 2012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897331151763162e-05,
+      "loss": 0.3079,
+      "step": 2013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897160259735193e-05,
+      "loss": 0.3046,
+      "step": 2014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896989226335893e-05,
+      "loss": 0.2972,
+      "step": 2015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896818051567713e-05,
+      "loss": 0.2956,
+      "step": 2016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896646735433088e-05,
+      "loss": 0.3029,
+      "step": 2017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896475277934465e-05,
+      "loss": 0.3325,
+      "step": 2018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896303679074306e-05,
+      "loss": 0.3007,
+      "step": 2019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896131938855048e-05,
+      "loss": 0.3038,
+      "step": 2020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989596005727915e-05,
+      "loss": 0.354,
+      "step": 2021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989578803434907e-05,
+      "loss": 0.3171,
+      "step": 2022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895615870067257e-05,
+      "loss": 0.3273,
+      "step": 2023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989544356443618e-05,
+      "loss": 0.3336,
+      "step": 2024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895271117458296e-05,
+      "loss": 0.2919,
+      "step": 2025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895098529136064e-05,
+      "loss": 0.3459,
+      "step": 2026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894925799471952e-05,
+      "loss": 0.3119,
+      "step": 2027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989475292846843e-05,
+      "loss": 0.3208,
+      "step": 2028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894579916127966e-05,
+      "loss": 0.3008,
+      "step": 2029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989440676245303e-05,
+      "loss": 0.2943,
+      "step": 2030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98942334674461e-05,
+      "loss": 0.3126,
+      "step": 2031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894060031109644e-05,
+      "loss": 0.3452,
+      "step": 2032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893886453446143e-05,
+      "loss": 0.2943,
+      "step": 2033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893712734458076e-05,
+      "loss": 0.3241,
+      "step": 2034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893538874147928e-05,
+      "loss": 0.3056,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893364872518176e-05,
+      "loss": 0.3436,
+      "step": 2036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893190729571308e-05,
+      "loss": 0.3326,
+      "step": 2037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893016445309813e-05,
+      "loss": 0.3038,
+      "step": 2038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892842019736178e-05,
+      "loss": 0.3146,
+      "step": 2039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892667452852897e-05,
+      "loss": 0.3436,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892492744662464e-05,
+      "loss": 0.3039,
+      "step": 2041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892317895167372e-05,
+      "loss": 0.3074,
+      "step": 2042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989214290437012e-05,
+      "loss": 0.3195,
+      "step": 2043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891967772273208e-05,
+      "loss": 0.3384,
+      "step": 2044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891792498879132e-05,
+      "loss": 0.332,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891617084190404e-05,
+      "loss": 0.3137,
+      "step": 2046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891441528209525e-05,
+      "loss": 0.2939,
+      "step": 2047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891265830939008e-05,
+      "loss": 0.3724,
+      "step": 2048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989108999238135e-05,
+      "loss": 0.2995,
+      "step": 2049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890914012539077e-05,
+      "loss": 0.3226,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890737891414696e-05,
+      "loss": 0.34,
+      "step": 2051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989056162901072e-05,
+      "loss": 0.3543,
+      "step": 2052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890385225329677e-05,
+      "loss": 0.3056,
+      "step": 2053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890208680374075e-05,
+      "loss": 0.2888,
+      "step": 2054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989003199414644e-05,
+      "loss": 0.3202,
+      "step": 2055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98898551666493e-05,
+      "loss": 0.3104,
+      "step": 2056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889678197885172e-05,
+      "loss": 0.3477,
+      "step": 2057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889501087856594e-05,
+      "loss": 0.319,
+      "step": 2058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988932383656609e-05,
+      "loss": 0.3246,
+      "step": 2059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988914644401619e-05,
+      "loss": 0.3286,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888968910209433e-05,
+      "loss": 0.3131,
+      "step": 2061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888791235148354e-05,
+      "loss": 0.3223,
+      "step": 2062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988861341883549e-05,
+      "loss": 0.36,
+      "step": 2063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888435461273374e-05,
+      "loss": 0.3158,
+      "step": 2064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888257362464558e-05,
+      "loss": 0.3132,
+      "step": 2065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888079122411585e-05,
+      "loss": 0.3344,
+      "step": 2066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887900741116993e-05,
+      "loss": 0.32,
+      "step": 2067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988772221858334e-05,
+      "loss": 0.3252,
+      "step": 2068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988754355481317e-05,
+      "loss": 0.3105,
+      "step": 2069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988736474980904e-05,
+      "loss": 0.319,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887185803573493e-05,
+      "loss": 0.3197,
+      "step": 2071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98870067161091e-05,
+      "loss": 0.3308,
+      "step": 2072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886827487418406e-05,
+      "loss": 0.3426,
+      "step": 2073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988664811750398e-05,
+      "loss": 0.3302,
+      "step": 2074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988646860636838e-05,
+      "loss": 0.3428,
+      "step": 2075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988628895401417e-05,
+      "loss": 0.3159,
+      "step": 2076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886109160443918e-05,
+      "loss": 0.3443,
+      "step": 2077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988592922566019e-05,
+      "loss": 0.3428,
+      "step": 2078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885749149665557e-05,
+      "loss": 0.3255,
+      "step": 2079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885568932462597e-05,
+      "loss": 0.3101,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885388574053874e-05,
+      "loss": 0.3382,
+      "step": 2081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988520807444197e-05,
+      "loss": 0.3312,
+      "step": 2082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885027433629465e-05,
+      "loss": 0.3257,
+      "step": 2083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884846651618935e-05,
+      "loss": 0.3274,
+      "step": 2084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884665728412963e-05,
+      "loss": 0.3413,
+      "step": 2085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884484664014137e-05,
+      "loss": 0.3332,
+      "step": 2086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988430345842504e-05,
+      "loss": 0.3079,
+      "step": 2087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884122111648255e-05,
+      "loss": 0.3145,
+      "step": 2088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883940623686386e-05,
+      "loss": 0.2951,
+      "step": 2089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883758994542017e-05,
+      "loss": 0.3515,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883577224217737e-05,
+      "loss": 0.3415,
+      "step": 2091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883395312716155e-05,
+      "loss": 0.3262,
+      "step": 2092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883213260039862e-05,
+      "loss": 0.3258,
+      "step": 2093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883031066191458e-05,
+      "loss": 0.3318,
+      "step": 2094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882848731173548e-05,
+      "loss": 0.3201,
+      "step": 2095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882666254988734e-05,
+      "loss": 0.3276,
+      "step": 2096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882483637639625e-05,
+      "loss": 0.3617,
+      "step": 2097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882300879128826e-05,
+      "loss": 0.3142,
+      "step": 2098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882117979458956e-05,
+      "loss": 0.291,
+      "step": 2099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881934938632615e-05,
+      "loss": 0.327,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881751756652427e-05,
+      "loss": 0.3179,
+      "step": 2101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881568433521002e-05,
+      "loss": 0.3347,
+      "step": 2102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988138496924097e-05,
+      "loss": 0.3249,
+      "step": 2103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881201363814937e-05,
+      "loss": 0.341,
+      "step": 2104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881017617245538e-05,
+      "loss": 0.3273,
+      "step": 2105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880833729535384e-05,
+      "loss": 0.3028,
+      "step": 2106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988064970068712e-05,
+      "loss": 0.3142,
+      "step": 2107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880465530703358e-05,
+      "loss": 0.3268,
+      "step": 2108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880281219586734e-05,
+      "loss": 0.3113,
+      "step": 2109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880096767339885e-05,
+      "loss": 0.3397,
+      "step": 2110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987991217396544e-05,
+      "loss": 0.3389,
+      "step": 2111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879727439466042e-05,
+      "loss": 0.3261,
+      "step": 2112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879542563844324e-05,
+      "loss": 0.3169,
+      "step": 2113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879357547102928e-05,
+      "loss": 0.3372,
+      "step": 2114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879172389244495e-05,
+      "loss": 0.2938,
+      "step": 2115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878987090271677e-05,
+      "loss": 0.3176,
+      "step": 2116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878801650187115e-05,
+      "loss": 0.3163,
+      "step": 2117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878616068993456e-05,
+      "loss": 0.3161,
+      "step": 2118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878430346693358e-05,
+      "loss": 0.326,
+      "step": 2119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878244483289467e-05,
+      "loss": 0.3205,
+      "step": 2120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878058478784442e-05,
+      "loss": 0.3319,
+      "step": 2121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987787233318094e-05,
+      "loss": 0.2895,
+      "step": 2122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987768604648162e-05,
+      "loss": 0.3232,
+      "step": 2123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877499618689136e-05,
+      "loss": 0.3234,
+      "step": 2124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877313049806156e-05,
+      "loss": 0.3296,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877126339835352e-05,
+      "loss": 0.3103,
+      "step": 2126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987693948877938e-05,
+      "loss": 0.3108,
+      "step": 2127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987675249664091e-05,
+      "loss": 0.3223,
+      "step": 2128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876565363422623e-05,
+      "loss": 0.3164,
+      "step": 2129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987637808912718e-05,
+      "loss": 0.3024,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876190673757264e-05,
+      "loss": 0.3133,
+      "step": 2131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876003117315546e-05,
+      "loss": 0.353,
+      "step": 2132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875815419804712e-05,
+      "loss": 0.3233,
+      "step": 2133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875627581227436e-05,
+      "loss": 0.3245,
+      "step": 2134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875439601586407e-05,
+      "loss": 0.3148,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987525148088431e-05,
+      "loss": 0.3067,
+      "step": 2136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875063219123826e-05,
+      "loss": 0.3635,
+      "step": 2137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874874816307648e-05,
+      "loss": 0.3264,
+      "step": 2138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874686272438467e-05,
+      "loss": 0.3017,
+      "step": 2139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987449758751898e-05,
+      "loss": 0.3198,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874308761551872e-05,
+      "loss": 0.3401,
+      "step": 2141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987411979453985e-05,
+      "loss": 0.271,
+      "step": 2142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873930686485613e-05,
+      "loss": 0.311,
+      "step": 2143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873741437391854e-05,
+      "loss": 0.2946,
+      "step": 2144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873552047261282e-05,
+      "loss": 0.3648,
+      "step": 2145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873362516096603e-05,
+      "loss": 0.3274,
+      "step": 2146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873172843900523e-05,
+      "loss": 0.3331,
+      "step": 2147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987298303067575e-05,
+      "loss": 0.3329,
+      "step": 2148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872793076424996e-05,
+      "loss": 0.3024,
+      "step": 2149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872602981150978e-05,
+      "loss": 0.3538,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872412744856402e-05,
+      "loss": 0.3553,
+      "step": 2151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872222367543997e-05,
+      "loss": 0.3066,
+      "step": 2152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872031849216473e-05,
+      "loss": 0.3158,
+      "step": 2153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871841189876557e-05,
+      "loss": 0.3172,
+      "step": 2154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987165038952697e-05,
+      "loss": 0.3165,
+      "step": 2155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871459448170437e-05,
+      "loss": 0.3256,
+      "step": 2156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871268365809687e-05,
+      "loss": 0.2902,
+      "step": 2157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987107714244745e-05,
+      "loss": 0.2958,
+      "step": 2158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870885778086456e-05,
+      "loss": 0.3612,
+      "step": 2159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987069427272944e-05,
+      "loss": 0.3142,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 0.2979,
+      "step": 2161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870310839038273e-05,
+      "loss": 0.3084,
+      "step": 2162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870118910709605e-05,
+      "loss": 0.3399,
+      "step": 2163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869926841395867e-05,
+      "loss": 0.3258,
+      "step": 2164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869734631099807e-05,
+      "loss": 0.3275,
+      "step": 2165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986954227982416e-05,
+      "loss": 0.3317,
+      "step": 2166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869349787571685e-05,
+      "loss": 0.3087,
+      "step": 2167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869157154345126e-05,
+      "loss": 0.3482,
+      "step": 2168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868964380147232e-05,
+      "loss": 0.333,
+      "step": 2169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868771464980762e-05,
+      "loss": 0.3209,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986857840884847e-05,
+      "loss": 0.3048,
+      "step": 2171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868385211753117e-05,
+      "loss": 0.3394,
+      "step": 2172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868191873697452e-05,
+      "loss": 0.3452,
+      "step": 2173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867998394684246e-05,
+      "loss": 0.3054,
+      "step": 2174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986780477471626e-05,
+      "loss": 0.3269,
+      "step": 2175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867611013796262e-05,
+      "loss": 0.2596,
+      "step": 2176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867417111927014e-05,
+      "loss": 0.3234,
+      "step": 2177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986722306911129e-05,
+      "loss": 0.3356,
+      "step": 2178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867028885351862e-05,
+      "loss": 0.328,
+      "step": 2179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98668345606515e-05,
+      "loss": 0.3194,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866640095012987e-05,
+      "loss": 0.3196,
+      "step": 2181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866445488439092e-05,
+      "loss": 0.3175,
+      "step": 2182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98662507409326e-05,
+      "loss": 0.3091,
+      "step": 2183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866055852496292e-05,
+      "loss": 0.3164,
+      "step": 2184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986586082313295e-05,
+      "loss": 0.3692,
+      "step": 2185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986566565284536e-05,
+      "loss": 0.3209,
+      "step": 2186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865470341636315e-05,
+      "loss": 0.3521,
+      "step": 2187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98652748895086e-05,
+      "loss": 0.3314,
+      "step": 2188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865079296465004e-05,
+      "loss": 0.3615,
+      "step": 2189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986488356250833e-05,
+      "loss": 0.3233,
+      "step": 2190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864687687641365e-05,
+      "loss": 0.3269,
+      "step": 2191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986449167186691e-05,
+      "loss": 0.322,
+      "step": 2192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864295515187765e-05,
+      "loss": 0.3105,
+      "step": 2193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864099217606734e-05,
+      "loss": 0.3,
+      "step": 2194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863902779126618e-05,
+      "loss": 0.3173,
+      "step": 2195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863706199750224e-05,
+      "loss": 0.3381,
+      "step": 2196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986350947948036e-05,
+      "loss": 0.3683,
+      "step": 2197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863312618319838e-05,
+      "loss": 0.3108,
+      "step": 2198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863115616271466e-05,
+      "loss": 0.3257,
+      "step": 2199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986291847333806e-05,
+      "loss": 0.3152,
+      "step": 2200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862721189522432e-05,
+      "loss": 0.3053,
+      "step": 2201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862523764827405e-05,
+      "loss": 0.3356,
+      "step": 2202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862326199255803e-05,
+      "loss": 0.3259,
+      "step": 2203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986212849281044e-05,
+      "loss": 0.302,
+      "step": 2204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861930645494142e-05,
+      "loss": 0.3331,
+      "step": 2205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861732657309735e-05,
+      "loss": 0.3051,
+      "step": 2206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861534528260052e-05,
+      "loss": 0.302,
+      "step": 2207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861336258347913e-05,
+      "loss": 0.3113,
+      "step": 2208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861137847576163e-05,
+      "loss": 0.3087,
+      "step": 2209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860939295947625e-05,
+      "loss": 0.3285,
+      "step": 2210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986074060346514e-05,
+      "loss": 0.3257,
+      "step": 2211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860541770131547e-05,
+      "loss": 0.3047,
+      "step": 2212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860342795949683e-05,
+      "loss": 0.3125,
+      "step": 2213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860143680922394e-05,
+      "loss": 0.3099,
+      "step": 2214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985994442505252e-05,
+      "loss": 0.3203,
+      "step": 2215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985974502834291e-05,
+      "loss": 0.3144,
+      "step": 2216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859545490796414e-05,
+      "loss": 0.352,
+      "step": 2217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859345812415874e-05,
+      "loss": 0.3176,
+      "step": 2218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859145993204153e-05,
+      "loss": 0.3545,
+      "step": 2219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858946033164098e-05,
+      "loss": 0.2879,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858745932298563e-05,
+      "loss": 0.3152,
+      "step": 2221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858545690610417e-05,
+      "loss": 0.2679,
+      "step": 2222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985834530810251e-05,
+      "loss": 0.2999,
+      "step": 2223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858144784777712e-05,
+      "loss": 0.3243,
+      "step": 2224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985794412063888e-05,
+      "loss": 0.3347,
+      "step": 2225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857743315688884e-05,
+      "loss": 0.3089,
+      "step": 2226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985754236993059e-05,
+      "loss": 0.3056,
+      "step": 2227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857341283366872e-05,
+      "loss": 0.3184,
+      "step": 2228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857140056000604e-05,
+      "loss": 0.2965,
+      "step": 2229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856938687834653e-05,
+      "loss": 0.3233,
+      "step": 2230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98567371788719e-05,
+      "loss": 0.3127,
+      "step": 2231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856535529115222e-05,
+      "loss": 0.3302,
+      "step": 2232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98563337385675e-05,
+      "loss": 0.3313,
+      "step": 2233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856131807231617e-05,
+      "loss": 0.3324,
+      "step": 2234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985592973511046e-05,
+      "loss": 0.3027,
+      "step": 2235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855727522206906e-05,
+      "loss": 0.3331,
+      "step": 2236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855525168523854e-05,
+      "loss": 0.3005,
+      "step": 2237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855322674064187e-05,
+      "loss": 0.3035,
+      "step": 2238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98551200388308e-05,
+      "loss": 0.3058,
+      "step": 2239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985491726282659e-05,
+      "loss": 0.3302,
+      "step": 2240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985471434605445e-05,
+      "loss": 0.3181,
+      "step": 2241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985451128851728e-05,
+      "loss": 0.3334,
+      "step": 2242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854308090217983e-05,
+      "loss": 0.3414,
+      "step": 2243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854104751159454e-05,
+      "loss": 0.325,
+      "step": 2244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853901271344605e-05,
+      "loss": 0.3191,
+      "step": 2245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853697650776343e-05,
+      "loss": 0.3336,
+      "step": 2246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853493889457568e-05,
+      "loss": 0.3483,
+      "step": 2247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853289987391197e-05,
+      "loss": 0.2909,
+      "step": 2248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853085944580145e-05,
+      "loss": 0.3525,
+      "step": 2249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852881761027318e-05,
+      "loss": 0.3315,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985267743673564e-05,
+      "loss": 0.3033,
+      "step": 2251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852472971708025e-05,
+      "loss": 0.3036,
+      "step": 2252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852268365947397e-05,
+      "loss": 0.3376,
+      "step": 2253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985206361945668e-05,
+      "loss": 0.3178,
+      "step": 2254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985185873223879e-05,
+      "loss": 0.301,
+      "step": 2255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851653704296664e-05,
+      "loss": 0.3481,
+      "step": 2256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851448535633226e-05,
+      "loss": 0.3216,
+      "step": 2257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851243226251406e-05,
+      "loss": 0.3279,
+      "step": 2258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851037776154136e-05,
+      "loss": 0.3028,
+      "step": 2259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850832185344355e-05,
+      "loss": 0.2935,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850626453824993e-05,
+      "loss": 0.3426,
+      "step": 2261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850420581599e-05,
+      "loss": 0.3307,
+      "step": 2262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850214568669298e-05,
+      "loss": 0.3107,
+      "step": 2263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985000841503885e-05,
+      "loss": 0.3019,
+      "step": 2264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849802120710584e-05,
+      "loss": 0.3157,
+      "step": 2265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984959568568746e-05,
+      "loss": 0.3302,
+      "step": 2266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849389109972415e-05,
+      "loss": 0.3185,
+      "step": 2267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984918239356841e-05,
+      "loss": 0.3105,
+      "step": 2268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848975536478394e-05,
+      "loss": 0.2969,
+      "step": 2269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984876853870532e-05,
+      "loss": 0.329,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848561400252143e-05,
+      "loss": 0.343,
+      "step": 2271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848354121121827e-05,
+      "loss": 0.2908,
+      "step": 2272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984814670131733e-05,
+      "loss": 0.316,
+      "step": 2273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847939140841616e-05,
+      "loss": 0.3346,
+      "step": 2274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847731439697647e-05,
+      "loss": 0.3038,
+      "step": 2275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847523597888396e-05,
+      "loss": 0.3232,
+      "step": 2276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847315615416826e-05,
+      "loss": 0.2997,
+      "step": 2277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847107492285907e-05,
+      "loss": 0.3433,
+      "step": 2278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984689922849862e-05,
+      "loss": 0.3279,
+      "step": 2279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846690824057928e-05,
+      "loss": 0.303,
+      "step": 2280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846482278966816e-05,
+      "loss": 0.3318,
+      "step": 2281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984627359322826e-05,
+      "loss": 0.3188,
+      "step": 2282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846064766845244e-05,
+      "loss": 0.3161,
+      "step": 2283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845855799820746e-05,
+      "loss": 0.3098,
+      "step": 2284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845646692157755e-05,
+      "loss": 0.3078,
+      "step": 2285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984543744385926e-05,
+      "loss": 0.3403,
+      "step": 2286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984522805492824e-05,
+      "loss": 0.2852,
+      "step": 2287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845018525367696e-05,
+      "loss": 0.3186,
+      "step": 2288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844808855180617e-05,
+      "loss": 0.2961,
+      "step": 2289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844599044369995e-05,
+      "loss": 0.3181,
+      "step": 2290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844389092938833e-05,
+      "loss": 0.3041,
+      "step": 2291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844179000890128e-05,
+      "loss": 0.3135,
+      "step": 2292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843968768226876e-05,
+      "loss": 0.3039,
+      "step": 2293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984375839495209e-05,
+      "loss": 0.3146,
+      "step": 2294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843547881068763e-05,
+      "loss": 0.3431,
+      "step": 2295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843337226579912e-05,
+      "loss": 0.3228,
+      "step": 2296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984312643148854e-05,
+      "loss": 0.3368,
+      "step": 2297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842915495797658e-05,
+      "loss": 0.3698,
+      "step": 2298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842704419510282e-05,
+      "loss": 0.3173,
+      "step": 2299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842493202629427e-05,
+      "loss": 0.3142,
+      "step": 2300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984228184515811e-05,
+      "loss": 0.3061,
+      "step": 2301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842070347099345e-05,
+      "loss": 0.3075,
+      "step": 2302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9841858708456164e-05,
+      "loss": 0.3287,
+      "step": 2303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841646929231578e-05,
+      "loss": 0.2957,
+      "step": 2304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841435009428617e-05,
+      "loss": 0.3337,
+      "step": 2305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984122294905031e-05,
+      "loss": 0.2881,
+      "step": 2306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841010748099685e-05,
+      "loss": 0.3244,
+      "step": 2307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984079840657977e-05,
+      "loss": 0.333,
+      "step": 2308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840585924493602e-05,
+      "loss": 0.3404,
+      "step": 2309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840373301844213e-05,
+      "loss": 0.3158,
+      "step": 2310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840160538634645e-05,
+      "loss": 0.3028,
+      "step": 2311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983994763486793e-05,
+      "loss": 0.3407,
+      "step": 2312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839734590547114e-05,
+      "loss": 0.3254,
+      "step": 2313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983952140567524e-05,
+      "loss": 0.3242,
+      "step": 2314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839308080255352e-05,
+      "loss": 0.3322,
+      "step": 2315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839094614290497e-05,
+      "loss": 0.3031,
+      "step": 2316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838881007783727e-05,
+      "loss": 0.2995,
+      "step": 2317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838667260738086e-05,
+      "loss": 0.3302,
+      "step": 2318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838453373156636e-05,
+      "loss": 0.3034,
+      "step": 2319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838239345042424e-05,
+      "loss": 0.3577,
+      "step": 2320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838025176398515e-05,
+      "loss": 0.3411,
+      "step": 2321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837810867227963e-05,
+      "loss": 0.333,
+      "step": 2322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983759641753383e-05,
+      "loss": 0.3043,
+      "step": 2323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983738182731918e-05,
+      "loss": 0.2942,
+      "step": 2324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983716709658708e-05,
+      "loss": 0.3209,
+      "step": 2325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836952225340592e-05,
+      "loss": 0.3164,
+      "step": 2326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836737213582787e-05,
+      "loss": 0.3129,
+      "step": 2327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836522061316745e-05,
+      "loss": 0.3416,
+      "step": 2328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836306768545524e-05,
+      "loss": 0.3238,
+      "step": 2329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983609133527221e-05,
+      "loss": 0.3356,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835875761499878e-05,
+      "loss": 0.3456,
+      "step": 2331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983566004723161e-05,
+      "loss": 0.2822,
+      "step": 2332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983544419247048e-05,
+      "loss": 0.3222,
+      "step": 2333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835228197219573e-05,
+      "loss": 0.3053,
+      "step": 2334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835012061481983e-05,
+      "loss": 0.3041,
+      "step": 2335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834795785260784e-05,
+      "loss": 0.3002,
+      "step": 2336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834579368559076e-05,
+      "loss": 0.2925,
+      "step": 2337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834362811379945e-05,
+      "loss": 0.3052,
+      "step": 2338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983414611372649e-05,
+      "loss": 0.3,
+      "step": 2339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833929275601796e-05,
+      "loss": 0.3383,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833712297008968e-05,
+      "loss": 0.3164,
+      "step": 2341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833495177951104e-05,
+      "loss": 0.3137,
+      "step": 2342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833277918431305e-05,
+      "loss": 0.3363,
+      "step": 2343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833060518452677e-05,
+      "loss": 0.2944,
+      "step": 2344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983284297801832e-05,
+      "loss": 0.3196,
+      "step": 2345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832625297131344e-05,
+      "loss": 0.334,
+      "step": 2346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983240747579486e-05,
+      "loss": 0.3211,
+      "step": 2347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832189514011976e-05,
+      "loss": 0.2851,
+      "step": 2348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983197141178581e-05,
+      "loss": 0.3234,
+      "step": 2349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831753169119473e-05,
+      "loss": 0.291,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983153478601608e-05,
+      "loss": 0.3449,
+      "step": 2351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831316262478758e-05,
+      "loss": 0.2993,
+      "step": 2352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831097598510623e-05,
+      "loss": 0.3187,
+      "step": 2353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830878794114804e-05,
+      "loss": 0.3414,
+      "step": 2354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983065984929442e-05,
+      "loss": 0.3285,
+      "step": 2355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98304407640526e-05,
+      "loss": 0.3219,
+      "step": 2356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830221538392477e-05,
+      "loss": 0.3464,
+      "step": 2357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830002172317178e-05,
+      "loss": 0.3429,
+      "step": 2358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982978266582984e-05,
+      "loss": 0.307,
+      "step": 2359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982956301893359e-05,
+      "loss": 0.304,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982934323163158e-05,
+      "loss": 0.3178,
+      "step": 2361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982912330392694e-05,
+      "loss": 0.3242,
+      "step": 2362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982890323582281e-05,
+      "loss": 0.3172,
+      "step": 2363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982868302732234e-05,
+      "loss": 0.3629,
+      "step": 2364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828462678428672e-05,
+      "loss": 0.308,
+      "step": 2365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982824218914495e-05,
+      "loss": 0.3396,
+      "step": 2366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828021559474326e-05,
+      "loss": 0.3094,
+      "step": 2367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827800789419955e-05,
+      "loss": 0.3618,
+      "step": 2368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827579878984987e-05,
+      "loss": 0.3069,
+      "step": 2369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982735882817258e-05,
+      "loss": 0.3288,
+      "step": 2370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982713763698589e-05,
+      "loss": 0.3059,
+      "step": 2371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826916305428075e-05,
+      "loss": 0.3159,
+      "step": 2372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826694833502295e-05,
+      "loss": 0.3233,
+      "step": 2373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982647322121172e-05,
+      "loss": 0.3336,
+      "step": 2374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826251468559513e-05,
+      "loss": 0.3228,
+      "step": 2375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826029575548834e-05,
+      "loss": 0.3096,
+      "step": 2376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825807542182867e-05,
+      "loss": 0.3055,
+      "step": 2377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825585368464768e-05,
+      "loss": 0.3019,
+      "step": 2378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825363054397723e-05,
+      "loss": 0.3247,
+      "step": 2379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98251405999849e-05,
+      "loss": 0.3179,
+      "step": 2380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982491800522948e-05,
+      "loss": 0.3278,
+      "step": 2381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824695270134643e-05,
+      "loss": 0.3234,
+      "step": 2382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824472394703567e-05,
+      "loss": 0.3266,
+      "step": 2383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982424937893944e-05,
+      "loss": 0.313,
+      "step": 2384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824026222845444e-05,
+      "loss": 0.2943,
+      "step": 2385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823802926424766e-05,
+      "loss": 0.3258,
+      "step": 2386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823579489680603e-05,
+      "loss": 0.296,
+      "step": 2387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823355912616134e-05,
+      "loss": 0.2923,
+      "step": 2388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823132195234563e-05,
+      "loss": 0.2863,
+      "step": 2389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982290833753908e-05,
+      "loss": 0.3164,
+      "step": 2390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822684339532888e-05,
+      "loss": 0.3177,
+      "step": 2391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822460201219184e-05,
+      "loss": 0.3165,
+      "step": 2392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822235922601165e-05,
+      "loss": 0.318,
+      "step": 2393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982201150368204e-05,
+      "loss": 0.3342,
+      "step": 2394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821786944465015e-05,
+      "loss": 0.3478,
+      "step": 2395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821562244953295e-05,
+      "loss": 0.296,
+      "step": 2396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982133740515009e-05,
+      "loss": 0.3236,
+      "step": 2397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982111242505861e-05,
+      "loss": 0.3382,
+      "step": 2398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820887304682074e-05,
+      "loss": 0.3133,
+      "step": 2399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820662044023695e-05,
+      "loss": 0.3242,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820436643086688e-05,
+      "loss": 0.288,
+      "step": 2401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820211101874273e-05,
+      "loss": 0.3121,
+      "step": 2402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981998542038967e-05,
+      "loss": 0.3207,
+      "step": 2403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981975959863611e-05,
+      "loss": 0.3256,
+      "step": 2404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819533636616816e-05,
+      "loss": 0.3231,
+      "step": 2405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981930753433501e-05,
+      "loss": 0.3381,
+      "step": 2406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819081291793932e-05,
+      "loss": 0.2891,
+      "step": 2407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818854908996804e-05,
+      "loss": 0.3235,
+      "step": 2408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818628385946863e-05,
+      "loss": 0.3213,
+      "step": 2409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818401722647344e-05,
+      "loss": 0.2837,
+      "step": 2410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818174919101487e-05,
+      "loss": 0.2685,
+      "step": 2411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817947975312527e-05,
+      "loss": 0.2861,
+      "step": 2412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981772089128371e-05,
+      "loss": 0.331,
+      "step": 2413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981749366701828e-05,
+      "loss": 0.2954,
+      "step": 2414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981726630251948e-05,
+      "loss": 0.3231,
+      "step": 2415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981703879779056e-05,
+      "loss": 0.2979,
+      "step": 2416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816811152834766e-05,
+      "loss": 0.3067,
+      "step": 2417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816583367655355e-05,
+      "loss": 0.3154,
+      "step": 2418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816355442255576e-05,
+      "loss": 0.3213,
+      "step": 2419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816127376638684e-05,
+      "loss": 0.315,
+      "step": 2420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815899170807943e-05,
+      "loss": 0.3658,
+      "step": 2421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815670824766608e-05,
+      "loss": 0.3239,
+      "step": 2422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815442338517942e-05,
+      "loss": 0.3188,
+      "step": 2423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815213712065208e-05,
+      "loss": 0.3391,
+      "step": 2424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981498494541167e-05,
+      "loss": 0.3317,
+      "step": 2425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98147560385606e-05,
+      "loss": 0.3198,
+      "step": 2426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814526991515266e-05,
+      "loss": 0.3109,
+      "step": 2427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981429780427894e-05,
+      "loss": 0.3157,
+      "step": 2428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981406847685489e-05,
+      "loss": 0.3498,
+      "step": 2429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98138390092464e-05,
+      "loss": 0.3254,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813609401456745e-05,
+      "loss": 0.3108,
+      "step": 2431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98133796534892e-05,
+      "loss": 0.3385,
+      "step": 2432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813149765347054e-05,
+      "loss": 0.3204,
+      "step": 2433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812919737033586e-05,
+      "loss": 0.2985,
+      "step": 2434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812689568552086e-05,
+      "loss": 0.3202,
+      "step": 2435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812459259905838e-05,
+      "loss": 0.3314,
+      "step": 2436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981222881109813e-05,
+      "loss": 0.3323,
+      "step": 2437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981199822213226e-05,
+      "loss": 0.3021,
+      "step": 2438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811767493011514e-05,
+      "loss": 0.3312,
+      "step": 2439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98115366237392e-05,
+      "loss": 0.299,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98113056143186e-05,
+      "loss": 0.3195,
+      "step": 2441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811074464753022e-05,
+      "loss": 0.3169,
+      "step": 2442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981084317504577e-05,
+      "loss": 0.3328,
+      "step": 2443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810611745200142e-05,
+      "loss": 0.3065,
+      "step": 2444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981038017521945e-05,
+      "loss": 0.3107,
+      "step": 2445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810148465106994e-05,
+      "loss": 0.2973,
+      "step": 2446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980991661486609e-05,
+      "loss": 0.3234,
+      "step": 2447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980968462450005e-05,
+      "loss": 0.348,
+      "step": 2448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809452494012186e-05,
+      "loss": 0.2928,
+      "step": 2449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809220223405812e-05,
+      "loss": 0.3383,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808987812684247e-05,
+      "loss": 0.3103,
+      "step": 2451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980875526185081e-05,
+      "loss": 0.2996,
+      "step": 2452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808522570908825e-05,
+      "loss": 0.3112,
+      "step": 2453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808289739861614e-05,
+      "loss": 0.3059,
+      "step": 2454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98080567687125e-05,
+      "loss": 0.3229,
+      "step": 2455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980782365746482e-05,
+      "loss": 0.3003,
+      "step": 2456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807590406121892e-05,
+      "loss": 0.3087,
+      "step": 2457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980735701468706e-05,
+      "loss": 0.2979,
+      "step": 2458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807123483163644e-05,
+      "loss": 0.3358,
+      "step": 2459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806889811554994e-05,
+      "loss": 0.3156,
+      "step": 2460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980665599986444e-05,
+      "loss": 0.2835,
+      "step": 2461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806422048095318e-05,
+      "loss": 0.3426,
+      "step": 2462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806187956250976e-05,
+      "loss": 0.307,
+      "step": 2463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805953724334758e-05,
+      "loss": 0.2998,
+      "step": 2464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805719352350007e-05,
+      "loss": 0.3429,
+      "step": 2465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980548484030007e-05,
+      "loss": 0.3039,
+      "step": 2466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98052501881883e-05,
+      "loss": 0.3294,
+      "step": 2467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980501539601805e-05,
+      "loss": 0.362,
+      "step": 2468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804780463792666e-05,
+      "loss": 0.3202,
+      "step": 2469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804545391515514e-05,
+      "loss": 0.3225,
+      "step": 2470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804310179189942e-05,
+      "loss": 0.2914,
+      "step": 2471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804074826819314e-05,
+      "loss": 0.3056,
+      "step": 2472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803839334406997e-05,
+      "loss": 0.328,
+      "step": 2473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980360370195635e-05,
+      "loss": 0.3464,
+      "step": 2474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803367929470734e-05,
+      "loss": 0.2985,
+      "step": 2475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803132016953525e-05,
+      "loss": 0.3051,
+      "step": 2476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802895964408088e-05,
+      "loss": 0.3042,
+      "step": 2477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802659771837794e-05,
+      "loss": 0.3204,
+      "step": 2478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802423439246023e-05,
+      "loss": 0.3309,
+      "step": 2479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802186966636144e-05,
+      "loss": 0.3116,
+      "step": 2480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980195035401154e-05,
+      "loss": 0.3348,
+      "step": 2481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9801713601375584e-05,
+      "loss": 0.3446,
+      "step": 2482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9801476708731665e-05,
+      "loss": 0.2989,
+      "step": 2483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980123967608316e-05,
+      "loss": 0.3102,
+      "step": 2484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801002503433464e-05,
+      "loss": 0.3668,
+      "step": 2485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800765190785958e-05,
+      "loss": 0.3192,
+      "step": 2486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980052773814403e-05,
+      "loss": 0.3108,
+      "step": 2487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980029014551108e-05,
+      "loss": 0.2842,
+      "step": 2488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800052412890492e-05,
+      "loss": 0.3357,
+      "step": 2489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979981454028567e-05,
+      "loss": 0.3208,
+      "step": 2490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799576527700005e-05,
+      "loss": 0.346,
+      "step": 2491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799338375136902e-05,
+      "loss": 0.3168,
+      "step": 2492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799100082599758e-05,
+      "loss": 0.2981,
+      "step": 2493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798861650091984e-05,
+      "loss": 0.3295,
+      "step": 2494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979862307761698e-05,
+      "loss": 0.3319,
+      "step": 2495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798384365178153e-05,
+      "loss": 0.2884,
+      "step": 2496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798145512778916e-05,
+      "loss": 0.333,
+      "step": 2497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979790652042268e-05,
+      "loss": 0.3078,
+      "step": 2498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797667388112856e-05,
+      "loss": 0.299,
+      "step": 2499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797428115852865e-05,
+      "loss": 0.2947,
+      "step": 2500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979718870364612e-05,
+      "loss": 0.305,
+      "step": 2501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796949151496045e-05,
+      "loss": 0.2912,
+      "step": 2502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979670945940606e-05,
+      "loss": 0.2946,
+      "step": 2503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796469627379585e-05,
+      "loss": 0.3172,
+      "step": 2504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796229655420052e-05,
+      "loss": 0.2746,
+      "step": 2505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795989543530886e-05,
+      "loss": 0.3056,
+      "step": 2506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795749291715516e-05,
+      "loss": 0.336,
+      "step": 2507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795508899977374e-05,
+      "loss": 0.3324,
+      "step": 2508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795268368319893e-05,
+      "loss": 0.2639,
+      "step": 2509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979502769674651e-05,
+      "loss": 0.2898,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794786885260663e-05,
+      "loss": 0.3335,
+      "step": 2511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794545933865793e-05,
+      "loss": 0.3214,
+      "step": 2512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979430484256534e-05,
+      "loss": 0.2879,
+      "step": 2513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979406361136275e-05,
+      "loss": 0.2849,
+      "step": 2514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979382224026146e-05,
+      "loss": 0.3079,
+      "step": 2515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979358072926493e-05,
+      "loss": 0.3053,
+      "step": 2516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793339078376605e-05,
+      "loss": 0.3142,
+      "step": 2517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979309728759993e-05,
+      "loss": 0.3289,
+      "step": 2518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792855356938368e-05,
+      "loss": 0.2856,
+      "step": 2519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979261328639537e-05,
+      "loss": 0.3278,
+      "step": 2520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792371075974398e-05,
+      "loss": 0.3238,
+      "step": 2521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792128725678906e-05,
+      "loss": 0.3177,
+      "step": 2522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791886235512362e-05,
+      "loss": 0.3353,
+      "step": 2523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791643605478222e-05,
+      "loss": 0.313,
+      "step": 2524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979140083557996e-05,
+      "loss": 0.3096,
+      "step": 2525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979115792582104e-05,
+      "loss": 0.2568,
+      "step": 2526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979091487620493e-05,
+      "loss": 0.3489,
+      "step": 2527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97906716867351e-05,
+      "loss": 0.3239,
+      "step": 2528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979042835741503e-05,
+      "loss": 0.3277,
+      "step": 2529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790184888248193e-05,
+      "loss": 0.3066,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978994127923807e-05,
+      "loss": 0.3038,
+      "step": 2531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789697530388135e-05,
+      "loss": 0.3249,
+      "step": 2532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978945364170187e-05,
+      "loss": 0.3198,
+      "step": 2533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789209613182758e-05,
+      "loss": 0.3256,
+      "step": 2534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788965444834295e-05,
+      "loss": 0.2827,
+      "step": 2535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788721136659953e-05,
+      "loss": 0.2962,
+      "step": 2536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788476688663234e-05,
+      "loss": 0.2831,
+      "step": 2537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788232100847623e-05,
+      "loss": 0.3278,
+      "step": 2538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978798737321662e-05,
+      "loss": 0.3478,
+      "step": 2539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787742505773715e-05,
+      "loss": 0.294,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787497498522408e-05,
+      "loss": 0.33,
+      "step": 2541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787252351466196e-05,
+      "loss": 0.3025,
+      "step": 2542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787007064608585e-05,
+      "loss": 0.3284,
+      "step": 2543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786761637953074e-05,
+      "loss": 0.3259,
+      "step": 2544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786516071503175e-05,
+      "loss": 0.3093,
+      "step": 2545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978627036526239e-05,
+      "loss": 0.3079,
+      "step": 2546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978602451923423e-05,
+      "loss": 0.2894,
+      "step": 2547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978577853342221e-05,
+      "loss": 0.3212,
+      "step": 2548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978553240782984e-05,
+      "loss": 0.2997,
+      "step": 2549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785286142460635e-05,
+      "loss": 0.3476,
+      "step": 2550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785039737318116e-05,
+      "loss": 0.3062,
+      "step": 2551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97847931924058e-05,
+      "loss": 0.3027,
+      "step": 2552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784546507727208e-05,
+      "loss": 0.3194,
+      "step": 2553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978429968328587e-05,
+      "loss": 0.3091,
+      "step": 2554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97840527190853e-05,
+      "loss": 0.301,
+      "step": 2555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978380561512904e-05,
+      "loss": 0.3083,
+      "step": 2556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783558371420607e-05,
+      "loss": 0.3236,
+      "step": 2557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978331098796354e-05,
+      "loss": 0.3037,
+      "step": 2558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978306346476137e-05,
+      "loss": 0.2894,
+      "step": 2559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978281580181763e-05,
+      "loss": 0.2973,
+      "step": 2560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782567999135866e-05,
+      "loss": 0.2908,
+      "step": 2561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978232005671961e-05,
+      "loss": 0.3127,
+      "step": 2562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782071974572408e-05,
+      "loss": 0.275,
+      "step": 2563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781823752697797e-05,
+      "loss": 0.3365,
+      "step": 2564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978157539109933e-05,
+      "loss": 0.2956,
+      "step": 2565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781326889780553e-05,
+      "loss": 0.3241,
+      "step": 2566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781078248745014e-05,
+      "loss": 0.3165,
+      "step": 2567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780829467996262e-05,
+      "loss": 0.3117,
+      "step": 2568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780580547537856e-05,
+      "loss": 0.2802,
+      "step": 2569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978033148737335e-05,
+      "loss": 0.3321,
+      "step": 2570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780082287506297e-05,
+      "loss": 0.3041,
+      "step": 2571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779832947940264e-05,
+      "loss": 0.2951,
+      "step": 2572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779583468678806e-05,
+      "loss": 0.3239,
+      "step": 2573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977933384972549e-05,
+      "loss": 0.3394,
+      "step": 2574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779084091083883e-05,
+      "loss": 0.3136,
+      "step": 2575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977883419275755e-05,
+      "loss": 0.3435,
+      "step": 2576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977858415475006e-05,
+      "loss": 0.3068,
+      "step": 2577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977833397706499e-05,
+      "loss": 0.3024,
+      "step": 2578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778083659705905e-05,
+      "loss": 0.3098,
+      "step": 2579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777833202676383e-05,
+      "loss": 0.3266,
+      "step": 2580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777582605980007e-05,
+      "loss": 0.308,
+      "step": 2581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977733186962035e-05,
+      "loss": 0.3104,
+      "step": 2582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777080993601e-05,
+      "loss": 0.3413,
+      "step": 2583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776829977925534e-05,
+      "loss": 0.3091,
+      "step": 2584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977657882259754e-05,
+      "loss": 0.3267,
+      "step": 2585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977632752762061e-05,
+      "loss": 0.3056,
+      "step": 2586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776076092998325e-05,
+      "loss": 0.333,
+      "step": 2587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775824518734283e-05,
+      "loss": 0.3294,
+      "step": 2588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775572804832076e-05,
+      "loss": 0.3395,
+      "step": 2589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775320951295298e-05,
+      "loss": 0.3193,
+      "step": 2590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775068958127548e-05,
+      "loss": 0.3269,
+      "step": 2591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774816825332423e-05,
+      "loss": 0.3257,
+      "step": 2592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977456455291353e-05,
+      "loss": 0.3131,
+      "step": 2593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774312140874462e-05,
+      "loss": 0.3229,
+      "step": 2594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774059589218838e-05,
+      "loss": 0.3183,
+      "step": 2595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773806897950253e-05,
+      "loss": 0.3199,
+      "step": 2596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977355406707233e-05,
+      "loss": 0.3552,
+      "step": 2597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773301096588667e-05,
+      "loss": 0.3068,
+      "step": 2598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773047986502886e-05,
+      "loss": 0.3144,
+      "step": 2599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772794736818594e-05,
+      "loss": 0.3222,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772541347539417e-05,
+      "loss": 0.3089,
+      "step": 2601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772287818668973e-05,
+      "loss": 0.3006,
+      "step": 2602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977203415021088e-05,
+      "loss": 0.3065,
+      "step": 2603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771780342168766e-05,
+      "loss": 0.3204,
+      "step": 2604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977152639454625e-05,
+      "loss": 0.3202,
+      "step": 2605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771272307346964e-05,
+      "loss": 0.302,
+      "step": 2606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771018080574534e-05,
+      "loss": 0.3079,
+      "step": 2607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770763714232597e-05,
+      "loss": 0.3106,
+      "step": 2608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770509208324783e-05,
+      "loss": 0.3415,
+      "step": 2609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977025456285473e-05,
+      "loss": 0.2939,
+      "step": 2610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976999977782607e-05,
+      "loss": 0.3001,
+      "step": 2611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769744853242443e-05,
+      "loss": 0.3194,
+      "step": 2612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769489789107492e-05,
+      "loss": 0.348,
+      "step": 2613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769234585424863e-05,
+      "loss": 0.3215,
+      "step": 2614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97689792421982e-05,
+      "loss": 0.318,
+      "step": 2615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976872375943115e-05,
+      "loss": 0.3009,
+      "step": 2616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976846813712736e-05,
+      "loss": 0.3,
+      "step": 2617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768212375290484e-05,
+      "loss": 0.3309,
+      "step": 2618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767956473924174e-05,
+      "loss": 0.3209,
+      "step": 2619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767700433032086e-05,
+      "loss": 0.3228,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767444252617876e-05,
+      "loss": 0.3056,
+      "step": 2621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767187932685207e-05,
+      "loss": 0.3242,
+      "step": 2622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766931473237738e-05,
+      "loss": 0.322,
+      "step": 2623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976667487427913e-05,
+      "loss": 0.3292,
+      "step": 2624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976641813581305e-05,
+      "loss": 0.311,
+      "step": 2625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766161257843166e-05,
+      "loss": 0.3198,
+      "step": 2626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765904240373145e-05,
+      "loss": 0.2951,
+      "step": 2627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765647083406665e-05,
+      "loss": 0.3405,
+      "step": 2628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976538978694739e-05,
+      "loss": 0.316,
+      "step": 2629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765132350999e-05,
+      "loss": 0.3302,
+      "step": 2630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764874775565173e-05,
+      "loss": 0.3057,
+      "step": 2631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764617060649585e-05,
+      "loss": 0.3291,
+      "step": 2632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976435920625592e-05,
+      "loss": 0.3046,
+      "step": 2633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764101212387862e-05,
+      "loss": 0.2977,
+      "step": 2634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763843079049092e-05,
+      "loss": 0.2988,
+      "step": 2635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763584806243298e-05,
+      "loss": 0.2891,
+      "step": 2636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763326393974172e-05,
+      "loss": 0.3301,
+      "step": 2637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763067842245406e-05,
+      "loss": 0.3229,
+      "step": 2638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976280915106069e-05,
+      "loss": 0.3312,
+      "step": 2639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976255032042372e-05,
+      "loss": 0.3082,
+      "step": 2640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762291350338195e-05,
+      "loss": 0.3277,
+      "step": 2641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976203224080781e-05,
+      "loss": 0.3127,
+      "step": 2642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761772991836273e-05,
+      "loss": 0.323,
+      "step": 2643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976151360342728e-05,
+      "loss": 0.3124,
+      "step": 2644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976125407558454e-05,
+      "loss": 0.3156,
+      "step": 2645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760994408311757e-05,
+      "loss": 0.3314,
+      "step": 2646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760734601612644e-05,
+      "loss": 0.3151,
+      "step": 2647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760474655490907e-05,
+      "loss": 0.2981,
+      "step": 2648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976021456995027e-05,
+      "loss": 0.3024,
+      "step": 2649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759954344994434e-05,
+      "loss": 0.3291,
+      "step": 2650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759693980627123e-05,
+      "loss": 0.2846,
+      "step": 2651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759433476852058e-05,
+      "loss": 0.3068,
+      "step": 2652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759172833672956e-05,
+      "loss": 0.3118,
+      "step": 2653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975891205109354e-05,
+      "loss": 0.308,
+      "step": 2654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758651129117535e-05,
+      "loss": 0.3018,
+      "step": 2655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758390067748673e-05,
+      "loss": 0.3005,
+      "step": 2656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975812886699068e-05,
+      "loss": 0.3609,
+      "step": 2657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975786752684729e-05,
+      "loss": 0.2891,
+      "step": 2658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757606047322225e-05,
+      "loss": 0.3308,
+      "step": 2659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975734442841923e-05,
+      "loss": 0.3062,
+      "step": 2660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975708267014204e-05,
+      "loss": 0.3075,
+      "step": 2661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756820772494395e-05,
+      "loss": 0.2937,
+      "step": 2662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756558735480035e-05,
+      "loss": 0.326,
+      "step": 2663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97562965591027e-05,
+      "loss": 0.288,
+      "step": 2664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756034243366137e-05,
+      "loss": 0.283,
+      "step": 2665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755771788274094e-05,
+      "loss": 0.2733,
+      "step": 2666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755509193830322e-05,
+      "loss": 0.3028,
+      "step": 2667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755246460038565e-05,
+      "loss": 0.3114,
+      "step": 2668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754983586902583e-05,
+      "loss": 0.3185,
+      "step": 2669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754720574426127e-05,
+      "loss": 0.3042,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754457422612955e-05,
+      "loss": 0.3003,
+      "step": 2671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754194131466823e-05,
+      "loss": 0.3095,
+      "step": 2672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753930700991497e-05,
+      "loss": 0.3208,
+      "step": 2673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975366713119074e-05,
+      "loss": 0.2987,
+      "step": 2674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753403422068313e-05,
+      "loss": 0.2905,
+      "step": 2675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975313957362798e-05,
+      "loss": 0.3458,
+      "step": 2676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752875585873513e-05,
+      "loss": 0.3232,
+      "step": 2677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975261145880869e-05,
+      "loss": 0.2862,
+      "step": 2678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752347192437272e-05,
+      "loss": 0.3377,
+      "step": 2679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752082786763044e-05,
+      "loss": 0.2869,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751818241789772e-05,
+      "loss": 0.3451,
+      "step": 2681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751553557521247e-05,
+      "loss": 0.3126,
+      "step": 2682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975128873396124e-05,
+      "loss": 0.296,
+      "step": 2683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975102377111354e-05,
+      "loss": 0.2866,
+      "step": 2684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750758668981925e-05,
+      "loss": 0.3547,
+      "step": 2685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750493427570188e-05,
+      "loss": 0.3246,
+      "step": 2686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750228046882118e-05,
+      "loss": 0.323,
+      "step": 2687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97499625269215e-05,
+      "loss": 0.2956,
+      "step": 2688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974969686769213e-05,
+      "loss": 0.3347,
+      "step": 2689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749431069197803e-05,
+      "loss": 0.3065,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749165131442313e-05,
+      "loss": 0.2768,
+      "step": 2691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748899054429464e-05,
+      "loss": 0.3531,
+      "step": 2692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748632838163052e-05,
+      "loss": 0.3241,
+      "step": 2693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974836648264688e-05,
+      "loss": 0.3703,
+      "step": 2694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748099987884757e-05,
+      "loss": 0.2897,
+      "step": 2695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747833353880487e-05,
+      "loss": 0.2798,
+      "step": 2696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747566580637874e-05,
+      "loss": 0.3381,
+      "step": 2697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747299668160735e-05,
+      "loss": 0.3104,
+      "step": 2698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747032616452883e-05,
+      "loss": 0.3253,
+      "step": 2699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746765425518124e-05,
+      "loss": 0.3179,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746498095360283e-05,
+      "loss": 0.3011,
+      "step": 2701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746230625983175e-05,
+      "loss": 0.3059,
+      "step": 2702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745963017390622e-05,
+      "loss": 0.3042,
+      "step": 2703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745695269586446e-05,
+      "loss": 0.3275,
+      "step": 2704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745427382574475e-05,
+      "loss": 0.3336,
+      "step": 2705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745159356358528e-05,
+      "loss": 0.3338,
+      "step": 2706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744891190942436e-05,
+      "loss": 0.3204,
+      "step": 2707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744622886330035e-05,
+      "loss": 0.2929,
+      "step": 2708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974435444252515e-05,
+      "loss": 0.3007,
+      "step": 2709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974408585953162e-05,
+      "loss": 0.3087,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743817137353283e-05,
+      "loss": 0.3203,
+      "step": 2711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743548275993973e-05,
+      "loss": 0.3215,
+      "step": 2712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974327927545753e-05,
+      "loss": 0.3294,
+      "step": 2713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743010135747802e-05,
+      "loss": 0.3202,
+      "step": 2714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974274085686863e-05,
+      "loss": 0.2848,
+      "step": 2715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742471438823857e-05,
+      "loss": 0.2889,
+      "step": 2716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742201881617338e-05,
+      "loss": 0.3146,
+      "step": 2717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741932185252918e-05,
+      "loss": 0.2813,
+      "step": 2718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741662349734453e-05,
+      "loss": 0.3268,
+      "step": 2719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741392375065795e-05,
+      "loss": 0.3086,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741122261250802e-05,
+      "loss": 0.3116,
+      "step": 2721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974085200829333e-05,
+      "loss": 0.3397,
+      "step": 2722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974058161619724e-05,
+      "loss": 0.3161,
+      "step": 2723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740311084966398e-05,
+      "loss": 0.3283,
+      "step": 2724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740040414604667e-05,
+      "loss": 0.3425,
+      "step": 2725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739769605115908e-05,
+      "loss": 0.3226,
+      "step": 2726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739498656503996e-05,
+      "loss": 0.311,
+      "step": 2727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739227568772795e-05,
+      "loss": 0.328,
+      "step": 2728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973895634192618e-05,
+      "loss": 0.29,
+      "step": 2729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738684975968025e-05,
+      "loss": 0.305,
+      "step": 2730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973841347090221e-05,
+      "loss": 0.2907,
+      "step": 2731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973814182673261e-05,
+      "loss": 0.3599,
+      "step": 2732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737870043463104e-05,
+      "loss": 0.311,
+      "step": 2733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737598121097573e-05,
+      "loss": 0.2704,
+      "step": 2734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737326059639904e-05,
+      "loss": 0.3163,
+      "step": 2735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737053859093984e-05,
+      "loss": 0.3313,
+      "step": 2736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97367815194637e-05,
+      "loss": 0.3106,
+      "step": 2737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973650904075294e-05,
+      "loss": 0.3154,
+      "step": 2738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97362364229656e-05,
+      "loss": 0.324,
+      "step": 2739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735963666105572e-05,
+      "loss": 0.3075,
+      "step": 2740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973569077017675e-05,
+      "loss": 0.3424,
+      "step": 2741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735417735183033e-05,
+      "loss": 0.3287,
+      "step": 2742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735144561128327e-05,
+      "loss": 0.3223,
+      "step": 2743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734871248016523e-05,
+      "loss": 0.3042,
+      "step": 2744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734597795851535e-05,
+      "loss": 0.3051,
+      "step": 2745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734324204637265e-05,
+      "loss": 0.3169,
+      "step": 2746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973405047437762e-05,
+      "loss": 0.3263,
+      "step": 2747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973377660507651e-05,
+      "loss": 0.323,
+      "step": 2748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973350259673785e-05,
+      "loss": 0.3063,
+      "step": 2749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733228449365554e-05,
+      "loss": 0.2837,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732954162963535e-05,
+      "loss": 0.2961,
+      "step": 2751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732679737535708e-05,
+      "loss": 0.3311,
+      "step": 2752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732405173085997e-05,
+      "loss": 0.3336,
+      "step": 2753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732130469618328e-05,
+      "loss": 0.2951,
+      "step": 2754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973185562713662e-05,
+      "loss": 0.2952,
+      "step": 2755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731580645644794e-05,
+      "loss": 0.3317,
+      "step": 2756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973130552514679e-05,
+      "loss": 0.3089,
+      "step": 2757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731030265646527e-05,
+      "loss": 0.3052,
+      "step": 2758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730754867147942e-05,
+      "loss": 0.308,
+      "step": 2759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730479329654967e-05,
+      "loss": 0.2912,
+      "step": 2760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973020365317154e-05,
+      "loss": 0.3421,
+      "step": 2761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729927837701595e-05,
+      "loss": 0.3087,
+      "step": 2762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729651883249075e-05,
+      "loss": 0.301,
+      "step": 2763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972937578981792e-05,
+      "loss": 0.2941,
+      "step": 2764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729099557412075e-05,
+      "loss": 0.3048,
+      "step": 2765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728823186035486e-05,
+      "loss": 0.3029,
+      "step": 2766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97285466756921e-05,
+      "loss": 0.3025,
+      "step": 2767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728270026385866e-05,
+      "loss": 0.3458,
+      "step": 2768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727993238120734e-05,
+      "loss": 0.3033,
+      "step": 2769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972771631090066e-05,
+      "loss": 0.2997,
+      "step": 2770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97274392447296e-05,
+      "loss": 0.3031,
+      "step": 2771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727162039611515e-05,
+      "loss": 0.2697,
+      "step": 2772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726884695550356e-05,
+      "loss": 0.2847,
+      "step": 2773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972660721255009e-05,
+      "loss": 0.3362,
+      "step": 2774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726329590614682e-05,
+      "loss": 0.3167,
+      "step": 2775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726051829748095e-05,
+      "loss": 0.3023,
+      "step": 2776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725773929954298e-05,
+      "loss": 0.2619,
+      "step": 2777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725495891237258e-05,
+      "loss": 0.3,
+      "step": 2778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725217713600947e-05,
+      "loss": 0.3161,
+      "step": 2779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972493939704934e-05,
+      "loss": 0.3281,
+      "step": 2780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724660941586414e-05,
+      "loss": 0.3256,
+      "step": 2781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724382347216143e-05,
+      "loss": 0.325,
+      "step": 2782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724103613942505e-05,
+      "loss": 0.3054,
+      "step": 2783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723824741769488e-05,
+      "loss": 0.3072,
+      "step": 2784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972354573070107e-05,
+      "loss": 0.2901,
+      "step": 2785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972326658074124e-05,
+      "loss": 0.3195,
+      "step": 2786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722987291893983e-05,
+      "loss": 0.2917,
+      "step": 2787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722707864163287e-05,
+      "loss": 0.315,
+      "step": 2788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972242829755315e-05,
+      "loss": 0.2899,
+      "step": 2789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722148592067553e-05,
+      "loss": 0.2986,
+      "step": 2790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721868747710505e-05,
+      "loss": 0.2998,
+      "step": 2791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721588764485993e-05,
+      "loss": 0.2968,
+      "step": 2792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721308642398026e-05,
+      "loss": 0.309,
+      "step": 2793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721028381450598e-05,
+      "loss": 0.3472,
+      "step": 2794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972074798164771e-05,
+      "loss": 0.2944,
+      "step": 2795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720467442993375e-05,
+      "loss": 0.3279,
+      "step": 2796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720186765491596e-05,
+      "loss": 0.3294,
+      "step": 2797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719905949146382e-05,
+      "loss": 0.2958,
+      "step": 2798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719624993961744e-05,
+      "loss": 0.2712,
+      "step": 2799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719343899941702e-05,
+      "loss": 0.2944,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971906266709026e-05,
+      "loss": 0.3011,
+      "step": 2801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971878129541144e-05,
+      "loss": 0.3276,
+      "step": 2802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718499784909262e-05,
+      "loss": 0.3534,
+      "step": 2803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718218135587745e-05,
+      "loss": 0.3283,
+      "step": 2804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717936347450918e-05,
+      "loss": 0.282,
+      "step": 2805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717654420502798e-05,
+      "loss": 0.308,
+      "step": 2806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717372354747416e-05,
+      "loss": 0.297,
+      "step": 2807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717090150188805e-05,
+      "loss": 0.3313,
+      "step": 2808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716807806830988e-05,
+      "loss": 0.3573,
+      "step": 2809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716525324678003e-05,
+      "loss": 0.3525,
+      "step": 2810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716242703733886e-05,
+      "loss": 0.2923,
+      "step": 2811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971595994400267e-05,
+      "loss": 0.3015,
+      "step": 2812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715677045488392e-05,
+      "loss": 0.3146,
+      "step": 2813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715394008195103e-05,
+      "loss": 0.3297,
+      "step": 2814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715110832126835e-05,
+      "loss": 0.3444,
+      "step": 2815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971482751728764e-05,
+      "loss": 0.3124,
+      "step": 2816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971454406368156e-05,
+      "loss": 0.3436,
+      "step": 2817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714260471312648e-05,
+      "loss": 0.3023,
+      "step": 2818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713976740184954e-05,
+      "loss": 0.3194,
+      "step": 2819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713692870302527e-05,
+      "loss": 0.2949,
+      "step": 2820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713408861669428e-05,
+      "loss": 0.3192,
+      "step": 2821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971312471428971e-05,
+      "loss": 0.2943,
+      "step": 2822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971284042816743e-05,
+      "loss": 0.3052,
+      "step": 2823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712556003306652e-05,
+      "loss": 0.2878,
+      "step": 2824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971227143971144e-05,
+      "loss": 0.3077,
+      "step": 2825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711986737385852e-05,
+      "loss": 0.331,
+      "step": 2826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971170189633396e-05,
+      "loss": 0.356,
+      "step": 2827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711416916559837e-05,
+      "loss": 0.3075,
+      "step": 2828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711131798067545e-05,
+      "loss": 0.3448,
+      "step": 2829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971084654086116e-05,
+      "loss": 0.3265,
+      "step": 2830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710561144944756e-05,
+      "loss": 0.3047,
+      "step": 2831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710275610322412e-05,
+      "loss": 0.2892,
+      "step": 2832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709989936998203e-05,
+      "loss": 0.3366,
+      "step": 2833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970970412497621e-05,
+      "loss": 0.3284,
+      "step": 2834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.3495,
+      "step": 2835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709132084855217e-05,
+      "loss": 0.3166,
+      "step": 2836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970884585676438e-05,
+      "loss": 0.2813,
+      "step": 2837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708559489992108e-05,
+      "loss": 0.3206,
+      "step": 2838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970827298454248e-05,
+      "loss": 0.3119,
+      "step": 2839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707986340419597e-05,
+      "loss": 0.3287,
+      "step": 2840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707699557627554e-05,
+      "loss": 0.3294,
+      "step": 2841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970741263617044e-05,
+      "loss": 0.295,
+      "step": 2842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970712557605236e-05,
+      "loss": 0.3241,
+      "step": 2843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706838377277412e-05,
+      "loss": 0.3138,
+      "step": 2844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97065510398497e-05,
+      "loss": 0.3255,
+      "step": 2845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706263563773326e-05,
+      "loss": 0.3413,
+      "step": 2846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705975949052396e-05,
+      "loss": 0.3293,
+      "step": 2847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970568819569102e-05,
+      "loss": 0.2891,
+      "step": 2848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705400303693305e-05,
+      "loss": 0.3126,
+      "step": 2849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705112273063373e-05,
+      "loss": 0.325,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704824103805325e-05,
+      "loss": 0.3254,
+      "step": 2851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704535795923284e-05,
+      "loss": 0.3064,
+      "step": 2852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704247349421367e-05,
+      "loss": 0.2917,
+      "step": 2853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703958764303698e-05,
+      "loss": 0.3065,
+      "step": 2854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703670040574393e-05,
+      "loss": 0.3288,
+      "step": 2855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703381178237582e-05,
+      "loss": 0.328,
+      "step": 2856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970309217729739e-05,
+      "loss": 0.2902,
+      "step": 2857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970280303775794e-05,
+      "loss": 0.3439,
+      "step": 2858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702513759623364e-05,
+      "loss": 0.3022,
+      "step": 2859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702224342897797e-05,
+      "loss": 0.3061,
+      "step": 2860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970193478758537e-05,
+      "loss": 0.3085,
+      "step": 2861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701645093690224e-05,
+      "loss": 0.3234,
+      "step": 2862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970135526121649e-05,
+      "loss": 0.2994,
+      "step": 2863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970106529016831e-05,
+      "loss": 0.2877,
+      "step": 2864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700775180549834e-05,
+      "loss": 0.2926,
+      "step": 2865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700484932365196e-05,
+      "loss": 0.3227,
+      "step": 2866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970019454561854e-05,
+      "loss": 0.2973,
+      "step": 2867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699904020314027e-05,
+      "loss": 0.3239,
+      "step": 2868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699613356455797e-05,
+      "loss": 0.3413,
+      "step": 2869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699322554048005e-05,
+      "loss": 0.3478,
+      "step": 2870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96990316130948e-05,
+      "loss": 0.3069,
+      "step": 2871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698740533600343e-05,
+      "loss": 0.3357,
+      "step": 2872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969844931556879e-05,
+      "loss": 0.322,
+      "step": 2873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698157959004302e-05,
+      "loss": 0.2986,
+      "step": 2874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969786646391104e-05,
+      "loss": 0.3298,
+      "step": 2875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969757483029317e-05,
+      "loss": 0.2905,
+      "step": 2876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969728305815485e-05,
+      "loss": 0.3518,
+      "step": 2877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696991147500254e-05,
+      "loss": 0.3567,
+      "step": 2878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696699098333557e-05,
+      "loss": 0.3132,
+      "step": 2879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696406910658918e-05,
+      "loss": 0.343,
+      "step": 2880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696114584480517e-05,
+      "loss": 0.3291,
+      "step": 2881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969582211980253e-05,
+      "loss": 0.338,
+      "step": 2882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695529516629136e-05,
+      "loss": 0.3079,
+      "step": 2883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695236774964514e-05,
+      "loss": 0.2731,
+      "step": 2884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969494389481284e-05,
+      "loss": 0.3332,
+      "step": 2885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96946508761783e-05,
+      "loss": 0.3356,
+      "step": 2886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694357719065085e-05,
+      "loss": 0.3242,
+      "step": 2887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694064423477378e-05,
+      "loss": 0.2916,
+      "step": 2888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969377098941937e-05,
+      "loss": 0.314,
+      "step": 2889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693477416895252e-05,
+      "loss": 0.3078,
+      "step": 2890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693183705909215e-05,
+      "loss": 0.2904,
+      "step": 2891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692889856465454e-05,
+      "loss": 0.3417,
+      "step": 2892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692595868568172e-05,
+      "loss": 0.3125,
+      "step": 2893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692301742221562e-05,
+      "loss": 0.3169,
+      "step": 2894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692007477429832e-05,
+      "loss": 0.3148,
+      "step": 2895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691713074197178e-05,
+      "loss": 0.3077,
+      "step": 2896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969141853252781e-05,
+      "loss": 0.3233,
+      "step": 2897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691123852425936e-05,
+      "loss": 0.3369,
+      "step": 2898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969082903389576e-05,
+      "loss": 0.3218,
+      "step": 2899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96905340769415e-05,
+      "loss": 0.2955,
+      "step": 2900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690238981567364e-05,
+      "loss": 0.3133,
+      "step": 2901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689943747777566e-05,
+      "loss": 0.307,
+      "step": 2902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968964837557633e-05,
+      "loss": 0.3079,
+      "step": 2903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689352864967867e-05,
+      "loss": 0.2951,
+      "step": 2904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689057215956403e-05,
+      "loss": 0.3351,
+      "step": 2905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688761428546162e-05,
+      "loss": 0.3175,
+      "step": 2906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688465502741366e-05,
+      "loss": 0.3157,
+      "step": 2907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688169438546244e-05,
+      "loss": 0.3035,
+      "step": 2908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687873235965023e-05,
+      "loss": 0.2989,
+      "step": 2909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687576895001936e-05,
+      "loss": 0.2777,
+      "step": 2910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687280415661218e-05,
+      "loss": 0.3211,
+      "step": 2911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686983797947096e-05,
+      "loss": 0.3511,
+      "step": 2912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686687041863814e-05,
+      "loss": 0.2843,
+      "step": 2913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968639014741561e-05,
+      "loss": 0.359,
+      "step": 2914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968609311460672e-05,
+      "loss": 0.2717,
+      "step": 2915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685795943441394e-05,
+      "loss": 0.3037,
+      "step": 2916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685498633923873e-05,
+      "loss": 0.3278,
+      "step": 2917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685201186058402e-05,
+      "loss": 0.3008,
+      "step": 2918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968490359984923e-05,
+      "loss": 0.2907,
+      "step": 2919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684605875300617e-05,
+      "loss": 0.3107,
+      "step": 2920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96843080124168e-05,
+      "loss": 0.3132,
+      "step": 2921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684010011202045e-05,
+      "loss": 0.3307,
+      "step": 2922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683711871660606e-05,
+      "loss": 0.3285,
+      "step": 2923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683413593796745e-05,
+      "loss": 0.3552,
+      "step": 2924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968311517761471e-05,
+      "loss": 0.3384,
+      "step": 2925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968281662311878e-05,
+      "loss": 0.3262,
+      "step": 2926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682517930313206e-05,
+      "loss": 0.2791,
+      "step": 2927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682219099202267e-05,
+      "loss": 0.3051,
+      "step": 2928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968192012979022e-05,
+      "loss": 0.318,
+      "step": 2929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968162102208134e-05,
+      "loss": 0.2947,
+      "step": 2930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681321776079906e-05,
+      "loss": 0.2951,
+      "step": 2931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968102239179018e-05,
+      "loss": 0.3079,
+      "step": 2932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680722869216445e-05,
+      "loss": 0.3046,
+      "step": 2933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680423208362985e-05,
+      "loss": 0.3165,
+      "step": 2934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968012340923407e-05,
+      "loss": 0.3215,
+      "step": 2935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679823471833988e-05,
+      "loss": 0.3023,
+      "step": 2936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679523396167023e-05,
+      "loss": 0.3186,
+      "step": 2937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967922318223746e-05,
+      "loss": 0.3435,
+      "step": 2938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678922830049583e-05,
+      "loss": 0.321,
+      "step": 2939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678622339607694e-05,
+      "loss": 0.3246,
+      "step": 2940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678321710916072e-05,
+      "loss": 0.3263,
+      "step": 2941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678020943979024e-05,
+      "loss": 0.2965,
+      "step": 2942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677720038800835e-05,
+      "loss": 0.3051,
+      "step": 2943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967741899538581e-05,
+      "loss": 0.2924,
+      "step": 2944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677117813738245e-05,
+      "loss": 0.3309,
+      "step": 2945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676816493862446e-05,
+      "loss": 0.3285,
+      "step": 2946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967651503576271e-05,
+      "loss": 0.2977,
+      "step": 2947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676213439443355e-05,
+      "loss": 0.3634,
+      "step": 2948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967591170490868e-05,
+      "loss": 0.3123,
+      "step": 2949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675609832162996e-05,
+      "loss": 0.296,
+      "step": 2950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675307821210613e-05,
+      "loss": 0.2829,
+      "step": 2951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967500567205585e-05,
+      "loss": 0.2939,
+      "step": 2952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674703384703023e-05,
+      "loss": 0.3143,
+      "step": 2953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674400959156443e-05,
+      "loss": 0.2972,
+      "step": 2954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967409839542044e-05,
+      "loss": 0.3073,
+      "step": 2955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673795693499326e-05,
+      "loss": 0.2904,
+      "step": 2956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967349285339743e-05,
+      "loss": 0.2918,
+      "step": 2957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673189875119082e-05,
+      "loss": 0.3207,
+      "step": 2958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672886758668598e-05,
+      "loss": 0.2681,
+      "step": 2959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967258350405032e-05,
+      "loss": 0.296,
+      "step": 2960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967228011126857e-05,
+      "loss": 0.3127,
+      "step": 2961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671976580327685e-05,
+      "loss": 0.3259,
+      "step": 2962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671672911232004e-05,
+      "loss": 0.3164,
+      "step": 2963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967136910398586e-05,
+      "loss": 0.2792,
+      "step": 2964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671065158593597e-05,
+      "loss": 0.3175,
+      "step": 2965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670761075059555e-05,
+      "loss": 0.3257,
+      "step": 2966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670456853388073e-05,
+      "loss": 0.3118,
+      "step": 2967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670152493583502e-05,
+      "loss": 0.2982,
+      "step": 2968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669847995650188e-05,
+      "loss": 0.2937,
+      "step": 2969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966954335959248e-05,
+      "loss": 0.3178,
+      "step": 2970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669238585414728e-05,
+      "loss": 0.288,
+      "step": 2971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668933673121292e-05,
+      "loss": 0.2929,
+      "step": 2972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966862862271652e-05,
+      "loss": 0.3469,
+      "step": 2973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668323434204768e-05,
+      "loss": 0.3244,
+      "step": 2974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668018107590404e-05,
+      "loss": 0.3154,
+      "step": 2975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966771264287778e-05,
+      "loss": 0.3174,
+      "step": 2976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966740704007127e-05,
+      "loss": 0.3231,
+      "step": 2977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667101299175225e-05,
+      "loss": 0.349,
+      "step": 2978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666795420194023e-05,
+      "loss": 0.3301,
+      "step": 2979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966648940313203e-05,
+      "loss": 0.2805,
+      "step": 2980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666183247993624e-05,
+      "loss": 0.2824,
+      "step": 2981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665876954783167e-05,
+      "loss": 0.2707,
+      "step": 2982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665570523505035e-05,
+      "loss": 0.3061,
+      "step": 2983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665263954163613e-05,
+      "loss": 0.2952,
+      "step": 2984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966495724676327e-05,
+      "loss": 0.3093,
+      "step": 2985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96646504013084e-05,
+      "loss": 0.304,
+      "step": 2986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664343417803377e-05,
+      "loss": 0.2893,
+      "step": 2987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664036296252584e-05,
+      "loss": 0.3229,
+      "step": 2988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663729036660415e-05,
+      "loss": 0.3113,
+      "step": 2989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663421639031254e-05,
+      "loss": 0.3034,
+      "step": 2990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663114103369495e-05,
+      "loss": 0.3117,
+      "step": 2991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966280642967953e-05,
+      "loss": 0.3066,
+      "step": 2992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966249861796575e-05,
+      "loss": 0.295,
+      "step": 2993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662190668232558e-05,
+      "loss": 0.3044,
+      "step": 2994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661882580484347e-05,
+      "loss": 0.27,
+      "step": 2995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661574354725523e-05,
+      "loss": 0.3076,
+      "step": 2996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661265990960486e-05,
+      "loss": 0.3369,
+      "step": 2997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966095748919364e-05,
+      "loss": 0.2916,
+      "step": 2998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966064884942939e-05,
+      "loss": 0.3066,
+      "step": 2999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660340071672152e-05,
+      "loss": 0.2823,
+      "step": 3000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660031155926327e-05,
+      "loss": 0.3047,
+      "step": 3001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659722102196335e-05,
+      "loss": 0.3087,
+      "step": 3002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659412910486587e-05,
+      "loss": 0.3159,
+      "step": 3003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96591035808015e-05,
+      "loss": 0.3585,
+      "step": 3004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658794113145495e-05,
+      "loss": 0.2974,
+      "step": 3005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658484507522987e-05,
+      "loss": 0.3048,
+      "step": 3006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96581747639384e-05,
+      "loss": 0.3287,
+      "step": 3007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657864882396165e-05,
+      "loss": 0.3499,
+      "step": 3008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657554862900697e-05,
+      "loss": 0.3131,
+      "step": 3009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657244705456438e-05,
+      "loss": 0.3519,
+      "step": 3010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656934410067807e-05,
+      "loss": 0.3209,
+      "step": 3011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656623976739242e-05,
+      "loss": 0.3258,
+      "step": 3012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656313405475172e-05,
+      "loss": 0.3059,
+      "step": 3013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656002696280038e-05,
+      "loss": 0.313,
+      "step": 3014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965569184915828e-05,
+      "loss": 0.3238,
+      "step": 3015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655380864114332e-05,
+      "loss": 0.2873,
+      "step": 3016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655069741152642e-05,
+      "loss": 0.3596,
+      "step": 3017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654758480277648e-05,
+      "loss": 0.3418,
+      "step": 3018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96544470814938e-05,
+      "loss": 0.3029,
+      "step": 3019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654135544805545e-05,
+      "loss": 0.2999,
+      "step": 3020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653823870217334e-05,
+      "loss": 0.3451,
+      "step": 3021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965351205773362e-05,
+      "loss": 0.3051,
+      "step": 3022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653200107358855e-05,
+      "loss": 0.3336,
+      "step": 3023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652888019097493e-05,
+      "loss": 0.3095,
+      "step": 3024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652575792953994e-05,
+      "loss": 0.2976,
+      "step": 3025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965226342893282e-05,
+      "loss": 0.2833,
+      "step": 3026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651950927038425e-05,
+      "loss": 0.3142,
+      "step": 3027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651638287275284e-05,
+      "loss": 0.289,
+      "step": 3028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651325509647857e-05,
+      "loss": 0.3036,
+      "step": 3029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965101259416061e-05,
+      "loss": 0.2815,
+      "step": 3030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650699540818013e-05,
+      "loss": 0.321,
+      "step": 3031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650386349624544e-05,
+      "loss": 0.3031,
+      "step": 3032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965007302058467e-05,
+      "loss": 0.3231,
+      "step": 3033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649759553702867e-05,
+      "loss": 0.2939,
+      "step": 3034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649445948983617e-05,
+      "loss": 0.2682,
+      "step": 3035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649132206431395e-05,
+      "loss": 0.3108,
+      "step": 3036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648818326050683e-05,
+      "loss": 0.305,
+      "step": 3037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648504307845967e-05,
+      "loss": 0.3019,
+      "step": 3038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964819015182173e-05,
+      "loss": 0.3145,
+      "step": 3039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647875857982458e-05,
+      "loss": 0.3195,
+      "step": 3040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964756142633265e-05,
+      "loss": 0.3387,
+      "step": 3041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647246856876784e-05,
+      "loss": 0.3159,
+      "step": 3042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646932149619362e-05,
+      "loss": 0.3055,
+      "step": 3043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646617304564878e-05,
+      "loss": 0.3086,
+      "step": 3044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646302321717825e-05,
+      "loss": 0.3029,
+      "step": 3045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645987201082707e-05,
+      "loss": 0.3078,
+      "step": 3046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645671942664025e-05,
+      "loss": 0.3056,
+      "step": 3047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964535654646628e-05,
+      "loss": 0.3292,
+      "step": 3048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645041012493974e-05,
+      "loss": 0.3091,
+      "step": 3049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964472534075162e-05,
+      "loss": 0.2847,
+      "step": 3050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644409531243726e-05,
+      "loss": 0.3242,
+      "step": 3051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644093583974802e-05,
+      "loss": 0.3035,
+      "step": 3052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964377749894936e-05,
+      "loss": 0.325,
+      "step": 3053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643461276171917e-05,
+      "loss": 0.2915,
+      "step": 3054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964314491564699e-05,
+      "loss": 0.3141,
+      "step": 3055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964282841737909e-05,
+      "loss": 0.313,
+      "step": 3056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642511781372753e-05,
+      "loss": 0.3201,
+      "step": 3057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964219500763249e-05,
+      "loss": 0.288,
+      "step": 3058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964187809616283e-05,
+      "loss": 0.287,
+      "step": 3059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641561046968297e-05,
+      "loss": 0.3023,
+      "step": 3060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641243860053424e-05,
+      "loss": 0.287,
+      "step": 3061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640926535422738e-05,
+      "loss": 0.3459,
+      "step": 3062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640609073080777e-05,
+      "loss": 0.3123,
+      "step": 3063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640291473032066e-05,
+      "loss": 0.3422,
+      "step": 3064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963997373528115e-05,
+      "loss": 0.3212,
+      "step": 3065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963965585983257e-05,
+      "loss": 0.3299,
+      "step": 3066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639337846690852e-05,
+      "loss": 0.338,
+      "step": 3067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639019695860556e-05,
+      "loss": 0.2971,
+      "step": 3068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638701407346213e-05,
+      "loss": 0.2954,
+      "step": 3069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638382981152377e-05,
+      "loss": 0.2937,
+      "step": 3070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638064417283594e-05,
+      "loss": 0.2941,
+      "step": 3071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637745715744416e-05,
+      "loss": 0.3177,
+      "step": 3072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637426876539395e-05,
+      "loss": 0.2955,
+      "step": 3073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963710789967308e-05,
+      "loss": 0.3385,
+      "step": 3074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636788785150037e-05,
+      "loss": 0.3,
+      "step": 3075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636469532974817e-05,
+      "loss": 0.2954,
+      "step": 3076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636150143151983e-05,
+      "loss": 0.3181,
+      "step": 3077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635830615686092e-05,
+      "loss": 0.3313,
+      "step": 3078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635510950581718e-05,
+      "loss": 0.297,
+      "step": 3079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963519114784342e-05,
+      "loss": 0.3115,
+      "step": 3080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634871207475765e-05,
+      "loss": 0.2559,
+      "step": 3081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634551129483327e-05,
+      "loss": 0.3185,
+      "step": 3082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963423091387068e-05,
+      "loss": 0.2746,
+      "step": 3083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963391056064239e-05,
+      "loss": 0.2958,
+      "step": 3084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963359006980304e-05,
+      "loss": 0.3111,
+      "step": 3085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633269441357202e-05,
+      "loss": 0.2978,
+      "step": 3086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632948675309463e-05,
+      "loss": 0.2889,
+      "step": 3087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632627771664402e-05,
+      "loss": 0.2913,
+      "step": 3088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96323067304266e-05,
+      "loss": 0.3209,
+      "step": 3089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631985551600645e-05,
+      "loss": 0.3206,
+      "step": 3090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631664235191124e-05,
+      "loss": 0.295,
+      "step": 3091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963134278120263e-05,
+      "loss": 0.2903,
+      "step": 3092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631021189639746e-05,
+      "loss": 0.3157,
+      "step": 3093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630699460507076e-05,
+      "loss": 0.3307,
+      "step": 3094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963037759380921e-05,
+      "loss": 0.3209,
+      "step": 3095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630055589550744e-05,
+      "loss": 0.3349,
+      "step": 3096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629733447736283e-05,
+      "loss": 0.3003,
+      "step": 3097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629411168370426e-05,
+      "loss": 0.326,
+      "step": 3098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629088751457773e-05,
+      "loss": 0.3061,
+      "step": 3099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628766197002934e-05,
+      "loss": 0.276,
+      "step": 3100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962844350501052e-05,
+      "loss": 0.31,
+      "step": 3101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628120675485127e-05,
+      "loss": 0.3107,
+      "step": 3102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962779770843138e-05,
+      "loss": 0.308,
+      "step": 3103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627474603853885e-05,
+      "loss": 0.3019,
+      "step": 3104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962715136175726e-05,
+      "loss": 0.2748,
+      "step": 3105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626827982146117e-05,
+      "loss": 0.3047,
+      "step": 3106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626504465025083e-05,
+      "loss": 0.3108,
+      "step": 3107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626180810398776e-05,
+      "loss": 0.3155,
+      "step": 3108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625857018271816e-05,
+      "loss": 0.3055,
+      "step": 3109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625533088648832e-05,
+      "loss": 0.2718,
+      "step": 3110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625209021534452e-05,
+      "loss": 0.337,
+      "step": 3111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624884816933298e-05,
+      "loss": 0.3219,
+      "step": 3112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624560474850007e-05,
+      "loss": 0.3189,
+      "step": 3113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624235995289212e-05,
+      "loss": 0.2976,
+      "step": 3114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9623911378255546e-05,
+      "loss": 0.3064,
+      "step": 3115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9623586623753644e-05,
+      "loss": 0.3127,
+      "step": 3116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623261731788147e-05,
+      "loss": 0.3097,
+      "step": 3117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96229367023637e-05,
+      "loss": 0.3066,
+      "step": 3118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962261153548494e-05,
+      "loss": 0.3233,
+      "step": 3119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962228623115651e-05,
+      "loss": 0.3189,
+      "step": 3120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962196078938306e-05,
+      "loss": 0.2843,
+      "step": 3121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621635210169245e-05,
+      "loss": 0.3243,
+      "step": 3122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621309493519704e-05,
+      "loss": 0.2935,
+      "step": 3123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962098363943909e-05,
+      "loss": 0.3224,
+      "step": 3124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962065764793207e-05,
+      "loss": 0.2867,
+      "step": 3125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620331519003288e-05,
+      "loss": 0.3045,
+      "step": 3126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962000525265741e-05,
+      "loss": 0.3087,
+      "step": 3127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961967884889909e-05,
+      "loss": 0.315,
+      "step": 3128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619352307732998e-05,
+      "loss": 0.2932,
+      "step": 3129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619025629163793e-05,
+      "loss": 0.2755,
+      "step": 3130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961869881319614e-05,
+      "loss": 0.3004,
+      "step": 3131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961837185983471e-05,
+      "loss": 0.2915,
+      "step": 3132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618044769084172e-05,
+      "loss": 0.3047,
+      "step": 3133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617717540949203e-05,
+      "loss": 0.3096,
+      "step": 3134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961739017543447e-05,
+      "loss": 0.2971,
+      "step": 3135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617062672544653e-05,
+      "loss": 0.3004,
+      "step": 3136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616735032284427e-05,
+      "loss": 0.3112,
+      "step": 3137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961640725465848e-05,
+      "loss": 0.3109,
+      "step": 3138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616079339671487e-05,
+      "loss": 0.3093,
+      "step": 3139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961575128732813e-05,
+      "loss": 0.3184,
+      "step": 3140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96154230976331e-05,
+      "loss": 0.3286,
+      "step": 3141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961509477059108e-05,
+      "loss": 0.3111,
+      "step": 3142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614766306206765e-05,
+      "loss": 0.3254,
+      "step": 3143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961443770448485e-05,
+      "loss": 0.3165,
+      "step": 3144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614108965430018e-05,
+      "loss": 0.3279,
+      "step": 3145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613780089046968e-05,
+      "loss": 0.3187,
+      "step": 3146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613451075340403e-05,
+      "loss": 0.3018,
+      "step": 3147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613121924315018e-05,
+      "loss": 0.2825,
+      "step": 3148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612792635975516e-05,
+      "loss": 0.3135,
+      "step": 3149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612463210326603e-05,
+      "loss": 0.3447,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612133647372982e-05,
+      "loss": 0.3063,
+      "step": 3151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961180394711936e-05,
+      "loss": 0.2942,
+      "step": 3152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611474109570446e-05,
+      "loss": 0.3562,
+      "step": 3153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961114413473095e-05,
+      "loss": 0.3315,
+      "step": 3154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610814022605592e-05,
+      "loss": 0.3331,
+      "step": 3155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610483773199082e-05,
+      "loss": 0.3209,
+      "step": 3156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961015338651614e-05,
+      "loss": 0.2801,
+      "step": 3157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960982286256148e-05,
+      "loss": 0.2708,
+      "step": 3158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960949220133983e-05,
+      "loss": 0.3045,
+      "step": 3159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960916140285591e-05,
+      "loss": 0.2817,
+      "step": 3160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608830467114445e-05,
+      "loss": 0.2848,
+      "step": 3161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608499394120162e-05,
+      "loss": 0.3187,
+      "step": 3162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960816818387779e-05,
+      "loss": 0.2829,
+      "step": 3163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960783683639206e-05,
+      "loss": 0.3122,
+      "step": 3164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607505351667705e-05,
+      "loss": 0.3309,
+      "step": 3165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607173729709462e-05,
+      "loss": 0.3145,
+      "step": 3166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960684197052207e-05,
+      "loss": 0.3574,
+      "step": 3167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606510074110257e-05,
+      "loss": 0.2996,
+      "step": 3168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606178040478775e-05,
+      "loss": 0.3479,
+      "step": 3169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605845869632363e-05,
+      "loss": 0.3227,
+      "step": 3170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605513561575762e-05,
+      "loss": 0.3155,
+      "step": 3171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605181116313725e-05,
+      "loss": 0.3236,
+      "step": 3172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604848533850996e-05,
+      "loss": 0.2912,
+      "step": 3173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604515814192327e-05,
+      "loss": 0.2813,
+      "step": 3174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604182957342473e-05,
+      "loss": 0.3271,
+      "step": 3175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603849963306187e-05,
+      "loss": 0.2875,
+      "step": 3176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603516832088227e-05,
+      "loss": 0.292,
+      "step": 3177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603183563693344e-05,
+      "loss": 0.3255,
+      "step": 3178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602850158126308e-05,
+      "loss": 0.3214,
+      "step": 3179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602516615391877e-05,
+      "loss": 0.323,
+      "step": 3180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602182935494813e-05,
+      "loss": 0.3127,
+      "step": 3181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601849118439885e-05,
+      "loss": 0.3106,
+      "step": 3182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601515164231865e-05,
+      "loss": 0.311,
+      "step": 3183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601181072875516e-05,
+      "loss": 0.2996,
+      "step": 3184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960084684437561e-05,
+      "loss": 0.3104,
+      "step": 3185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600512478736933e-05,
+      "loss": 0.3436,
+      "step": 3186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600177975964246e-05,
+      "loss": 0.3169,
+      "step": 3187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599843336062337e-05,
+      "loss": 0.3127,
+      "step": 3188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599508559035985e-05,
+      "loss": 0.2854,
+      "step": 3189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599173644889964e-05,
+      "loss": 0.3216,
+      "step": 3190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959883859362907e-05,
+      "loss": 0.3048,
+      "step": 3191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959850340525808e-05,
+      "loss": 0.2947,
+      "step": 3192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598168079781783e-05,
+      "loss": 0.3107,
+      "step": 3193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597832617204968e-05,
+      "loss": 0.3111,
+      "step": 3194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597497017532432e-05,
+      "loss": 0.2954,
+      "step": 3195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597161280768965e-05,
+      "loss": 0.3388,
+      "step": 3196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596825406919368e-05,
+      "loss": 0.2981,
+      "step": 3197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959648939598843e-05,
+      "loss": 0.3132,
+      "step": 3198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596153247980952e-05,
+      "loss": 0.2938,
+      "step": 3199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595816962901742e-05,
+      "loss": 0.3167,
+      "step": 3200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.95954805407556e-05,
+      "loss": 0.2956,
+      "step": 3201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595143981547328e-05,
+      "loss": 0.3044,
+      "step": 3202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594807285281737e-05,
+      "loss": 0.2998,
+      "step": 3203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959447045196364e-05,
+      "loss": 0.2934,
+      "step": 3204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594133481597837e-05,
+      "loss": 0.2996,
+      "step": 3205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9593796374189155e-05,
+      "loss": 0.2932,
+      "step": 3206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593459129742404e-05,
+      "loss": 0.3213,
+      "step": 3207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593121748262395e-05,
+      "loss": 0.3054,
+      "step": 3208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592784229753953e-05,
+      "loss": 0.3191,
+      "step": 3209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95924465742219e-05,
+      "loss": 0.301,
+      "step": 3210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592108781671054e-05,
+      "loss": 0.3072,
+      "step": 3211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591770852106247e-05,
+      "loss": 0.324,
+      "step": 3212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591432785532302e-05,
+      "loss": 0.286,
+      "step": 3213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591094581954047e-05,
+      "loss": 0.2722,
+      "step": 3214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590756241376314e-05,
+      "loss": 0.3196,
+      "step": 3215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590417763803938e-05,
+      "loss": 0.3053,
+      "step": 3216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959007914924175e-05,
+      "loss": 0.3436,
+      "step": 3217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958974039769459e-05,
+      "loss": 0.3326,
+      "step": 3218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589401509167295e-05,
+      "loss": 0.3285,
+      "step": 3219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589062483664708e-05,
+      "loss": 0.342,
+      "step": 3220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958872332119167e-05,
+      "loss": 0.294,
+      "step": 3221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588384021753022e-05,
+      "loss": 0.2928,
+      "step": 3222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588044585353617e-05,
+      "loss": 0.2606,
+      "step": 3223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587705011998295e-05,
+      "loss": 0.3094,
+      "step": 3224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958736530169192e-05,
+      "loss": 0.2859,
+      "step": 3225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958702545443933e-05,
+      "loss": 0.2877,
+      "step": 3226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958668547024539e-05,
+      "loss": 0.2717,
+      "step": 3227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586345349114952e-05,
+      "loss": 0.3064,
+      "step": 3228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958600509105287e-05,
+      "loss": 0.3126,
+      "step": 3229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958566469606401e-05,
+      "loss": 0.3051,
+      "step": 3230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585324164153236e-05,
+      "loss": 0.319,
+      "step": 3231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958498349532541e-05,
+      "loss": 0.276,
+      "step": 3232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584642689585393e-05,
+      "loss": 0.3185,
+      "step": 3233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584301746938062e-05,
+      "loss": 0.2924,
+      "step": 3234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958396066738828e-05,
+      "loss": 0.3024,
+      "step": 3235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958361945094092e-05,
+      "loss": 0.2688,
+      "step": 3236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583278097600858e-05,
+      "loss": 0.2683,
+      "step": 3237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582936607372968e-05,
+      "loss": 0.3011,
+      "step": 3238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582594980262134e-05,
+      "loss": 0.3071,
+      "step": 3239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582253216273223e-05,
+      "loss": 0.3001,
+      "step": 3240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581911315411132e-05,
+      "loss": 0.3077,
+      "step": 3241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581569277680735e-05,
+      "loss": 0.3061,
+      "step": 3242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581227103086917e-05,
+      "loss": 0.3094,
+      "step": 3243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958088479163457e-05,
+      "loss": 0.2902,
+      "step": 3244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580542343328584e-05,
+      "loss": 0.3184,
+      "step": 3245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580199758173848e-05,
+      "loss": 0.3064,
+      "step": 3246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579857036175257e-05,
+      "loss": 0.3188,
+      "step": 3247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579514177337704e-05,
+      "loss": 0.2954,
+      "step": 3248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579171181666088e-05,
+      "loss": 0.3176,
+      "step": 3249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578828049165312e-05,
+      "loss": 0.3209,
+      "step": 3250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957848477984027e-05,
+      "loss": 0.3086,
+      "step": 3251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957814137369587e-05,
+      "loss": 0.295,
+      "step": 3252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577797830737015e-05,
+      "loss": 0.324,
+      "step": 3253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577454150968614e-05,
+      "loss": 0.3096,
+      "step": 3254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577110334395577e-05,
+      "loss": 0.3226,
+      "step": 3255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576766381022815e-05,
+      "loss": 0.2906,
+      "step": 3256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957642229085524e-05,
+      "loss": 0.3148,
+      "step": 3257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957607806389776e-05,
+      "loss": 0.2981,
+      "step": 3258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575733700155303e-05,
+      "loss": 0.2801,
+      "step": 3259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575389199632786e-05,
+      "loss": 0.3283,
+      "step": 3260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575044562335126e-05,
+      "loss": 0.3187,
+      "step": 3261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574699788267243e-05,
+      "loss": 0.2944,
+      "step": 3262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574354877434072e-05,
+      "loss": 0.3433,
+      "step": 3263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574009829840534e-05,
+      "loss": 0.3041,
+      "step": 3264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957366464549155e-05,
+      "loss": 0.3009,
+      "step": 3265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573319324392063e-05,
+      "loss": 0.3027,
+      "step": 3266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572973866547e-05,
+      "loss": 0.2832,
+      "step": 3267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95726282719613e-05,
+      "loss": 0.2929,
+      "step": 3268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957228254063989e-05,
+      "loss": 0.3432,
+      "step": 3269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571936672587718e-05,
+      "loss": 0.298,
+      "step": 3270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571590667809714e-05,
+      "loss": 0.3029,
+      "step": 3271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571244526310834e-05,
+      "loss": 0.2858,
+      "step": 3272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570898248096012e-05,
+      "loss": 0.312,
+      "step": 3273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570551833170194e-05,
+      "loss": 0.3088,
+      "step": 3274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570205281538337e-05,
+      "loss": 0.2958,
+      "step": 3275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569858593205385e-05,
+      "loss": 0.3042,
+      "step": 3276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569511768176287e-05,
+      "loss": 0.3116,
+      "step": 3277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569164806456004e-05,
+      "loss": 0.2982,
+      "step": 3278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956881770804949e-05,
+      "loss": 0.2862,
+      "step": 3279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95684704729617e-05,
+      "loss": 0.3122,
+      "step": 3280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568123101197594e-05,
+      "loss": 0.3221,
+      "step": 3281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567775592762138e-05,
+      "loss": 0.3232,
+      "step": 3282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956742794766029e-05,
+      "loss": 0.3178,
+      "step": 3283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567080165897023e-05,
+      "loss": 0.3222,
+      "step": 3284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95667322474773e-05,
+      "loss": 0.3082,
+      "step": 3285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566384192406088e-05,
+      "loss": 0.3095,
+      "step": 3286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566036000688367e-05,
+      "loss": 0.2878,
+      "step": 3287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565687672329104e-05,
+      "loss": 0.291,
+      "step": 3288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956533920733328e-05,
+      "loss": 0.2703,
+      "step": 3289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564990605705863e-05,
+      "loss": 0.2982,
+      "step": 3290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956464186745184e-05,
+      "loss": 0.2922,
+      "step": 3291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564292992576192e-05,
+      "loss": 0.3093,
+      "step": 3292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95639439810839e-05,
+      "loss": 0.273,
+      "step": 3293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956359483297995e-05,
+      "loss": 0.3174,
+      "step": 3294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956324554826933e-05,
+      "loss": 0.2861,
+      "step": 3295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9562896126957026e-05,
+      "loss": 0.3193,
+      "step": 3296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9562546569048034e-05,
+      "loss": 0.3059,
+      "step": 3297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956219687454735e-05,
+      "loss": 0.2881,
+      "step": 3298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561847043459958e-05,
+      "loss": 0.3092,
+      "step": 3299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561497075790862e-05,
+      "loss": 0.3145,
+      "step": 3300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956114697154506e-05,
+      "loss": 0.2967,
+      "step": 3301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560796730727554e-05,
+      "loss": 0.3499,
+      "step": 3302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560446353343344e-05,
+      "loss": 0.3009,
+      "step": 3303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956009583939744e-05,
+      "loss": 0.2958,
+      "step": 3304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559745188894837e-05,
+      "loss": 0.2866,
+      "step": 3305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955939440184056e-05,
+      "loss": 0.3142,
+      "step": 3306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559043478239608e-05,
+      "loss": 0.3048,
+      "step": 3307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558692418097e-05,
+      "loss": 0.2878,
+      "step": 3308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558341221417744e-05,
+      "loss": 0.3065,
+      "step": 3309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955798988820686e-05,
+      "loss": 0.3015,
+      "step": 3310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955763841846937e-05,
+      "loss": 0.2596,
+      "step": 3311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557286812210293e-05,
+      "loss": 0.3435,
+      "step": 3312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556935069434643e-05,
+      "loss": 0.3176,
+      "step": 3313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556583190147457e-05,
+      "loss": 0.2895,
+      "step": 3314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955623117435375e-05,
+      "loss": 0.3039,
+      "step": 3315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955587902205856e-05,
+      "loss": 0.3244,
+      "step": 3316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955552673326691e-05,
+      "loss": 0.2862,
+      "step": 3317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555174307983833e-05,
+      "loss": 0.2805,
+      "step": 3318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554821746214365e-05,
+      "loss": 0.298,
+      "step": 3319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955446904796354e-05,
+      "loss": 0.3272,
+      "step": 3320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95541162132364e-05,
+      "loss": 0.3203,
+      "step": 3321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553763242037985e-05,
+      "loss": 0.3075,
+      "step": 3322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955341013437333e-05,
+      "loss": 0.2913,
+      "step": 3323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553056890247485e-05,
+      "loss": 0.3227,
+      "step": 3324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552703509665493e-05,
+      "loss": 0.3045,
+      "step": 3325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95523499926324e-05,
+      "loss": 0.3119,
+      "step": 3326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955199633915326e-05,
+      "loss": 0.2924,
+      "step": 3327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551642549233127e-05,
+      "loss": 0.3098,
+      "step": 3328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551288622877046e-05,
+      "loss": 0.2907,
+      "step": 3329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550934560090076e-05,
+      "loss": 0.2929,
+      "step": 3330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550580360877277e-05,
+      "loss": 0.3205,
+      "step": 3331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550226025243705e-05,
+      "loss": 0.3147,
+      "step": 3332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549871553194427e-05,
+      "loss": 0.285,
+      "step": 3333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549516944734498e-05,
+      "loss": 0.2894,
+      "step": 3334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954916219986899e-05,
+      "loss": 0.2876,
+      "step": 3335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548807318602967e-05,
+      "loss": 0.3247,
+      "step": 3336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95484523009415e-05,
+      "loss": 0.3214,
+      "step": 3337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548097146889654e-05,
+      "loss": 0.3318,
+      "step": 3338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954774185645251e-05,
+      "loss": 0.297,
+      "step": 3339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547386429635142e-05,
+      "loss": 0.3075,
+      "step": 3340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547030866442622e-05,
+      "loss": 0.3105,
+      "step": 3341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546675166880035e-05,
+      "loss": 0.3271,
+      "step": 3342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546319330952454e-05,
+      "loss": 0.3318,
+      "step": 3343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954596335866497e-05,
+      "loss": 0.3175,
+      "step": 3344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954560725002266e-05,
+      "loss": 0.31,
+      "step": 3345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954525100503062e-05,
+      "loss": 0.2849,
+      "step": 3346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544894623693927e-05,
+      "loss": 0.3163,
+      "step": 3347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544538106017682e-05,
+      "loss": 0.2797,
+      "step": 3348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544181452006975e-05,
+      "loss": 0.2947,
+      "step": 3349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954382466166689e-05,
+      "loss": 0.3136,
+      "step": 3350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543467735002543e-05,
+      "loss": 0.3117,
+      "step": 3351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543110672019017e-05,
+      "loss": 0.2873,
+      "step": 3352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542753472721417e-05,
+      "loss": 0.3025,
+      "step": 3353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542396137114844e-05,
+      "loss": 0.3059,
+      "step": 3354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542038665204407e-05,
+      "loss": 0.3101,
+      "step": 3355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541681056995205e-05,
+      "loss": 0.3113,
+      "step": 3356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954132331249235e-05,
+      "loss": 0.3245,
+      "step": 3357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540965431700952e-05,
+      "loss": 0.3696,
+      "step": 3358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954060741462612e-05,
+      "loss": 0.2602,
+      "step": 3359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954024926127297e-05,
+      "loss": 0.3391,
+      "step": 3360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539890971646624e-05,
+      "loss": 0.2699,
+      "step": 3361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953953254575219e-05,
+      "loss": 0.2802,
+      "step": 3362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539173983594794e-05,
+      "loss": 0.288,
+      "step": 3363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953881528517956e-05,
+      "loss": 0.3194,
+      "step": 3364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95384564505116e-05,
+      "loss": 0.2968,
+      "step": 3365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953809747959605e-05,
+      "loss": 0.305,
+      "step": 3366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953773837243803e-05,
+      "loss": 0.3119,
+      "step": 3367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537379129042677e-05,
+      "loss": 0.3172,
+      "step": 3368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537019749415122e-05,
+      "loss": 0.3302,
+      "step": 3369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536660233560495e-05,
+      "loss": 0.2865,
+      "step": 3370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953630058148393e-05,
+      "loss": 0.3015,
+      "step": 3371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535940793190566e-05,
+      "loss": 0.3175,
+      "step": 3372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535580868685544e-05,
+      "loss": 0.3137,
+      "step": 3373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535220807974002e-05,
+      "loss": 0.3402,
+      "step": 3374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534860611061087e-05,
+      "loss": 0.3147,
+      "step": 3375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953450027795194e-05,
+      "loss": 0.319,
+      "step": 3376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534139808651713e-05,
+      "loss": 0.2754,
+      "step": 3377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533779203165547e-05,
+      "loss": 0.3497,
+      "step": 3378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533418461498604e-05,
+      "loss": 0.2889,
+      "step": 3379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533057583656028e-05,
+      "loss": 0.3065,
+      "step": 3380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532696569642975e-05,
+      "loss": 0.3239,
+      "step": 3381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532335419464605e-05,
+      "loss": 0.3181,
+      "step": 3382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531974133126075e-05,
+      "loss": 0.3246,
+      "step": 3383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531612710632542e-05,
+      "loss": 0.3188,
+      "step": 3384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531251151989176e-05,
+      "loss": 0.3044,
+      "step": 3385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9530889457201137e-05,
+      "loss": 0.2934,
+      "step": 3386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9530527626273592e-05,
+      "loss": 0.2748,
+      "step": 3387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530165659211708e-05,
+      "loss": 0.306,
+      "step": 3388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952980355602066e-05,
+      "loss": 0.2838,
+      "step": 3389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529441316705614e-05,
+      "loss": 0.2985,
+      "step": 3390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529078941271748e-05,
+      "loss": 0.3258,
+      "step": 3391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528716429724238e-05,
+      "loss": 0.3378,
+      "step": 3392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528353782068264e-05,
+      "loss": 0.2888,
+      "step": 3393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527990998309e-05,
+      "loss": 0.3045,
+      "step": 3394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952762807845163e-05,
+      "loss": 0.3035,
+      "step": 3395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527265022501345e-05,
+      "loss": 0.3062,
+      "step": 3396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526901830463323e-05,
+      "loss": 0.2893,
+      "step": 3397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526538502342753e-05,
+      "loss": 0.2853,
+      "step": 3398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952617503814483e-05,
+      "loss": 0.2925,
+      "step": 3399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952581143787474e-05,
+      "loss": 0.264,
+      "step": 3400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525447701537677e-05,
+      "loss": 0.3213,
+      "step": 3401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952508382913884e-05,
+      "loss": 0.3094,
+      "step": 3402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524719820683424e-05,
+      "loss": 0.295,
+      "step": 3403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524355676176633e-05,
+      "loss": 0.3255,
+      "step": 3404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523991395623662e-05,
+      "loss": 0.2875,
+      "step": 3405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952362697902971e-05,
+      "loss": 0.3053,
+      "step": 3406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95232624264e-05,
+      "loss": 0.3106,
+      "step": 3407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522897737739725e-05,
+      "loss": 0.3135,
+      "step": 3408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522532913054096e-05,
+      "loss": 0.2826,
+      "step": 3409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952216795234833e-05,
+      "loss": 0.3271,
+      "step": 3410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521802855627634e-05,
+      "loss": 0.3005,
+      "step": 3411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521437622897225e-05,
+      "loss": 0.2868,
+      "step": 3412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521072254162324e-05,
+      "loss": 0.2977,
+      "step": 3413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520706749428142e-05,
+      "loss": 0.3104,
+      "step": 3414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520341108699906e-05,
+      "loss": 0.2967,
+      "step": 3415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519975331982835e-05,
+      "loss": 0.2869,
+      "step": 3416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519609419282156e-05,
+      "loss": 0.298,
+      "step": 3417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519243370603097e-05,
+      "loss": 0.2926,
+      "step": 3418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518877185950885e-05,
+      "loss": 0.3204,
+      "step": 3419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951851086533075e-05,
+      "loss": 0.2969,
+      "step": 3420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518144408747925e-05,
+      "loss": 0.3211,
+      "step": 3421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517777816207648e-05,
+      "loss": 0.3137,
+      "step": 3422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517411087715148e-05,
+      "loss": 0.2955,
+      "step": 3423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951704422327567e-05,
+      "loss": 0.3048,
+      "step": 3424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516677222894453e-05,
+      "loss": 0.3536,
+      "step": 3425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516310086576734e-05,
+      "loss": 0.3056,
+      "step": 3426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515942814327764e-05,
+      "loss": 0.2912,
+      "step": 3427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951557540615278e-05,
+      "loss": 0.2993,
+      "step": 3428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515207862057047e-05,
+      "loss": 0.2986,
+      "step": 3429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95148401820458e-05,
+      "loss": 0.3084,
+      "step": 3430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514472366124294e-05,
+      "loss": 0.3244,
+      "step": 3431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514104414297785e-05,
+      "loss": 0.3167,
+      "step": 3432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513736326571528e-05,
+      "loss": 0.2997,
+      "step": 3433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513368102950777e-05,
+      "loss": 0.3186,
+      "step": 3434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512999743440803e-05,
+      "loss": 0.2748,
+      "step": 3435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512631248046854e-05,
+      "loss": 0.2671,
+      "step": 3436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512262616774205e-05,
+      "loss": 0.2794,
+      "step": 3437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511893849628113e-05,
+      "loss": 0.28,
+      "step": 3438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951152494661385e-05,
+      "loss": 0.3095,
+      "step": 3439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511155907736682e-05,
+      "loss": 0.277,
+      "step": 3440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510786733001887e-05,
+      "loss": 0.3208,
+      "step": 3441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951041742241473e-05,
+      "loss": 0.2914,
+      "step": 3442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510047975980493e-05,
+      "loss": 0.2803,
+      "step": 3443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509678393704448e-05,
+      "loss": 0.2712,
+      "step": 3444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509308675591878e-05,
+      "loss": 0.3233,
+      "step": 3445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950893882164806e-05,
+      "loss": 0.276,
+      "step": 3446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950856883187828e-05,
+      "loss": 0.2872,
+      "step": 3447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508198706287827e-05,
+      "loss": 0.3016,
+      "step": 3448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950782844488198e-05,
+      "loss": 0.3226,
+      "step": 3449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950745804766603e-05,
+      "loss": 0.3209,
+      "step": 3450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950708751464527e-05,
+      "loss": 0.3112,
+      "step": 3451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506716845824998e-05,
+      "loss": 0.307,
+      "step": 3452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506346041210497e-05,
+      "loss": 0.2898,
+      "step": 3453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950597510080707e-05,
+      "loss": 0.3092,
+      "step": 3454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950560402462001e-05,
+      "loss": 0.2938,
+      "step": 3455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505232812654626e-05,
+      "loss": 0.3472,
+      "step": 3456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950486146491622e-05,
+      "loss": 0.298,
+      "step": 3457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504489981410085e-05,
+      "loss": 0.3138,
+      "step": 3458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950411836214154e-05,
+      "loss": 0.3422,
+      "step": 3459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503746607115883e-05,
+      "loss": 0.3382,
+      "step": 3460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503374716338436e-05,
+      "loss": 0.2902,
+      "step": 3461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95030026898145e-05,
+      "loss": 0.3125,
+      "step": 3462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502630527549393e-05,
+      "loss": 0.3102,
+      "step": 3463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502258229548436e-05,
+      "loss": 0.2971,
+      "step": 3464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501885795816937e-05,
+      "loss": 0.3346,
+      "step": 3465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501513226360226e-05,
+      "loss": 0.2878,
+      "step": 3466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501140521183616e-05,
+      "loss": 0.3179,
+      "step": 3467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500767680292436e-05,
+      "loss": 0.3173,
+      "step": 3468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950039470369201e-05,
+      "loss": 0.3255,
+      "step": 3469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500021591387667e-05,
+      "loss": 0.304,
+      "step": 3470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499648343384738e-05,
+      "loss": 0.2962,
+      "step": 3471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499274959688546e-05,
+      "loss": 0.3023,
+      "step": 3472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498901440304436e-05,
+      "loss": 0.3092,
+      "step": 3473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498527785237737e-05,
+      "loss": 0.317,
+      "step": 3474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949815399449378e-05,
+      "loss": 0.3038,
+      "step": 3475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497780068077922e-05,
+      "loss": 0.3043,
+      "step": 3476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497406005995486e-05,
+      "loss": 0.2804,
+      "step": 3477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497031808251826e-05,
+      "loss": 0.3163,
+      "step": 3478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496657474852286e-05,
+      "loss": 0.2911,
+      "step": 3479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496283005802205e-05,
+      "loss": 0.3007,
+      "step": 3480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495908401106942e-05,
+      "loss": 0.2957,
+      "step": 3481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495533660771842e-05,
+      "loss": 0.2987,
+      "step": 3482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495158784802258e-05,
+      "loss": 0.3156,
+      "step": 3483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949478377320355e-05,
+      "loss": 0.2864,
+      "step": 3484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494408625981064e-05,
+      "loss": 0.3,
+      "step": 3485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949403334314017e-05,
+      "loss": 0.2971,
+      "step": 3486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493657924686223e-05,
+      "loss": 0.3215,
+      "step": 3487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493282370624588e-05,
+      "loss": 0.2898,
+      "step": 3488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492906680960626e-05,
+      "loss": 0.3205,
+      "step": 3489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492530855699706e-05,
+      "loss": 0.3195,
+      "step": 3490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492154894847192e-05,
+      "loss": 0.2949,
+      "step": 3491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949177879840846e-05,
+      "loss": 0.2955,
+      "step": 3492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491402566388882e-05,
+      "loss": 0.3313,
+      "step": 3493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491026198793826e-05,
+      "loss": 0.2932,
+      "step": 3494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490649695628678e-05,
+      "loss": 0.2781,
+      "step": 3495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490273056898803e-05,
+      "loss": 0.3002,
+      "step": 3496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489896282609592e-05,
+      "loss": 0.3161,
+      "step": 3497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489519372766422e-05,
+      "loss": 0.3283,
+      "step": 3498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948914232737468e-05,
+      "loss": 0.2898,
+      "step": 3499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488765146439745e-05,
+      "loss": 0.2909,
+      "step": 3500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948838782996701e-05,
+      "loss": 0.3012,
+      "step": 3501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488010377961864e-05,
+      "loss": 0.3163,
+      "step": 3502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487632790429697e-05,
+      "loss": 0.2973,
+      "step": 3503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487255067375907e-05,
+      "loss": 0.32,
+      "step": 3504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486877208805885e-05,
+      "loss": 0.3069,
+      "step": 3505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948649921472503e-05,
+      "loss": 0.2927,
+      "step": 3506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486121085138734e-05,
+      "loss": 0.2837,
+      "step": 3507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948574282005241e-05,
+      "loss": 0.2928,
+      "step": 3508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 0.2951,
+      "step": 3509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948498588340128e-05,
+      "loss": 0.2902,
+      "step": 3510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948460721184728e-05,
+      "loss": 0.292,
+      "step": 3511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484228404814873e-05,
+      "loss": 0.2911,
+      "step": 3512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948384946230947e-05,
+      "loss": 0.324,
+      "step": 3513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948347038433648e-05,
+      "loss": 0.28,
+      "step": 3514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948309117090132e-05,
+      "loss": 0.3172,
+      "step": 3515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482711822009404e-05,
+      "loss": 0.2755,
+      "step": 3516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482332337666156e-05,
+      "loss": 0.2852,
+      "step": 3517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481952717876994e-05,
+      "loss": 0.2624,
+      "step": 3518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481572962647334e-05,
+      "loss": 0.299,
+      "step": 3519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948119307198261e-05,
+      "loss": 0.3125,
+      "step": 3520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480813045888248e-05,
+      "loss": 0.2978,
+      "step": 3521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948043288436967e-05,
+      "loss": 0.2716,
+      "step": 3522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480052587432306e-05,
+      "loss": 0.2931,
+      "step": 3523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479672155081598e-05,
+      "loss": 0.2703,
+      "step": 3524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947929158732297e-05,
+      "loss": 0.2935,
+      "step": 3525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947891088416186e-05,
+      "loss": 0.3375,
+      "step": 3526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947853004560371e-05,
+      "loss": 0.3012,
+      "step": 3527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478149071653956e-05,
+      "loss": 0.3261,
+      "step": 3528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477767962318043e-05,
+      "loss": 0.3417,
+      "step": 3529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477386717601413e-05,
+      "loss": 0.3261,
+      "step": 3530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477005337509507e-05,
+      "loss": 0.357,
+      "step": 3531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947662382204778e-05,
+      "loss": 0.33,
+      "step": 3532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476242171221683e-05,
+      "loss": 0.3225,
+      "step": 3533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947586038503666e-05,
+      "loss": 0.2719,
+      "step": 3534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475478463498167e-05,
+      "loss": 0.3066,
+      "step": 3535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475096406611663e-05,
+      "loss": 0.31,
+      "step": 3536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474714214382602e-05,
+      "loss": 0.3023,
+      "step": 3537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474331886816444e-05,
+      "loss": 0.3248,
+      "step": 3538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947394942391865e-05,
+      "loss": 0.329,
+      "step": 3539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947356682569468e-05,
+      "loss": 0.2914,
+      "step": 3540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473184092150007e-05,
+      "loss": 0.2993,
+      "step": 3541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947280122329009e-05,
+      "loss": 0.3184,
+      "step": 3542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472418219120403e-05,
+      "loss": 0.2928,
+      "step": 3543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947203507964641e-05,
+      "loss": 0.2943,
+      "step": 3544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471651804873595e-05,
+      "loss": 0.2975,
+      "step": 3545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471268394807424e-05,
+      "loss": 0.2945,
+      "step": 3546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470884849453378e-05,
+      "loss": 0.3084,
+      "step": 3547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470501168816928e-05,
+      "loss": 0.3299,
+      "step": 3548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470117352903567e-05,
+      "loss": 0.2986,
+      "step": 3549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469733401718766e-05,
+      "loss": 0.2675,
+      "step": 3550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469349315268015e-05,
+      "loss": 0.2929,
+      "step": 3551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468965093556802e-05,
+      "loss": 0.2909,
+      "step": 3552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946858073659061e-05,
+      "loss": 0.3183,
+      "step": 3553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946819624437493e-05,
+      "loss": 0.3089,
+      "step": 3554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467811616915258e-05,
+      "loss": 0.3216,
+      "step": 3555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467426854217085e-05,
+      "loss": 0.3284,
+      "step": 3556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467041956285907e-05,
+      "loss": 0.3086,
+      "step": 3557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466656923127224e-05,
+      "loss": 0.2882,
+      "step": 3558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466271754746537e-05,
+      "loss": 0.2885,
+      "step": 3559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465886451149342e-05,
+      "loss": 0.3127,
+      "step": 3560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465501012341146e-05,
+      "loss": 0.2897,
+      "step": 3561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465115438327453e-05,
+      "loss": 0.3161,
+      "step": 3562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464729729113777e-05,
+      "loss": 0.3143,
+      "step": 3563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464343884705615e-05,
+      "loss": 0.3056,
+      "step": 3564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463957905108492e-05,
+      "loss": 0.3032,
+      "step": 3565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463571790327915e-05,
+      "loss": 0.3001,
+      "step": 3566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463185540369398e-05,
+      "loss": 0.3211,
+      "step": 3567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9462799155238462e-05,
+      "loss": 0.2858,
+      "step": 3568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946241263494062e-05,
+      "loss": 0.3346,
+      "step": 3569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94620259794814e-05,
+      "loss": 0.304,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946163918886632e-05,
+      "loss": 0.2993,
+      "step": 3571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946125226310091e-05,
+      "loss": 0.3101,
+      "step": 3572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946086520219069e-05,
+      "loss": 0.3115,
+      "step": 3573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460478006141193e-05,
+      "loss": 0.3131,
+      "step": 3574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460090674957953e-05,
+      "loss": 0.3348,
+      "step": 3575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459703208646496e-05,
+      "loss": 0.2806,
+      "step": 3576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459315607212363e-05,
+      "loss": 0.2917,
+      "step": 3577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458927870661082e-05,
+      "loss": 0.3065,
+      "step": 3578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458539998998197e-05,
+      "loss": 0.2885,
+      "step": 3579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945815199222925e-05,
+      "loss": 0.3221,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457763850359782e-05,
+      "loss": 0.3117,
+      "step": 3581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457375573395334e-05,
+      "loss": 0.3333,
+      "step": 3582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456987161341455e-05,
+      "loss": 0.3229,
+      "step": 3583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456598614203696e-05,
+      "loss": 0.2981,
+      "step": 3584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456209931987597e-05,
+      "loss": 0.2944,
+      "step": 3585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455821114698722e-05,
+      "loss": 0.2972,
+      "step": 3586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455432162342617e-05,
+      "loss": 0.3029,
+      "step": 3587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945504307492484e-05,
+      "loss": 0.3083,
+      "step": 3588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945465385245095e-05,
+      "loss": 0.2887,
+      "step": 3589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454264494926507e-05,
+      "loss": 0.3227,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945387500235707e-05,
+      "loss": 0.2974,
+      "step": 3591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453485374748204e-05,
+      "loss": 0.3077,
+      "step": 3592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453095612105474e-05,
+      "loss": 0.3034,
+      "step": 3593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945270571443445e-05,
+      "loss": 0.3384,
+      "step": 3594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452315681740693e-05,
+      "loss": 0.2826,
+      "step": 3595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451925514029785e-05,
+      "loss": 0.2999,
+      "step": 3596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451535211307294e-05,
+      "loss": 0.3226,
+      "step": 3597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451144773578797e-05,
+      "loss": 0.3261,
+      "step": 3598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450754200849866e-05,
+      "loss": 0.3062,
+      "step": 3599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450363493126083e-05,
+      "loss": 0.2816,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944997265041303e-05,
+      "loss": 0.3178,
+      "step": 3601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449581672716293e-05,
+      "loss": 0.294,
+      "step": 3602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944919056004145e-05,
+      "loss": 0.3056,
+      "step": 3603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944879931239409e-05,
+      "loss": 0.3117,
+      "step": 3604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94484079297798e-05,
+      "loss": 0.3223,
+      "step": 3605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448016412204177e-05,
+      "loss": 0.3455,
+      "step": 3606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447624759672807e-05,
+      "loss": 0.2976,
+      "step": 3607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447232972191288e-05,
+      "loss": 0.2943,
+      "step": 3608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446841049765212e-05,
+      "loss": 0.2832,
+      "step": 3609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944644899240018e-05,
+      "loss": 0.3195,
+      "step": 3610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446056800101796e-05,
+      "loss": 0.3151,
+      "step": 3611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445664472875652e-05,
+      "loss": 0.2952,
+      "step": 3612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445272010727362e-05,
+      "loss": 0.2992,
+      "step": 3613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944487941366253e-05,
+      "loss": 0.3007,
+      "step": 3614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944448668168676e-05,
+      "loss": 0.3163,
+      "step": 3615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944409381480566e-05,
+      "loss": 0.3055,
+      "step": 3616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443700813024854e-05,
+      "loss": 0.3075,
+      "step": 3617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443307676349943e-05,
+      "loss": 0.2964,
+      "step": 3618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442914404786547e-05,
+      "loss": 0.2986,
+      "step": 3619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944252099834028e-05,
+      "loss": 0.3062,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442127457016768e-05,
+      "loss": 0.321,
+      "step": 3621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944173378082163e-05,
+      "loss": 0.3018,
+      "step": 3622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441339969760486e-05,
+      "loss": 0.3344,
+      "step": 3623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440946023838964e-05,
+      "loss": 0.287,
+      "step": 3624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944055194306269e-05,
+      "loss": 0.2992,
+      "step": 3625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440157727437295e-05,
+      "loss": 0.2692,
+      "step": 3626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943976337696841e-05,
+      "loss": 0.3021,
+      "step": 3627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439368891661665e-05,
+      "loss": 0.2994,
+      "step": 3628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94389742715227e-05,
+      "loss": 0.2977,
+      "step": 3629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438579516557144e-05,
+      "loss": 0.2805,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943818462677064e-05,
+      "loss": 0.3283,
+      "step": 3631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943778960216883e-05,
+      "loss": 0.3107,
+      "step": 3632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437394442757354e-05,
+      "loss": 0.2942,
+      "step": 3633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436999148541858e-05,
+      "loss": 0.3011,
+      "step": 3634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436603719527988e-05,
+      "loss": 0.2729,
+      "step": 3635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436208155721394e-05,
+      "loss": 0.3057,
+      "step": 3636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435812457127724e-05,
+      "loss": 0.3062,
+      "step": 3637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943541662375263e-05,
+      "loss": 0.2938,
+      "step": 3638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435020655601767e-05,
+      "loss": 0.3042,
+      "step": 3639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943462455268079e-05,
+      "loss": 0.3128,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943422831499536e-05,
+      "loss": 0.2934,
+      "step": 3641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943383194255113e-05,
+      "loss": 0.3284,
+      "step": 3642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943343543535377e-05,
+      "loss": 0.2729,
+      "step": 3643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433038793408944e-05,
+      "loss": 0.2966,
+      "step": 3644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432642016722307e-05,
+      "loss": 0.305,
+      "step": 3645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432245105299534e-05,
+      "loss": 0.3211,
+      "step": 3646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94318480591463e-05,
+      "loss": 0.305,
+      "step": 3647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431450878268265e-05,
+      "loss": 0.2762,
+      "step": 3648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431053562671108e-05,
+      "loss": 0.298,
+      "step": 3649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430656112360506e-05,
+      "loss": 0.3004,
+      "step": 3650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943025852734213e-05,
+      "loss": 0.2938,
+      "step": 3651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942986080762167e-05,
+      "loss": 0.2954,
+      "step": 3652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429462953204794e-05,
+      "loss": 0.2864,
+      "step": 3653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942906496409719e-05,
+      "loss": 0.3172,
+      "step": 3654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942866684030455e-05,
+      "loss": 0.3323,
+      "step": 3655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428268581832553e-05,
+      "loss": 0.2789,
+      "step": 3656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9427870188686888e-05,
+      "loss": 0.3035,
+      "step": 3657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942747166087325e-05,
+      "loss": 0.2941,
+      "step": 3658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427072998397327e-05,
+      "loss": 0.2826,
+      "step": 3659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426674201264814e-05,
+      "loss": 0.278,
+      "step": 3660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942627526948141e-05,
+      "loss": 0.2795,
+      "step": 3661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425876203052816e-05,
+      "loss": 0.275,
+      "step": 3662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425477001984726e-05,
+      "loss": 0.3195,
+      "step": 3663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942507766628284e-05,
+      "loss": 0.2743,
+      "step": 3664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424678195952873e-05,
+      "loss": 0.3102,
+      "step": 3665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942427859100052e-05,
+      "loss": 0.2685,
+      "step": 3666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94238788514315e-05,
+      "loss": 0.3124,
+      "step": 3667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942347897725151e-05,
+      "loss": 0.3264,
+      "step": 3668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423078968466274e-05,
+      "loss": 0.3104,
+      "step": 3669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422678825081495e-05,
+      "loss": 0.3267,
+      "step": 3670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94222785471029e-05,
+      "loss": 0.2792,
+      "step": 3671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421878134536196e-05,
+      "loss": 0.3092,
+      "step": 3672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942147758738711e-05,
+      "loss": 0.3205,
+      "step": 3673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421076905661356e-05,
+      "loss": 0.3035,
+      "step": 3674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420676089364666e-05,
+      "loss": 0.32,
+      "step": 3675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420275138502758e-05,
+      "loss": 0.285,
+      "step": 3676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419874053081365e-05,
+      "loss": 0.2958,
+      "step": 3677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941947283310621e-05,
+      "loss": 0.2818,
+      "step": 3678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941907147858303e-05,
+      "loss": 0.3164,
+      "step": 3679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941866998951756e-05,
+      "loss": 0.2951,
+      "step": 3680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418268365915522e-05,
+      "loss": 0.2859,
+      "step": 3681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417866607782667e-05,
+      "loss": 0.333,
+      "step": 3682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941746471512472e-05,
+      "loss": 0.2975,
+      "step": 3683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417062687947434e-05,
+      "loss": 0.3145,
+      "step": 3684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416660526256546e-05,
+      "loss": 0.3024,
+      "step": 3685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941625823005781e-05,
+      "loss": 0.3005,
+      "step": 3686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415855799356954e-05,
+      "loss": 0.2896,
+      "step": 3687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415453234159736e-05,
+      "loss": 0.3012,
+      "step": 3688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941505053447191e-05,
+      "loss": 0.3056,
+      "step": 3689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414647700299225e-05,
+      "loss": 0.2795,
+      "step": 3690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414244731647434e-05,
+      "loss": 0.3056,
+      "step": 3691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413841628522295e-05,
+      "loss": 0.278,
+      "step": 3692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941343839092956e-05,
+      "loss": 0.3116,
+      "step": 3693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413035018875e-05,
+      "loss": 0.3,
+      "step": 3694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412631512364368e-05,
+      "loss": 0.2892,
+      "step": 3695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412227871403428e-05,
+      "loss": 0.289,
+      "step": 3696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411824095997945e-05,
+      "loss": 0.3391,
+      "step": 3697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411420186153694e-05,
+      "loss": 0.3117,
+      "step": 3698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411016141876438e-05,
+      "loss": 0.3049,
+      "step": 3699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941061196317195e-05,
+      "loss": 0.2794,
+      "step": 3700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410207650046005e-05,
+      "loss": 0.2844,
+      "step": 3701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409803202504375e-05,
+      "loss": 0.3015,
+      "step": 3702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940939862055284e-05,
+      "loss": 0.2983,
+      "step": 3703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408993904197176e-05,
+      "loss": 0.291,
+      "step": 3704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408589053443167e-05,
+      "loss": 0.3116,
+      "step": 3705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408184068296593e-05,
+      "loss": 0.2854,
+      "step": 3706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940777894876324e-05,
+      "loss": 0.2858,
+      "step": 3707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94073736948489e-05,
+      "loss": 0.289,
+      "step": 3708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940696830655935e-05,
+      "loss": 0.2907,
+      "step": 3709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406562783900394e-05,
+      "loss": 0.3408,
+      "step": 3710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406157126877814e-05,
+      "loss": 0.2961,
+      "step": 3711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940575133549741e-05,
+      "loss": 0.295,
+      "step": 3712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405345409764973e-05,
+      "loss": 0.3142,
+      "step": 3713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940493934968631e-05,
+      "loss": 0.3058,
+      "step": 3714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404533155267215e-05,
+      "loss": 0.3036,
+      "step": 3715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404126826513487e-05,
+      "loss": 0.3113,
+      "step": 3716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940372036343094e-05,
+      "loss": 0.2851,
+      "step": 3717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403313766025372e-05,
+      "loss": 0.3219,
+      "step": 3718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940290703430259e-05,
+      "loss": 0.3129,
+      "step": 3719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940250016826841e-05,
+      "loss": 0.3062,
+      "step": 3720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402093167928638e-05,
+      "loss": 0.2824,
+      "step": 3721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401686033289092e-05,
+      "loss": 0.3081,
+      "step": 3722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401278764355586e-05,
+      "loss": 0.2967,
+      "step": 3723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400871361133937e-05,
+      "loss": 0.3305,
+      "step": 3724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400463823629963e-05,
+      "loss": 0.3108,
+      "step": 3725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400056151849487e-05,
+      "loss": 0.3016,
+      "step": 3726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399648345798332e-05,
+      "loss": 0.3096,
+      "step": 3727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399240405482324e-05,
+      "loss": 0.3035,
+      "step": 3728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398832330907287e-05,
+      "loss": 0.3152,
+      "step": 3729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398424122079053e-05,
+      "loss": 0.2895,
+      "step": 3730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398015779003453e-05,
+      "loss": 0.3054,
+      "step": 3731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397607301686317e-05,
+      "loss": 0.3014,
+      "step": 3732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397198690133483e-05,
+      "loss": 0.3126,
+      "step": 3733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396789944350782e-05,
+      "loss": 0.3383,
+      "step": 3734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396381064344062e-05,
+      "loss": 0.3049,
+      "step": 3735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395972050119155e-05,
+      "loss": 0.3025,
+      "step": 3736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395562901681904e-05,
+      "loss": 0.3415,
+      "step": 3737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395153619038158e-05,
+      "loss": 0.3021,
+      "step": 3738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939474420219376e-05,
+      "loss": 0.2977,
+      "step": 3739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939433465115456e-05,
+      "loss": 0.304,
+      "step": 3740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393924965926408e-05,
+      "loss": 0.2987,
+      "step": 3741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393515146515152e-05,
+      "loss": 0.3132,
+      "step": 3742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393105192926647e-05,
+      "loss": 0.3052,
+      "step": 3743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9392695105166758e-05,
+      "loss": 0.309,
+      "step": 3744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939228488324133e-05,
+      "loss": 0.2986,
+      "step": 3745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939187452715623e-05,
+      "loss": 0.3361,
+      "step": 3746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391464036917315e-05,
+      "loss": 0.3027,
+      "step": 3747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391053412530452e-05,
+      "loss": 0.2992,
+      "step": 3748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390642654001506e-05,
+      "loss": 0.2864,
+      "step": 3749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939023176133634e-05,
+      "loss": 0.3918,
+      "step": 3750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938982073454083e-05,
+      "loss": 0.2951,
+      "step": 3751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389409573620845e-05,
+      "loss": 0.322,
+      "step": 3752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388998278582253e-05,
+      "loss": 0.3025,
+      "step": 3753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388586849430936e-05,
+      "loss": 0.2826,
+      "step": 3754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388175286172765e-05,
+      "loss": 0.32,
+      "step": 3755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387763588813625e-05,
+      "loss": 0.2864,
+      "step": 3756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938735175735939e-05,
+      "loss": 0.3023,
+      "step": 3757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938693979181595e-05,
+      "loss": 0.2986,
+      "step": 3758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386527692189177e-05,
+      "loss": 0.3291,
+      "step": 3759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386115458484977e-05,
+      "loss": 0.2937,
+      "step": 3760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385703090709217e-05,
+      "loss": 0.2804,
+      "step": 3761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385290588867803e-05,
+      "loss": 0.3309,
+      "step": 3762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938487795296662e-05,
+      "loss": 0.2805,
+      "step": 3763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384465183011563e-05,
+      "loss": 0.2777,
+      "step": 3764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938405227900853e-05,
+      "loss": 0.3123,
+      "step": 3765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383639240963416e-05,
+      "loss": 0.3076,
+      "step": 3766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383226068882128e-05,
+      "loss": 0.3007,
+      "step": 3767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382812762770557e-05,
+      "loss": 0.2841,
+      "step": 3768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382399322634612e-05,
+      "loss": 0.3024,
+      "step": 3769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93819857484802e-05,
+      "loss": 0.2954,
+      "step": 3770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381572040313226e-05,
+      "loss": 0.2848,
+      "step": 3771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93811581981396e-05,
+      "loss": 0.3329,
+      "step": 3772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380744221965234e-05,
+      "loss": 0.3101,
+      "step": 3773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380330111796044e-05,
+      "loss": 0.3119,
+      "step": 3774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937991586763794e-05,
+      "loss": 0.3028,
+      "step": 3775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937950148949684e-05,
+      "loss": 0.2857,
+      "step": 3776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379086977378664e-05,
+      "loss": 0.3087,
+      "step": 3777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378672331289336e-05,
+      "loss": 0.28,
+      "step": 3778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378257551234774e-05,
+      "loss": 0.3006,
+      "step": 3779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377842637220902e-05,
+      "loss": 0.3421,
+      "step": 3780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377427589253654e-05,
+      "loss": 0.2841,
+      "step": 3781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377012407338955e-05,
+      "loss": 0.314,
+      "step": 3782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376597091482728e-05,
+      "loss": 0.3156,
+      "step": 3783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376181641690918e-05,
+      "loss": 0.3148,
+      "step": 3784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937576605796945e-05,
+      "loss": 0.3066,
+      "step": 3785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375350340324263e-05,
+      "loss": 0.3048,
+      "step": 3786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374934488761297e-05,
+      "loss": 0.2964,
+      "step": 3787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374518503286487e-05,
+      "loss": 0.2911,
+      "step": 3788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374102383905784e-05,
+      "loss": 0.3119,
+      "step": 3789
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373686130625124e-05,
+      "loss": 0.3052,
+      "step": 3790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373269743450455e-05,
+      "loss": 0.302,
+      "step": 3791
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937285322238772e-05,
+      "loss": 0.2997,
+      "step": 3792
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937243656744288e-05,
+      "loss": 0.2866,
+      "step": 3793
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372019778621876e-05,
+      "loss": 0.304,
+      "step": 3794
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371602855930667e-05,
+      "loss": 0.3052,
+      "step": 3795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93711857993752e-05,
+      "loss": 0.3044,
+      "step": 3796
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937076860896145e-05,
+      "loss": 0.2829,
+      "step": 3797
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937035128469536e-05,
+      "loss": 0.2898,
+      "step": 3798
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369933826582895e-05,
+      "loss": 0.286,
+      "step": 3799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936951623463002e-05,
+      "loss": 0.2972,
+      "step": 3800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369098508842695e-05,
+      "loss": 0.3191,
+      "step": 3801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368680649226897e-05,
+      "loss": 0.325,
+      "step": 3802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368262655788586e-05,
+      "loss": 0.2934,
+      "step": 3803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367844528533734e-05,
+      "loss": 0.3036,
+      "step": 3804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367426267468313e-05,
+      "loss": 0.3088,
+      "step": 3805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93670078725983e-05,
+      "loss": 0.2883,
+      "step": 3806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366589343929676e-05,
+      "loss": 0.2933,
+      "step": 3807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366170681468407e-05,
+      "loss": 0.298,
+      "step": 3808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365751885220482e-05,
+      "loss": 0.2862,
+      "step": 3809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365332955191882e-05,
+      "loss": 0.318,
+      "step": 3810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936491389138859e-05,
+      "loss": 0.3042,
+      "step": 3811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936449469381659e-05,
+      "loss": 0.3144,
+      "step": 3812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364075362481876e-05,
+      "loss": 0.2769,
+      "step": 3813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936365589739043e-05,
+      "loss": 0.3097,
+      "step": 3814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363236298548248e-05,
+      "loss": 0.3231,
+      "step": 3815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936281656596132e-05,
+      "loss": 0.3032,
+      "step": 3816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362396699635647e-05,
+      "loss": 0.2691,
+      "step": 3817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361976699577226e-05,
+      "loss": 0.3128,
+      "step": 3818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936155656579205e-05,
+      "loss": 0.3321,
+      "step": 3819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361136298286124e-05,
+      "loss": 0.3012,
+      "step": 3820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360715897065456e-05,
+      "loss": 0.2719,
+      "step": 3821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936029536213604e-05,
+      "loss": 0.2905,
+      "step": 3822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359874693503892e-05,
+      "loss": 0.2748,
+      "step": 3823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359453891175017e-05,
+      "loss": 0.304,
+      "step": 3824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935903295515543e-05,
+      "loss": 0.289,
+      "step": 3825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358611885451134e-05,
+      "loss": 0.3113,
+      "step": 3826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358190682068154e-05,
+      "loss": 0.2943,
+      "step": 3827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357769345012506e-05,
+      "loss": 0.2832,
+      "step": 3828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93573478742902e-05,
+      "loss": 0.3092,
+      "step": 3829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935692626990726e-05,
+      "loss": 0.2923,
+      "step": 3830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356504531869712e-05,
+      "loss": 0.2878,
+      "step": 3831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356082660183578e-05,
+      "loss": 0.3294,
+      "step": 3832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355660654854882e-05,
+      "loss": 0.2827,
+      "step": 3833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355238515889653e-05,
+      "loss": 0.3099,
+      "step": 3834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935481624329392e-05,
+      "loss": 0.2658,
+      "step": 3835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935439383707372e-05,
+      "loss": 0.2929,
+      "step": 3836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935397129723508e-05,
+      "loss": 0.2758,
+      "step": 3837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353548623784043e-05,
+      "loss": 0.3057,
+      "step": 3838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9353125816726638e-05,
+      "loss": 0.316,
+      "step": 3839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352702876068905e-05,
+      "loss": 0.2769,
+      "step": 3840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352279801816892e-05,
+      "loss": 0.2813,
+      "step": 3841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935185659397664e-05,
+      "loss": 0.3169,
+      "step": 3842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935143325255419e-05,
+      "loss": 0.3146,
+      "step": 3843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935100977755559e-05,
+      "loss": 0.3225,
+      "step": 3844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350586168986894e-05,
+      "loss": 0.3121,
+      "step": 3845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 0.3012,
+      "step": 3846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934973855116341e-05,
+      "loss": 0.2715,
+      "step": 3847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934931454192073e-05,
+      "loss": 0.3428,
+      "step": 3848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934889039913216e-05,
+      "loss": 0.3019,
+      "step": 3849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348466122803768e-05,
+      "loss": 0.2874,
+      "step": 3850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348041712941608e-05,
+      "loss": 0.3259,
+      "step": 3851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347617169551748e-05,
+      "loss": 0.2884,
+      "step": 3852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934719249264025e-05,
+      "loss": 0.3325,
+      "step": 3853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346767682213176e-05,
+      "loss": 0.3167,
+      "step": 3854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346342738276593e-05,
+      "loss": 0.3159,
+      "step": 3855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345917660836583e-05,
+      "loss": 0.3093,
+      "step": 3856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345492449899203e-05,
+      "loss": 0.293,
+      "step": 3857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345067105470535e-05,
+      "loss": 0.3023,
+      "step": 3858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344641627556655e-05,
+      "loss": 0.2682,
+      "step": 3859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344216016163636e-05,
+      "loss": 0.3068,
+      "step": 3860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343790271297564e-05,
+      "loss": 0.3072,
+      "step": 3861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343364392964514e-05,
+      "loss": 0.3244,
+      "step": 3862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934293838117057e-05,
+      "loss": 0.3125,
+      "step": 3863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934251223592182e-05,
+      "loss": 0.3025,
+      "step": 3864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934208595722435e-05,
+      "loss": 0.3147,
+      "step": 3865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934165954508425e-05,
+      "loss": 0.2984,
+      "step": 3866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934123299950761e-05,
+      "loss": 0.2954,
+      "step": 3867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934080632050052e-05,
+      "loss": 0.3164,
+      "step": 3868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340379508069076e-05,
+      "loss": 0.343,
+      "step": 3869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933995256221938e-05,
+      "loss": 0.2698,
+      "step": 3870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933952548295752e-05,
+      "loss": 0.2858,
+      "step": 3871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933909827028961e-05,
+      "loss": 0.2845,
+      "step": 3872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933867092422174e-05,
+      "loss": 0.2965,
+      "step": 3873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933824344476002e-05,
+      "loss": 0.2801,
+      "step": 3874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337815831910557e-05,
+      "loss": 0.2995,
+      "step": 3875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933738808567946e-05,
+      "loss": 0.3088,
+      "step": 3876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933696020607283e-05,
+      "loss": 0.2923,
+      "step": 3877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933653219309679e-05,
+      "loss": 0.3128,
+      "step": 3878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933610404675745e-05,
+      "loss": 0.2498,
+      "step": 3879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933567576706092e-05,
+      "loss": 0.2619,
+      "step": 3880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335247354013323e-05,
+      "loss": 0.3057,
+      "step": 3881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933481880762078e-05,
+      "loss": 0.2654,
+      "step": 3882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933439012788941e-05,
+      "loss": 0.2824,
+      "step": 3883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333961314825333e-05,
+      "loss": 0.2928,
+      "step": 3884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933353236843468e-05,
+      "loss": 0.2784,
+      "step": 3885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333103288723575e-05,
+      "loss": 0.3148,
+      "step": 3886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332674075698146e-05,
+      "loss": 0.2974,
+      "step": 3887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332244729364528e-05,
+      "loss": 0.2881,
+      "step": 3888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933181524972885e-05,
+      "loss": 0.2885,
+      "step": 3889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331385636797247e-05,
+      "loss": 0.3103,
+      "step": 3890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933095589057586e-05,
+      "loss": 0.3077,
+      "step": 3891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933052601107082e-05,
+      "loss": 0.2975,
+      "step": 3892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330095998288274e-05,
+      "loss": 0.3101,
+      "step": 3893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932966585223436e-05,
+      "loss": 0.2716,
+      "step": 3894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329235572915227e-05,
+      "loss": 0.305,
+      "step": 3895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328805160337015e-05,
+      "loss": 0.2763,
+      "step": 3896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328374614505874e-05,
+      "loss": 0.2605,
+      "step": 3897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327943935427956e-05,
+      "loss": 0.3474,
+      "step": 3898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327513123109415e-05,
+      "loss": 0.327,
+      "step": 3899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93270821775564e-05,
+      "loss": 0.3059,
+      "step": 3900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326651098775065e-05,
+      "loss": 0.306,
+      "step": 3901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326219886771573e-05,
+      "loss": 0.3173,
+      "step": 3902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325788541552085e-05,
+      "loss": 0.2956,
+      "step": 3903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325357063122754e-05,
+      "loss": 0.3329,
+      "step": 3904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324925451489747e-05,
+      "loss": 0.3087,
+      "step": 3905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324493706659234e-05,
+      "loss": 0.2892,
+      "step": 3906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324061828637374e-05,
+      "loss": 0.3019,
+      "step": 3907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323629817430344e-05,
+      "loss": 0.2948,
+      "step": 3908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932319767304431e-05,
+      "loss": 0.2828,
+      "step": 3909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932276539548544e-05,
+      "loss": 0.3082,
+      "step": 3910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322332984759922e-05,
+      "loss": 0.304,
+      "step": 3911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932190044087392e-05,
+      "loss": 0.3061,
+      "step": 3912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932146776383362e-05,
+      "loss": 0.3121,
+      "step": 3913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93210349536452e-05,
+      "loss": 0.3177,
+      "step": 3914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932060201031484e-05,
+      "loss": 0.2767,
+      "step": 3915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320168933848723e-05,
+      "loss": 0.3025,
+      "step": 3916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319735724253045e-05,
+      "loss": 0.3084,
+      "step": 3917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319302381533986e-05,
+      "loss": 0.2705,
+      "step": 3918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318868905697732e-05,
+      "loss": 0.2803,
+      "step": 3919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318435296750485e-05,
+      "loss": 0.2866,
+      "step": 3920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318001554698428e-05,
+      "loss": 0.2916,
+      "step": 3921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931756767954777e-05,
+      "loss": 0.2507,
+      "step": 3922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317133671304695e-05,
+      "loss": 0.3173,
+      "step": 3923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316699529975408e-05,
+      "loss": 0.275,
+      "step": 3924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316265255566114e-05,
+      "loss": 0.2712,
+      "step": 3925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315830848083008e-05,
+      "loss": 0.2982,
+      "step": 3926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93153963075323e-05,
+      "loss": 0.3065,
+      "step": 3927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93149616339202e-05,
+      "loss": 0.3072,
+      "step": 3928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9314526827252913e-05,
+      "loss": 0.2846,
+      "step": 3929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314091887536647e-05,
+      "loss": 0.3119,
+      "step": 3930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313656814777618e-05,
+      "loss": 0.2858,
+      "step": 3931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313221608982046e-05,
+      "loss": 0.3009,
+      "step": 3932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312786270156135e-05,
+      "loss": 0.2765,
+      "step": 3933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312350798306118e-05,
+      "loss": 0.2909,
+      "step": 3934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311915193438203e-05,
+      "loss": 0.2969,
+      "step": 3935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311479455558614e-05,
+      "loss": 0.306,
+      "step": 3936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311043584673582e-05,
+      "loss": 0.2905,
+      "step": 3937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931060758078933e-05,
+      "loss": 0.2988,
+      "step": 3938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310171443912083e-05,
+      "loss": 0.3468,
+      "step": 3939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309735174048074e-05,
+      "loss": 0.3282,
+      "step": 3940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930929877120353e-05,
+      "loss": 0.336,
+      "step": 3941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930886223538469e-05,
+      "loss": 0.287,
+      "step": 3942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930842556659779e-05,
+      "loss": 0.3037,
+      "step": 3943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930798876484906e-05,
+      "loss": 0.3404,
+      "step": 3944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307551830144747e-05,
+      "loss": 0.3069,
+      "step": 3945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930711476249109e-05,
+      "loss": 0.2537,
+      "step": 3946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306677561894328e-05,
+      "loss": 0.3179,
+      "step": 3947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306240228360714e-05,
+      "loss": 0.3019,
+      "step": 3948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305802761896486e-05,
+      "loss": 0.3018,
+      "step": 3949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93053651625079e-05,
+      "loss": 0.2847,
+      "step": 3950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304927430201203e-05,
+      "loss": 0.2739,
+      "step": 3951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304489564982646e-05,
+      "loss": 0.3084,
+      "step": 3952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304051566858487e-05,
+      "loss": 0.3308,
+      "step": 3953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303613435834984e-05,
+      "loss": 0.2962,
+      "step": 3954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303175171918388e-05,
+      "loss": 0.2541,
+      "step": 3955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302736775114967e-05,
+      "loss": 0.321,
+      "step": 3956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302298245430976e-05,
+      "loss": 0.3325,
+      "step": 3957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301859582872685e-05,
+      "loss": 0.3076,
+      "step": 3958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930142078744636e-05,
+      "loss": 0.2755,
+      "step": 3959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300981859158266e-05,
+      "loss": 0.281,
+      "step": 3960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930054279801467e-05,
+      "loss": 0.3105,
+      "step": 3961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300103604021848e-05,
+      "loss": 0.267,
+      "step": 3962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299664277186073e-05,
+      "loss": 0.2686,
+      "step": 3963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299224817513614e-05,
+      "loss": 0.3233,
+      "step": 3964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929878522501076e-05,
+      "loss": 0.3015,
+      "step": 3965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929834549968378e-05,
+      "loss": 0.3232,
+      "step": 3966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929790564153896e-05,
+      "loss": 0.3362,
+      "step": 3967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929746565058258e-05,
+      "loss": 0.2784,
+      "step": 3968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929702552682093e-05,
+      "loss": 0.3128,
+      "step": 3969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296585270260292e-05,
+      "loss": 0.3032,
+      "step": 3970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296144880906954e-05,
+      "loss": 0.3184,
+      "step": 3971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929570435876721e-05,
+      "loss": 0.288,
+      "step": 3972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929526370384735e-05,
+      "loss": 0.2781,
+      "step": 3973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9294822916153672e-05,
+      "loss": 0.2927,
+      "step": 3974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929438199569247e-05,
+      "loss": 0.3018,
+      "step": 3975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293940942470038e-05,
+      "loss": 0.2889,
+      "step": 3976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929349975649268e-05,
+      "loss": 0.3258,
+      "step": 3977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293058437766697e-05,
+      "loss": 0.3034,
+      "step": 3978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292616986298397e-05,
+      "loss": 0.3033,
+      "step": 3979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292175402094078e-05,
+      "loss": 0.3004,
+      "step": 3980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291733685160056e-05,
+      "loss": 0.3376,
+      "step": 3981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929129183550263e-05,
+      "loss": 0.3241,
+      "step": 3982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290849853128124e-05,
+      "loss": 0.2973,
+      "step": 3983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290407738042843e-05,
+      "loss": 0.2949,
+      "step": 3984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289965490253103e-05,
+      "loss": 0.2777,
+      "step": 3985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289523109765223e-05,
+      "loss": 0.3001,
+      "step": 3986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928908059658552e-05,
+      "loss": 0.3009,
+      "step": 3987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288637950720315e-05,
+      "loss": 0.2664,
+      "step": 3988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288195172175932e-05,
+      "loss": 0.3254,
+      "step": 3989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287752260958695e-05,
+      "loss": 0.3364,
+      "step": 3990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287309217074936e-05,
+      "loss": 0.3033,
+      "step": 3991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286866040530974e-05,
+      "loss": 0.2959,
+      "step": 3992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928642273133314e-05,
+      "loss": 0.2864,
+      "step": 3993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285979289487778e-05,
+      "loss": 0.2893,
+      "step": 3994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928553571500121e-05,
+      "loss": 0.2901,
+      "step": 3995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285092007879776e-05,
+      "loss": 0.2971,
+      "step": 3996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284648168129812e-05,
+      "loss": 0.2735,
+      "step": 3997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284204195757664e-05,
+      "loss": 0.3133,
+      "step": 3998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928376009076967e-05,
+      "loss": 0.2991,
+      "step": 3999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928331585317217e-05,
+      "loss": 0.2856,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282871482971514e-05,
+      "loss": 0.2781,
+      "step": 4001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928242698017405e-05,
+      "loss": 0.3043,
+      "step": 4002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281982344786123e-05,
+      "loss": 0.2854,
+      "step": 4003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281537576814085e-05,
+      "loss": 0.2896,
+      "step": 4004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281092676264294e-05,
+      "loss": 0.2731,
+      "step": 4005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92806476431431e-05,
+      "loss": 0.2898,
+      "step": 4006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928020247745686e-05,
+      "loss": 0.3065,
+      "step": 4007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279757179211938e-05,
+      "loss": 0.2667,
+      "step": 4008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927931174841469e-05,
+      "loss": 0.2868,
+      "step": 4009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278866185071478e-05,
+      "loss": 0.3421,
+      "step": 4010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927842048918867e-05,
+      "loss": 0.308,
+      "step": 4011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277974660772628e-05,
+      "loss": 0.3041,
+      "step": 4012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927752869982972e-05,
+      "loss": 0.2689,
+      "step": 4013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277082606366323e-05,
+      "loss": 0.2939,
+      "step": 4014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276636380388805e-05,
+      "loss": 0.3167,
+      "step": 4015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927619002190354e-05,
+      "loss": 0.3379,
+      "step": 4016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275743530916897e-05,
+      "loss": 0.2737,
+      "step": 4017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275296907435264e-05,
+      "loss": 0.3105,
+      "step": 4018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9274850151465016e-05,
+      "loss": 0.268,
+      "step": 4019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274403263012537e-05,
+      "loss": 0.3124,
+      "step": 4020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273956242084207e-05,
+      "loss": 0.2778,
+      "step": 4021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273509088686412e-05,
+      "loss": 0.2959,
+      "step": 4022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273061802825542e-05,
+      "loss": 0.314,
+      "step": 4023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927261438450798e-05,
+      "loss": 0.2859,
+      "step": 4024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272166833740126e-05,
+      "loss": 0.2795,
+      "step": 4025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271719150528362e-05,
+      "loss": 0.3086,
+      "step": 4026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271271334879094e-05,
+      "loss": 0.2705,
+      "step": 4027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270823386798708e-05,
+      "loss": 0.2798,
+      "step": 4028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270375306293608e-05,
+      "loss": 0.3088,
+      "step": 4029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269927093370197e-05,
+      "loss": 0.2985,
+      "step": 4030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269478748034873e-05,
+      "loss": 0.2819,
+      "step": 4031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926903027029404e-05,
+      "loss": 0.2947,
+      "step": 4032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268581660154106e-05,
+      "loss": 0.2882,
+      "step": 4033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268132917621482e-05,
+      "loss": 0.3102,
+      "step": 4034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926768404270257e-05,
+      "loss": 0.3266,
+      "step": 4035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267235035403785e-05,
+      "loss": 0.3096,
+      "step": 4036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266785895731543e-05,
+      "loss": 0.2718,
+      "step": 4037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926633662369226e-05,
+      "loss": 0.3382,
+      "step": 4038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926588721929235e-05,
+      "loss": 0.2956,
+      "step": 4039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265437682538236e-05,
+      "loss": 0.2904,
+      "step": 4040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926498801343633e-05,
+      "loss": 0.3047,
+      "step": 4041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926453821199307e-05,
+      "loss": 0.2766,
+      "step": 4042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926408827821487e-05,
+      "loss": 0.318,
+      "step": 4043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926363821210816e-05,
+      "loss": 0.2973,
+      "step": 4044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263188013679368e-05,
+      "loss": 0.3274,
+      "step": 4045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262737682934926e-05,
+      "loss": 0.2862,
+      "step": 4046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262287219881267e-05,
+      "loss": 0.2801,
+      "step": 4047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261836624524823e-05,
+      "loss": 0.2979,
+      "step": 4048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926138589687203e-05,
+      "loss": 0.2709,
+      "step": 4049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926093503692933e-05,
+      "loss": 0.279,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260484044703163e-05,
+      "loss": 0.2639,
+      "step": 4051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260032920199968e-05,
+      "loss": 0.323,
+      "step": 4052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259581663426185e-05,
+      "loss": 0.2847,
+      "step": 4053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925913027438827e-05,
+      "loss": 0.3259,
+      "step": 4054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258678753092662e-05,
+      "loss": 0.2948,
+      "step": 4055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258227099545818e-05,
+      "loss": 0.3023,
+      "step": 4056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257775313754182e-05,
+      "loss": 0.2958,
+      "step": 4057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925732339572421e-05,
+      "loss": 0.3128,
+      "step": 4058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256871345462356e-05,
+      "loss": 0.3011,
+      "step": 4059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256419162975085e-05,
+      "loss": 0.2735,
+      "step": 4060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255966848268847e-05,
+      "loss": 0.2904,
+      "step": 4061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255514401350103e-05,
+      "loss": 0.3653,
+      "step": 4062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255061822225322e-05,
+      "loss": 0.2603,
+      "step": 4063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925460911090096e-05,
+      "loss": 0.3195,
+      "step": 4064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254156267383496e-05,
+      "loss": 0.309,
+      "step": 4065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253703291679383e-05,
+      "loss": 0.301,
+      "step": 4066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253250183795105e-05,
+      "loss": 0.3467,
+      "step": 4067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925279694373713e-05,
+      "loss": 0.3274,
+      "step": 4068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252343571511926e-05,
+      "loss": 0.2878,
+      "step": 4069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925189006712598e-05,
+      "loss": 0.3159,
+      "step": 4070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251436430585758e-05,
+      "loss": 0.2979,
+      "step": 4071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250982661897745e-05,
+      "loss": 0.2924,
+      "step": 4072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250528761068422e-05,
+      "loss": 0.301,
+      "step": 4073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250074728104276e-05,
+      "loss": 0.2766,
+      "step": 4074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249620563011793e-05,
+      "loss": 0.3094,
+      "step": 4075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924916626579745e-05,
+      "loss": 0.2797,
+      "step": 4076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924871183646775e-05,
+      "loss": 0.308,
+      "step": 4077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248257275029174e-05,
+      "loss": 0.2941,
+      "step": 4078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924780258148822e-05,
+      "loss": 0.3003,
+      "step": 4079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924734775585138e-05,
+      "loss": 0.3063,
+      "step": 4080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246892798125155e-05,
+      "loss": 0.2984,
+      "step": 4081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924643770831604e-05,
+      "loss": 0.297,
+      "step": 4082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245982486430536e-05,
+      "loss": 0.2792,
+      "step": 4083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245527132475144e-05,
+      "loss": 0.3063,
+      "step": 4084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924507164645637e-05,
+      "loss": 0.3041,
+      "step": 4085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924461602838072e-05,
+      "loss": 0.2979,
+      "step": 4086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244160278254702e-05,
+      "loss": 0.2768,
+      "step": 4087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243704396084828e-05,
+      "loss": 0.2652,
+      "step": 4088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243248381877605e-05,
+      "loss": 0.2816,
+      "step": 4089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242792235639554e-05,
+      "loss": 0.335,
+      "step": 4090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924233595737718e-05,
+      "loss": 0.2889,
+      "step": 4091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924187954709701e-05,
+      "loss": 0.2637,
+      "step": 4092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241423004805565e-05,
+      "loss": 0.2664,
+      "step": 4093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240966330509354e-05,
+      "loss": 0.2815,
+      "step": 4094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924050952421491e-05,
+      "loss": 0.3032,
+      "step": 4095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240052585928757e-05,
+      "loss": 0.2862,
+      "step": 4096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923959551565742e-05,
+      "loss": 0.2905,
+      "step": 4097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923913831340743e-05,
+      "loss": 0.3125,
+      "step": 4098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238680979185308e-05,
+      "loss": 0.3047,
+      "step": 4099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92382235129976e-05,
+      "loss": 0.2732,
+      "step": 4100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923776591485084e-05,
+      "loss": 0.3051,
+      "step": 4101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923730818475155e-05,
+      "loss": 0.2908,
+      "step": 4102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236850322706284e-05,
+      "loss": 0.2828,
+      "step": 4103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236392328721577e-05,
+      "loss": 0.3064,
+      "step": 4104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235934202803967e-05,
+      "loss": 0.2787,
+      "step": 4105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923547594496e-05,
+      "loss": 0.2836,
+      "step": 4106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923501755519622e-05,
+      "loss": 0.285,
+      "step": 4107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923455903351918e-05,
+      "loss": 0.307,
+      "step": 4108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923410037993543e-05,
+      "loss": 0.3186,
+      "step": 4109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9233641594451514e-05,
+      "loss": 0.2953,
+      "step": 4110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233182677073985e-05,
+      "loss": 0.305,
+      "step": 4111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923272362780941e-05,
+      "loss": 0.2847,
+      "step": 4112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232264446664336e-05,
+      "loss": 0.2899,
+      "step": 4113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231805133645324e-05,
+      "loss": 0.3095,
+      "step": 4114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231345688758932e-05,
+      "loss": 0.2928,
+      "step": 4115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230886112011733e-05,
+      "loss": 0.2866,
+      "step": 4116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923042640341028e-05,
+      "loss": 0.2911,
+      "step": 4117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229966562961144e-05,
+      "loss": 0.2785,
+      "step": 4118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229506590670897e-05,
+      "loss": 0.2929,
+      "step": 4119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229046486546104e-05,
+      "loss": 0.2942,
+      "step": 4120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228586250593338e-05,
+      "loss": 0.2709,
+      "step": 4121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228125882819176e-05,
+      "loss": 0.3177,
+      "step": 4122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922766538323019e-05,
+      "loss": 0.3052,
+      "step": 4123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227204751832958e-05,
+      "loss": 0.305,
+      "step": 4124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226743988634065e-05,
+      "loss": 0.2967,
+      "step": 4125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226283093640088e-05,
+      "loss": 0.2859,
+      "step": 4126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922582206685761e-05,
+      "loss": 0.3183,
+      "step": 4127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225360908293217e-05,
+      "loss": 0.3231,
+      "step": 4128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92248996179535e-05,
+      "loss": 0.3064,
+      "step": 4129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224438195845045e-05,
+      "loss": 0.2901,
+      "step": 4130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223976641974438e-05,
+      "loss": 0.3007,
+      "step": 4131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922351495634828e-05,
+      "loss": 0.2455,
+      "step": 4132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223053138973164e-05,
+      "loss": 0.3024,
+      "step": 4133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222591189855683e-05,
+      "loss": 0.3174,
+      "step": 4134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222129109002437e-05,
+      "loss": 0.2619,
+      "step": 4135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221666896420026e-05,
+      "loss": 0.322,
+      "step": 4136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221204552115054e-05,
+      "loss": 0.2795,
+      "step": 4137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922074207609413e-05,
+      "loss": 0.2978,
+      "step": 4138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220279468363848e-05,
+      "loss": 0.2766,
+      "step": 4139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219816728930824e-05,
+      "loss": 0.2889,
+      "step": 4140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219353857801666e-05,
+      "loss": 0.2932,
+      "step": 4141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218890854982984e-05,
+      "loss": 0.2942,
+      "step": 4142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218427720481396e-05,
+      "loss": 0.2976,
+      "step": 4143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217964454303515e-05,
+      "loss": 0.307,
+      "step": 4144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921750105645596e-05,
+      "loss": 0.2899,
+      "step": 4145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217037526945343e-05,
+      "loss": 0.2642,
+      "step": 4146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216573865778292e-05,
+      "loss": 0.2992,
+      "step": 4147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921611007296143e-05,
+      "loss": 0.2918,
+      "step": 4148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215646148501378e-05,
+      "loss": 0.2994,
+      "step": 4149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215182092404765e-05,
+      "loss": 0.2949,
+      "step": 4150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921471790467822e-05,
+      "loss": 0.2949,
+      "step": 4151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921425358532837e-05,
+      "loss": 0.3103,
+      "step": 4152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213789134361854e-05,
+      "loss": 0.2841,
+      "step": 4153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213324551785302e-05,
+      "loss": 0.3047,
+      "step": 4154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921285983760535e-05,
+      "loss": 0.2852,
+      "step": 4155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212394991828635e-05,
+      "loss": 0.3209,
+      "step": 4156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92119300144618e-05,
+      "loss": 0.3071,
+      "step": 4157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211464905511484e-05,
+      "loss": 0.2871,
+      "step": 4158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9210999664984334e-05,
+      "loss": 0.3272,
+      "step": 4159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921053429288699e-05,
+      "loss": 0.2928,
+      "step": 4160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9210068789226106e-05,
+      "loss": 0.3255,
+      "step": 4161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209603154008328e-05,
+      "loss": 0.2644,
+      "step": 4162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209137387240306e-05,
+      "loss": 0.2988,
+      "step": 4163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208671488928693e-05,
+      "loss": 0.3017,
+      "step": 4164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208205459080145e-05,
+      "loss": 0.2669,
+      "step": 4165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920773929770132e-05,
+      "loss": 0.3043,
+      "step": 4166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207273004798873e-05,
+      "loss": 0.2709,
+      "step": 4167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206806580379468e-05,
+      "loss": 0.2932,
+      "step": 4168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920634002444977e-05,
+      "loss": 0.3051,
+      "step": 4169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205873337016437e-05,
+      "loss": 0.2647,
+      "step": 4170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205406518086137e-05,
+      "loss": 0.2832,
+      "step": 4171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920493956766554e-05,
+      "loss": 0.2669,
+      "step": 4172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204472485761315e-05,
+      "loss": 0.3001,
+      "step": 4173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204005272380132e-05,
+      "loss": 0.2948,
+      "step": 4174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920353792752867e-05,
+      "loss": 0.2951,
+      "step": 4175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203070451213598e-05,
+      "loss": 0.3281,
+      "step": 4176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92026028434416e-05,
+      "loss": 0.3444,
+      "step": 4177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202135104219347e-05,
+      "loss": 0.3495,
+      "step": 4178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201667233553525e-05,
+      "loss": 0.2941,
+      "step": 4179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920119923145082e-05,
+      "loss": 0.3357,
+      "step": 4180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200731097917912e-05,
+      "loss": 0.2959,
+      "step": 4181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920026283296149e-05,
+      "loss": 0.3223,
+      "step": 4182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 0.3036,
+      "step": 4183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199325908804864e-05,
+      "loss": 0.2771,
+      "step": 4184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198857249618037e-05,
+      "loss": 0.3143,
+      "step": 4185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198388459034465e-05,
+      "loss": 0.2811,
+      "step": 4186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919791953706084e-05,
+      "loss": 0.2707,
+      "step": 4187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197450483703864e-05,
+      "loss": 0.3294,
+      "step": 4188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919698129897023e-05,
+      "loss": 0.2802,
+      "step": 4189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196511982866646e-05,
+      "loss": 0.2657,
+      "step": 4190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196042535399814e-05,
+      "loss": 0.2944,
+      "step": 4191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195572956576442e-05,
+      "loss": 0.2991,
+      "step": 4192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919510324640323e-05,
+      "loss": 0.3337,
+      "step": 4193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194633404886893e-05,
+      "loss": 0.2858,
+      "step": 4194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194163432034144e-05,
+      "loss": 0.3238,
+      "step": 4195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919369332785169e-05,
+      "loss": 0.3139,
+      "step": 4196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193223092346253e-05,
+      "loss": 0.3054,
+      "step": 4197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192752725524545e-05,
+      "loss": 0.2835,
+      "step": 4198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192282227393286e-05,
+      "loss": 0.3027,
+      "step": 4199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9191811597959197e-05,
+      "loss": 0.3008,
+      "step": 4200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9191340837229e-05,
+      "loss": 0.304,
+      "step": 4201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190869945209417e-05,
+      "loss": 0.2823,
+      "step": 4202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919039892190718e-05,
+      "loss": 0.3057,
+      "step": 4203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189927767329015e-05,
+      "loss": 0.2933,
+      "step": 4204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189456481481648e-05,
+      "loss": 0.3094,
+      "step": 4205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188985064371818e-05,
+      "loss": 0.3154,
+      "step": 4206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188513516006247e-05,
+      "loss": 0.3257,
+      "step": 4207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188041836391685e-05,
+      "loss": 0.3074,
+      "step": 4208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918757002553486e-05,
+      "loss": 0.2869,
+      "step": 4209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187098083442512e-05,
+      "loss": 0.3035,
+      "step": 4210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186626010121388e-05,
+      "loss": 0.3179,
+      "step": 4211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186153805578224e-05,
+      "loss": 0.2811,
+      "step": 4212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918568146981977e-05,
+      "loss": 0.3059,
+      "step": 4213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918520900285277e-05,
+      "loss": 0.2887,
+      "step": 4214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184736404683975e-05,
+      "loss": 0.2659,
+      "step": 4215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184263675320135e-05,
+      "loss": 0.3052,
+      "step": 4216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183790814768e-05,
+      "loss": 0.2978,
+      "step": 4217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183317823034325e-05,
+      "loss": 0.324,
+      "step": 4218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918284470012587e-05,
+      "loss": 0.2807,
+      "step": 4219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918237144604939e-05,
+      "loss": 0.3023,
+      "step": 4220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181898060811648e-05,
+      "loss": 0.2948,
+      "step": 4221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181424544419403e-05,
+      "loss": 0.2791,
+      "step": 4222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180950896879418e-05,
+      "loss": 0.3097,
+      "step": 4223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180477118198458e-05,
+      "loss": 0.3187,
+      "step": 4224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180003208383293e-05,
+      "loss": 0.2948,
+      "step": 4225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179529167440695e-05,
+      "loss": 0.2785,
+      "step": 4226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179054995377427e-05,
+      "loss": 0.2873,
+      "step": 4227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917858069220027e-05,
+      "loss": 0.3045,
+      "step": 4228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178106257915995e-05,
+      "loss": 0.2823,
+      "step": 4229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917763169253138e-05,
+      "loss": 0.2869,
+      "step": 4230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177156996053198e-05,
+      "loss": 0.3135,
+      "step": 4231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917668216848824e-05,
+      "loss": 0.2843,
+      "step": 4232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176207209843282e-05,
+      "loss": 0.2928,
+      "step": 4233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175732120125112e-05,
+      "loss": 0.3076,
+      "step": 4234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175256899340514e-05,
+      "loss": 0.2813,
+      "step": 4235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174781547496276e-05,
+      "loss": 0.2821,
+      "step": 4236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174306064599185e-05,
+      "loss": 0.2964,
+      "step": 4237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173830450656037e-05,
+      "loss": 0.2972,
+      "step": 4238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173354705673627e-05,
+      "loss": 0.279,
+      "step": 4239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172878829658743e-05,
+      "loss": 0.3192,
+      "step": 4240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917240282261819e-05,
+      "loss": 0.292,
+      "step": 4241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171926684558766e-05,
+      "loss": 0.2825,
+      "step": 4242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917145041548727e-05,
+      "loss": 0.3022,
+      "step": 4243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170974015410503e-05,
+      "loss": 0.3191,
+      "step": 4244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170497484335276e-05,
+      "loss": 0.2908,
+      "step": 4245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170020822268393e-05,
+      "loss": 0.2963,
+      "step": 4246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169544029216664e-05,
+      "loss": 0.3221,
+      "step": 4247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169067105186895e-05,
+      "loss": 0.2801,
+      "step": 4248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91685900501859e-05,
+      "loss": 0.3193,
+      "step": 4249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168112864220492e-05,
+      "loss": 0.3262,
+      "step": 4250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167635547297496e-05,
+      "loss": 0.2915,
+      "step": 4251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916715809942372e-05,
+      "loss": 0.2787,
+      "step": 4252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916668052060599e-05,
+      "loss": 0.3375,
+      "step": 4253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166202810851123e-05,
+      "loss": 0.2993,
+      "step": 4254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165724970165948e-05,
+      "loss": 0.3133,
+      "step": 4255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165246998557285e-05,
+      "loss": 0.2887,
+      "step": 4256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164768896031962e-05,
+      "loss": 0.3021,
+      "step": 4257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164290662596813e-05,
+      "loss": 0.3024,
+      "step": 4258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163812298258665e-05,
+      "loss": 0.2893,
+      "step": 4259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163333803024348e-05,
+      "loss": 0.3023,
+      "step": 4260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162855176900706e-05,
+      "loss": 0.3155,
+      "step": 4261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916237641989457e-05,
+      "loss": 0.3009,
+      "step": 4262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161897532012777e-05,
+      "loss": 0.2717,
+      "step": 4263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161418513262174e-05,
+      "loss": 0.3093,
+      "step": 4264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160939363649592e-05,
+      "loss": 0.2876,
+      "step": 4265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160460083181886e-05,
+      "loss": 0.3114,
+      "step": 4266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159980671865898e-05,
+      "loss": 0.3035,
+      "step": 4267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159501129708477e-05,
+      "loss": 0.2765,
+      "step": 4268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915902145671647e-05,
+      "loss": 0.2847,
+      "step": 4269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915854165289673e-05,
+      "loss": 0.3122,
+      "step": 4270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158061718256115e-05,
+      "loss": 0.2731,
+      "step": 4271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157581652801475e-05,
+      "loss": 0.2877,
+      "step": 4272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157101456539665e-05,
+      "loss": 0.2912,
+      "step": 4273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915662112947755e-05,
+      "loss": 0.305,
+      "step": 4274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156140671621992e-05,
+      "loss": 0.2867,
+      "step": 4275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155660082979848e-05,
+      "loss": 0.2897,
+      "step": 4276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155179363557984e-05,
+      "loss": 0.2785,
+      "step": 4277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154698513363274e-05,
+      "loss": 0.2965,
+      "step": 4278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154217532402573e-05,
+      "loss": 0.294,
+      "step": 4279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153736420682763e-05,
+      "loss": 0.2858,
+      "step": 4280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153255178210713e-05,
+      "loss": 0.3054,
+      "step": 4281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152773804993293e-05,
+      "loss": 0.2841,
+      "step": 4282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152292301037384e-05,
+      "loss": 0.3339,
+      "step": 4283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915181066634986e-05,
+      "loss": 0.2939,
+      "step": 4284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151328900937606e-05,
+      "loss": 0.2947,
+      "step": 4285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150847004807496e-05,
+      "loss": 0.2811,
+      "step": 4286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150364977966423e-05,
+      "loss": 0.3122,
+      "step": 4287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149882820421264e-05,
+      "loss": 0.2726,
+      "step": 4288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914940053217891e-05,
+      "loss": 0.2936,
+      "step": 4289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914891811324625e-05,
+      "loss": 0.2741,
+      "step": 4290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914843556363017e-05,
+      "loss": 0.292,
+      "step": 4291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147952883337568e-05,
+      "loss": 0.3193,
+      "step": 4292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914747007237534e-05,
+      "loss": 0.3252,
+      "step": 4293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914698713075038e-05,
+      "loss": 0.3083,
+      "step": 4294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146504058469585e-05,
+      "loss": 0.2943,
+      "step": 4295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146020855539852e-05,
+      "loss": 0.3157,
+      "step": 4296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145537521968095e-05,
+      "loss": 0.3054,
+      "step": 4297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145054057761204e-05,
+      "loss": 0.2913,
+      "step": 4298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144570462926095e-05,
+      "loss": 0.2988,
+      "step": 4299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144086737469673e-05,
+      "loss": 0.3194,
+      "step": 4300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143602881398846e-05,
+      "loss": 0.3061,
+      "step": 4301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914311889472052e-05,
+      "loss": 0.2619,
+      "step": 4302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142634777441623e-05,
+      "loss": 0.286,
+      "step": 4303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142150529569055e-05,
+      "loss": 0.2726,
+      "step": 4304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914166615110974e-05,
+      "loss": 0.3136,
+      "step": 4305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91411816420706e-05,
+      "loss": 0.2715,
+      "step": 4306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140697002458547e-05,
+      "loss": 0.281,
+      "step": 4307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140212232280513e-05,
+      "loss": 0.2962,
+      "step": 4308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139727331543417e-05,
+      "loss": 0.3168,
+      "step": 4309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139242300254185e-05,
+      "loss": 0.2858,
+      "step": 4310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138757138419745e-05,
+      "loss": 0.3121,
+      "step": 4311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913827184604703e-05,
+      "loss": 0.3206,
+      "step": 4312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137786423142973e-05,
+      "loss": 0.2847,
+      "step": 4313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137300869714505e-05,
+      "loss": 0.3208,
+      "step": 4314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136815185768563e-05,
+      "loss": 0.2999,
+      "step": 4315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136329371312082e-05,
+      "loss": 0.2995,
+      "step": 4316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135843426352e-05,
+      "loss": 0.2982,
+      "step": 4317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135357350895262e-05,
+      "loss": 0.2868,
+      "step": 4318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913487114494881e-05,
+      "loss": 0.2717,
+      "step": 4319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134384808519596e-05,
+      "loss": 0.2863,
+      "step": 4320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133898341614555e-05,
+      "loss": 0.2879,
+      "step": 4321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133411744240643e-05,
+      "loss": 0.2863,
+      "step": 4322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132925016404805e-05,
+      "loss": 0.2779,
+      "step": 4323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132438158113998e-05,
+      "loss": 0.2965,
+      "step": 4324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131951169375176e-05,
+      "loss": 0.2817,
+      "step": 4325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131464050195294e-05,
+      "loss": 0.3006,
+      "step": 4326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913097680058131e-05,
+      "loss": 0.3098,
+      "step": 4327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130489420540185e-05,
+      "loss": 0.2858,
+      "step": 4328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130001910078878e-05,
+      "loss": 0.306,
+      "step": 4329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129514269204356e-05,
+      "loss": 0.3132,
+      "step": 4330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129026497923585e-05,
+      "loss": 0.2828,
+      "step": 4331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912853859624353e-05,
+      "loss": 0.2742,
+      "step": 4332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128050564171157e-05,
+      "loss": 0.31,
+      "step": 4333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912756240171344e-05,
+      "loss": 0.2989,
+      "step": 4334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912707410887736e-05,
+      "loss": 0.3179,
+      "step": 4335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126585685669877e-05,
+      "loss": 0.2746,
+      "step": 4336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126097132097978e-05,
+      "loss": 0.2851,
+      "step": 4337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912560844816864e-05,
+      "loss": 0.2837,
+      "step": 4338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912511963388884e-05,
+      "loss": 0.2858,
+      "step": 4339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912463068926556e-05,
+      "loss": 0.2851,
+      "step": 4340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912414161430579e-05,
+      "loss": 0.2912,
+      "step": 4341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912365240901651e-05,
+      "loss": 0.2908,
+      "step": 4342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123163073404714e-05,
+      "loss": 0.2796,
+      "step": 4343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122673607477385e-05,
+      "loss": 0.2849,
+      "step": 4344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122184011241517e-05,
+      "loss": 0.3122,
+      "step": 4345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121694284704105e-05,
+      "loss": 0.2923,
+      "step": 4346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912120442787214e-05,
+      "loss": 0.3344,
+      "step": 4347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120714440752625e-05,
+      "loss": 0.3145,
+      "step": 4348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120224323352553e-05,
+      "loss": 0.3233,
+      "step": 4349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911973407567893e-05,
+      "loss": 0.288,
+      "step": 4350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119243697738758e-05,
+      "loss": 0.2975,
+      "step": 4351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118753189539037e-05,
+      "loss": 0.3026,
+      "step": 4352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911826255108678e-05,
+      "loss": 0.2974,
+      "step": 4353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911777178238899e-05,
+      "loss": 0.2804,
+      "step": 4354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117280883452677e-05,
+      "loss": 0.3139,
+      "step": 4355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911678985428486e-05,
+      "loss": 0.2934,
+      "step": 4356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116298694892545e-05,
+      "loss": 0.2903,
+      "step": 4357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911580740528275e-05,
+      "loss": 0.3034,
+      "step": 4358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115315985462496e-05,
+      "loss": 0.2834,
+      "step": 4359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114824435438795e-05,
+      "loss": 0.2592,
+      "step": 4360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114332755218677e-05,
+      "loss": 0.3201,
+      "step": 4361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911384094480916e-05,
+      "loss": 0.2793,
+      "step": 4362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911334900421727e-05,
+      "loss": 0.301,
+      "step": 4363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112856933450038e-05,
+      "loss": 0.327,
+      "step": 4364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112364732514484e-05,
+      "loss": 0.3348,
+      "step": 4365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111872401417647e-05,
+      "loss": 0.2861,
+      "step": 4366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111379940166555e-05,
+      "loss": 0.3027,
+      "step": 4367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110887348768243e-05,
+      "loss": 0.2941,
+      "step": 4368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911039462722975e-05,
+      "loss": 0.3167,
+      "step": 4369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109901775558106e-05,
+      "loss": 0.3092,
+      "step": 4370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910940879376036e-05,
+      "loss": 0.3192,
+      "step": 4371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108915681843552e-05,
+      "loss": 0.3062,
+      "step": 4372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910842243981472e-05,
+      "loss": 0.2943,
+      "step": 4373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107929067680917e-05,
+      "loss": 0.3203,
+      "step": 4374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107435565449185e-05,
+      "loss": 0.3037,
+      "step": 4375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106941933126576e-05,
+      "loss": 0.2992,
+      "step": 4376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106448170720136e-05,
+      "loss": 0.3107,
+      "step": 4377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105954278236924e-05,
+      "loss": 0.2869,
+      "step": 4378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105460255683992e-05,
+      "loss": 0.2826,
+      "step": 4379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91049661030684e-05,
+      "loss": 0.3089,
+      "step": 4380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9104471820397204e-05,
+      "loss": 0.286,
+      "step": 4381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910397740767746e-05,
+      "loss": 0.31,
+      "step": 4382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103482864916237e-05,
+      "loss": 0.2741,
+      "step": 4383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102988192120598e-05,
+      "loss": 0.302,
+      "step": 4384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102493389297606e-05,
+      "loss": 0.3027,
+      "step": 4385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101998456454328e-05,
+      "loss": 0.2895,
+      "step": 4386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910150339359784e-05,
+      "loss": 0.3085,
+      "step": 4387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910100820073521e-05,
+      "loss": 0.3122,
+      "step": 4388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100512877873502e-05,
+      "loss": 0.2813,
+      "step": 4389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100017425019808e-05,
+      "loss": 0.276,
+      "step": 4390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099521842181193e-05,
+      "loss": 0.307,
+      "step": 4391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099026129364742e-05,
+      "loss": 0.3404,
+      "step": 4392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098530286577533e-05,
+      "loss": 0.2786,
+      "step": 4393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098034313826648e-05,
+      "loss": 0.2767,
+      "step": 4394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909753821111917e-05,
+      "loss": 0.3058,
+      "step": 4395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097041978462194e-05,
+      "loss": 0.2912,
+      "step": 4396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90965456158628e-05,
+      "loss": 0.3025,
+      "step": 4397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909604912332808e-05,
+      "loss": 0.2906,
+      "step": 4398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095552500865126e-05,
+      "loss": 0.2627,
+      "step": 4399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095055748481033e-05,
+      "loss": 0.3069,
+      "step": 4400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094558866182892e-05,
+      "loss": 0.301,
+      "step": 4401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909406185397781e-05,
+      "loss": 0.3072,
+      "step": 4402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093564711872873e-05,
+      "loss": 0.2811,
+      "step": 4403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093067439875195e-05,
+      "loss": 0.3243,
+      "step": 4404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092570037991873e-05,
+      "loss": 0.2907,
+      "step": 4405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909207250623001e-05,
+      "loss": 0.3032,
+      "step": 4406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909157484459672e-05,
+      "loss": 0.3087,
+      "step": 4407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091077053099104e-05,
+      "loss": 0.282,
+      "step": 4408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090579131744275e-05,
+      "loss": 0.2938,
+      "step": 4409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090081080539345e-05,
+      "loss": 0.2812,
+      "step": 4410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908958289949143e-05,
+      "loss": 0.2914,
+      "step": 4411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089084588607643e-05,
+      "loss": 0.2995,
+      "step": 4412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908858614789511e-05,
+      "loss": 0.3041,
+      "step": 4413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908808757736094e-05,
+      "loss": 0.3342,
+      "step": 4414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087588877012256e-05,
+      "loss": 0.2992,
+      "step": 4415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908709004685619e-05,
+      "loss": 0.2845,
+      "step": 4416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908659108689986e-05,
+      "loss": 0.3005,
+      "step": 4417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086091997150396e-05,
+      "loss": 0.3089,
+      "step": 4418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085592777614925e-05,
+      "loss": 0.2797,
+      "step": 4419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908509342830058e-05,
+      "loss": 0.3054,
+      "step": 4420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908459394921449e-05,
+      "loss": 0.3031,
+      "step": 4421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084094340363794e-05,
+      "loss": 0.282,
+      "step": 4422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908359460175563e-05,
+      "loss": 0.3,
+      "step": 4423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908309473339713e-05,
+      "loss": 0.2983,
+      "step": 4424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082594735295433e-05,
+      "loss": 0.3011,
+      "step": 4425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908209460745769e-05,
+      "loss": 0.2817,
+      "step": 4426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908159434989104e-05,
+      "loss": 0.3009,
+      "step": 4427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081093962602628e-05,
+      "loss": 0.305,
+      "step": 4428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90805934455996e-05,
+      "loss": 0.2989,
+      "step": 4429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908009279888911e-05,
+      "loss": 0.2946,
+      "step": 4430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079592022478307e-05,
+      "loss": 0.3099,
+      "step": 4431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907909111637434e-05,
+      "loss": 0.3022,
+      "step": 4432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907859008058437e-05,
+      "loss": 0.3026,
+      "step": 4433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078088915115555e-05,
+      "loss": 0.2812,
+      "step": 4434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077587619975045e-05,
+      "loss": 0.3032,
+      "step": 4435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907708619517001e-05,
+      "loss": 0.2746,
+      "step": 4436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076584640707606e-05,
+      "loss": 0.3315,
+      "step": 4437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076082956595e-05,
+      "loss": 0.2825,
+      "step": 4438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907558114283936e-05,
+      "loss": 0.3089,
+      "step": 4439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907507919944785e-05,
+      "loss": 0.2884,
+      "step": 4440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907457712642764e-05,
+      "loss": 0.3034,
+      "step": 4441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074074923785902e-05,
+      "loss": 0.2838,
+      "step": 4442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073572591529812e-05,
+      "loss": 0.2815,
+      "step": 4443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073070129666545e-05,
+      "loss": 0.2963,
+      "step": 4444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072567538203272e-05,
+      "loss": 0.2761,
+      "step": 4445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907206481714718e-05,
+      "loss": 0.2874,
+      "step": 4446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907156196650545e-05,
+      "loss": 0.2932,
+      "step": 4447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071058986285255e-05,
+      "loss": 0.3359,
+      "step": 4448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070555876493787e-05,
+      "loss": 0.3179,
+      "step": 4449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070052637138238e-05,
+      "loss": 0.3088,
+      "step": 4450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906954926822578e-05,
+      "loss": 0.2779,
+      "step": 4451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906904576976362e-05,
+      "loss": 0.2969,
+      "step": 4452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9068542141758943e-05,
+      "loss": 0.2789,
+      "step": 4453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906803838421894e-05,
+      "loss": 0.2723,
+      "step": 4454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906753449715081e-05,
+      "loss": 0.2946,
+      "step": 4455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906703048056175e-05,
+      "loss": 0.2828,
+      "step": 4456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906652633445896e-05,
+      "loss": 0.2949,
+      "step": 4457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066022058849644e-05,
+      "loss": 0.288,
+      "step": 4458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065517653740993e-05,
+      "loss": 0.3113,
+      "step": 4459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906501311914023e-05,
+      "loss": 0.28,
+      "step": 4460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064508455054548e-05,
+      "loss": 0.2815,
+      "step": 4461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064003661491157e-05,
+      "loss": 0.336,
+      "step": 4462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063498738457275e-05,
+      "loss": 0.3153,
+      "step": 4463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062993685960106e-05,
+      "loss": 0.3057,
+      "step": 4464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906248850400687e-05,
+      "loss": 0.295,
+      "step": 4465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061983192604784e-05,
+      "loss": 0.3047,
+      "step": 4466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906147775176106e-05,
+      "loss": 0.3063,
+      "step": 4467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906097218148292e-05,
+      "loss": 0.2862,
+      "step": 4468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060466481777587e-05,
+      "loss": 0.3007,
+      "step": 4469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059960652652284e-05,
+      "loss": 0.3009,
+      "step": 4470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059454694114237e-05,
+      "loss": 0.2976,
+      "step": 4471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905894860617067e-05,
+      "loss": 0.3033,
+      "step": 4472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058442388828816e-05,
+      "loss": 0.3017,
+      "step": 4473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057936042095902e-05,
+      "loss": 0.2888,
+      "step": 4474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057429565979166e-05,
+      "loss": 0.2541,
+      "step": 4475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056922960485835e-05,
+      "loss": 0.277,
+      "step": 4476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056416225623153e-05,
+      "loss": 0.3004,
+      "step": 4477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055909361398353e-05,
+      "loss": 0.2973,
+      "step": 4478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055402367818673e-05,
+      "loss": 0.2863,
+      "step": 4479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054895244891366e-05,
+      "loss": 0.31,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054387992623663e-05,
+      "loss": 0.3337,
+      "step": 4481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053880611022818e-05,
+      "loss": 0.3096,
+      "step": 4482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053373100096075e-05,
+      "loss": 0.2597,
+      "step": 4483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052865459850683e-05,
+      "loss": 0.3002,
+      "step": 4484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905235769029389e-05,
+      "loss": 0.2915,
+      "step": 4485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905184979143296e-05,
+      "loss": 0.3354,
+      "step": 4486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051341763275135e-05,
+      "loss": 0.3097,
+      "step": 4487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050833605827682e-05,
+      "loss": 0.2763,
+      "step": 4488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905032531909785e-05,
+      "loss": 0.3089,
+      "step": 4489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904981690309291e-05,
+      "loss": 0.3012,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049308357820116e-05,
+      "loss": 0.2863,
+      "step": 4491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048799683286734e-05,
+      "loss": 0.342,
+      "step": 4492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904829087950003e-05,
+      "loss": 0.3368,
+      "step": 4493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047781946467275e-05,
+      "loss": 0.2837,
+      "step": 4494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047272884195735e-05,
+      "loss": 0.3345,
+      "step": 4495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904676369269268e-05,
+      "loss": 0.272,
+      "step": 4496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904625437196539e-05,
+      "loss": 0.2953,
+      "step": 4497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045744922021133e-05,
+      "loss": 0.3134,
+      "step": 4498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904523534286719e-05,
+      "loss": 0.2912,
+      "step": 4499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904472563451084e-05,
+      "loss": 0.324,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904421579695936e-05,
+      "loss": 0.3398,
+      "step": 4501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043705830220033e-05,
+      "loss": 0.2847,
+      "step": 4502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043195734300147e-05,
+      "loss": 0.2554,
+      "step": 4503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904268550920699e-05,
+      "loss": 0.3123,
+      "step": 4504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042175154947842e-05,
+      "loss": 0.288,
+      "step": 4505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904166467153e-05,
+      "loss": 0.2704,
+      "step": 4506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041154058960755e-05,
+      "loss": 0.31,
+      "step": 4507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040643317247394e-05,
+      "loss": 0.3024,
+      "step": 4508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040132446397224e-05,
+      "loss": 0.2837,
+      "step": 4509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039621446417528e-05,
+      "loss": 0.2444,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039110317315618e-05,
+      "loss": 0.3185,
+      "step": 4511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903859905909879e-05,
+      "loss": 0.2953,
+      "step": 4512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038087671774344e-05,
+      "loss": 0.2854,
+      "step": 4513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903757615534959e-05,
+      "loss": 0.3011,
+      "step": 4514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903706450983183e-05,
+      "loss": 0.2872,
+      "step": 4515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036552735228375e-05,
+      "loss": 0.3091,
+      "step": 4516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903604083154653e-05,
+      "loss": 0.3264,
+      "step": 4517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903552879879362e-05,
+      "loss": 0.2866,
+      "step": 4518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035016636976942e-05,
+      "loss": 0.3074,
+      "step": 4519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034504346103825e-05,
+      "loss": 0.3003,
+      "step": 4520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903399192618158e-05,
+      "loss": 0.3128,
+      "step": 4521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033479377217527e-05,
+      "loss": 0.2787,
+      "step": 4522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032966699218992e-05,
+      "loss": 0.2665,
+      "step": 4523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903245389219329e-05,
+      "loss": 0.3016,
+      "step": 4524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903194095614775e-05,
+      "loss": 0.3085,
+      "step": 4525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9031427891089702e-05,
+      "loss": 0.3349,
+      "step": 4526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903091469702647e-05,
+      "loss": 0.2992,
+      "step": 4527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903040137396539e-05,
+      "loss": 0.3144,
+      "step": 4528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029887921913787e-05,
+      "loss": 0.3,
+      "step": 4529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029374340879e-05,
+      "loss": 0.2834,
+      "step": 4530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028860630868364e-05,
+      "loss": 0.305,
+      "step": 4531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028346791889216e-05,
+      "loss": 0.308,
+      "step": 4532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027832823948896e-05,
+      "loss": 0.2982,
+      "step": 4533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027318727054748e-05,
+      "loss": 0.3321,
+      "step": 4534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902680450121411e-05,
+      "loss": 0.2684,
+      "step": 4535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902629014643433e-05,
+      "loss": 0.3532,
+      "step": 4536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025775662722757e-05,
+      "loss": 0.2848,
+      "step": 4537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902526105008674e-05,
+      "loss": 0.288,
+      "step": 4538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024746308533628e-05,
+      "loss": 0.31,
+      "step": 4539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902423143807077e-05,
+      "loss": 0.2748,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902371643870553e-05,
+      "loss": 0.3202,
+      "step": 4541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023201310445253e-05,
+      "loss": 0.2898,
+      "step": 4542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022686053297304e-05,
+      "loss": 0.3345,
+      "step": 4543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022170667269046e-05,
+      "loss": 0.3372,
+      "step": 4544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902165515236783e-05,
+      "loss": 0.287,
+      "step": 4545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021139508601028e-05,
+      "loss": 0.3063,
+      "step": 4546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020623735976003e-05,
+      "loss": 0.2657,
+      "step": 4547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020107834500123e-05,
+      "loss": 0.2708,
+      "step": 4548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019591804180756e-05,
+      "loss": 0.262,
+      "step": 4549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019075645025277e-05,
+      "loss": 0.2977,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018559357041052e-05,
+      "loss": 0.2816,
+      "step": 4551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901804294023546e-05,
+      "loss": 0.2846,
+      "step": 4552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901752639461588e-05,
+      "loss": 0.3472,
+      "step": 4553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901700972018968e-05,
+      "loss": 0.2714,
+      "step": 4554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016492916964253e-05,
+      "loss": 0.297,
+      "step": 4555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015975984946978e-05,
+      "loss": 0.279,
+      "step": 4556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901545892414523e-05,
+      "loss": 0.3088,
+      "step": 4557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014941734566405e-05,
+      "loss": 0.3118,
+      "step": 4558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014424416217883e-05,
+      "loss": 0.3131,
+      "step": 4559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901390696910706e-05,
+      "loss": 0.2888,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9013389393241324e-05,
+      "loss": 0.2954,
+      "step": 4561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012871688628065e-05,
+      "loss": 0.3107,
+      "step": 4562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012353855274685e-05,
+      "loss": 0.3072,
+      "step": 4563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011835893188574e-05,
+      "loss": 0.3071,
+      "step": 4564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011317802377135e-05,
+      "loss": 0.2806,
+      "step": 4565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010799582847767e-05,
+      "loss": 0.3075,
+      "step": 4566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010281234607873e-05,
+      "loss": 0.2779,
+      "step": 4567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009762757664857e-05,
+      "loss": 0.29,
+      "step": 4568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009244152026122e-05,
+      "loss": 0.3165,
+      "step": 4569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008725417699077e-05,
+      "loss": 0.3194,
+      "step": 4570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900820655469113e-05,
+      "loss": 0.285,
+      "step": 4571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007687563009704e-05,
+      "loss": 0.3026,
+      "step": 4572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90071684426622e-05,
+      "loss": 0.2832,
+      "step": 4573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006649193656034e-05,
+      "loss": 0.3122,
+      "step": 4574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006129815998622e-05,
+      "loss": 0.2775,
+      "step": 4575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900561030969739e-05,
+      "loss": 0.2849,
+      "step": 4576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005090674759756e-05,
+      "loss": 0.2734,
+      "step": 4577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004570911193144e-05,
+      "loss": 0.2841,
+      "step": 4578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004051019004972e-05,
+      "loss": 0.2994,
+      "step": 4579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003530998202667e-05,
+      "loss": 0.3095,
+      "step": 4580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003010848793666e-05,
+      "loss": 0.2883,
+      "step": 4581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900249057078539e-05,
+      "loss": 0.3077,
+      "step": 4582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900197016418527e-05,
+      "loss": 0.2784,
+      "step": 4583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001449629000746e-05,
+      "loss": 0.3276,
+      "step": 4584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900092896523925e-05,
+      "loss": 0.3145,
+      "step": 4585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900040817290822e-05,
+      "loss": 0.327,
+      "step": 4586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999887252015092e-05,
+      "loss": 0.313,
+      "step": 4587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999366202567313e-05,
+      "loss": 0.3119,
+      "step": 4588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899884502457232e-05,
+      "loss": 0.3037,
+      "step": 4589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998323718037558e-05,
+      "loss": 0.283,
+      "step": 4590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997802282970477e-05,
+      "loss": 0.2768,
+      "step": 4591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899728071937852e-05,
+      "loss": 0.3043,
+      "step": 4592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996759027269144e-05,
+      "loss": 0.3006,
+      "step": 4593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996237206649795e-05,
+      "loss": 0.279,
+      "step": 4594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995715257527926e-05,
+      "loss": 0.3104,
+      "step": 4595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995193179911e-05,
+      "loss": 0.2862,
+      "step": 4596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994670973806465e-05,
+      "loss": 0.2882,
+      "step": 4597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994148639221785e-05,
+      "loss": 0.3057,
+      "step": 4598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993626176164423e-05,
+      "loss": 0.2799,
+      "step": 4599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899310358464184e-05,
+      "loss": 0.2805,
+      "step": 4600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89925808646615e-05,
+      "loss": 0.2897,
+      "step": 4601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992058016230868e-05,
+      "loss": 0.2823,
+      "step": 4602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991535039357415e-05,
+      "loss": 0.2886,
+      "step": 4603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899101193404861e-05,
+      "loss": 0.2885,
+      "step": 4604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990488700311927e-05,
+      "loss": 0.3174,
+      "step": 4605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989965338154838e-05,
+      "loss": 0.2952,
+      "step": 4606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989441847584817e-05,
+      "loss": 0.2901,
+      "step": 4607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988918228609347e-05,
+      "loss": 0.2626,
+      "step": 4608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89883944812359e-05,
+      "loss": 0.2698,
+      "step": 4609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987870605471964e-05,
+      "loss": 0.2907,
+      "step": 4610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898734660132502e-05,
+      "loss": 0.2992,
+      "step": 4611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898682246880255e-05,
+      "loss": 0.3071,
+      "step": 4612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986298207912046e-05,
+      "loss": 0.3023,
+      "step": 4613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985773818660992e-05,
+      "loss": 0.2838,
+      "step": 4614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985249301056878e-05,
+      "loss": 0.3051,
+      "step": 4615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89847246551072e-05,
+      "loss": 0.2689,
+      "step": 4616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898419988081945e-05,
+      "loss": 0.2798,
+      "step": 4617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983674978201123e-05,
+      "loss": 0.3064,
+      "step": 4618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983149947259722e-05,
+      "loss": 0.3066,
+      "step": 4619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982624788002736e-05,
+      "loss": 0.2972,
+      "step": 4620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898209950043768e-05,
+      "loss": 0.2783,
+      "step": 4621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981574084572042e-05,
+      "loss": 0.2635,
+      "step": 4622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981048540413337e-05,
+      "loss": 0.2994,
+      "step": 4623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980522867969077e-05,
+      "loss": 0.3022,
+      "step": 4624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979997067246755e-05,
+      "loss": 0.2864,
+      "step": 4625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979471138253894e-05,
+      "loss": 0.2706,
+      "step": 4626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978945080998002e-05,
+      "loss": 0.269,
+      "step": 4627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978418895486593e-05,
+      "loss": 0.2669,
+      "step": 4628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977892581727186e-05,
+      "loss": 0.3038,
+      "step": 4629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977366139727296e-05,
+      "loss": 0.2993,
+      "step": 4630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976839569494444e-05,
+      "loss": 0.2922,
+      "step": 4631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897631287103615e-05,
+      "loss": 0.2789,
+      "step": 4632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975786044359936e-05,
+      "loss": 0.2839,
+      "step": 4633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975259089473336e-05,
+      "loss": 0.2767,
+      "step": 4634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897473200638386e-05,
+      "loss": 0.2991,
+      "step": 4635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974204795099056e-05,
+      "loss": 0.2837,
+      "step": 4636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973677455626444e-05,
+      "loss": 0.2992,
+      "step": 4637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973149987973562e-05,
+      "loss": 0.2956,
+      "step": 4638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972622392147935e-05,
+      "loss": 0.3381,
+      "step": 4639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897209466815711e-05,
+      "loss": 0.2779,
+      "step": 4640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971566816008617e-05,
+      "loss": 0.2741,
+      "step": 4641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897103883571e-05,
+      "loss": 0.3031,
+      "step": 4642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970510727268802e-05,
+      "loss": 0.2935,
+      "step": 4643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8969982490692563e-05,
+      "loss": 0.2771,
+      "step": 4644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896945412598883e-05,
+      "loss": 0.3387,
+      "step": 4645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968925633165147e-05,
+      "loss": 0.3292,
+      "step": 4646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896839701222907e-05,
+      "loss": 0.2898,
+      "step": 4647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967868263188144e-05,
+      "loss": 0.3004,
+      "step": 4648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967339386049923e-05,
+      "loss": 0.2874,
+      "step": 4649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966810380821964e-05,
+      "loss": 0.3006,
+      "step": 4650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896628124751182e-05,
+      "loss": 0.2857,
+      "step": 4651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965751986127046e-05,
+      "loss": 0.2797,
+      "step": 4652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965222596675212e-05,
+      "loss": 0.274,
+      "step": 4653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896469307916387e-05,
+      "loss": 0.3159,
+      "step": 4654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896416343360059e-05,
+      "loss": 0.2839,
+      "step": 4655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8963633659992937e-05,
+      "loss": 0.3005,
+      "step": 4656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8963103758348475e-05,
+      "loss": 0.2684,
+      "step": 4657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962573728674773e-05,
+      "loss": 0.2726,
+      "step": 4658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962043570979407e-05,
+      "loss": 0.287,
+      "step": 4659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961513285269943e-05,
+      "loss": 0.3037,
+      "step": 4660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960982871553963e-05,
+      "loss": 0.3241,
+      "step": 4661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960452329839035e-05,
+      "loss": 0.2938,
+      "step": 4662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959921660132744e-05,
+      "loss": 0.2639,
+      "step": 4663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959390862442668e-05,
+      "loss": 0.2496,
+      "step": 4664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895885993677639e-05,
+      "loss": 0.2821,
+      "step": 4665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958328883141492e-05,
+      "loss": 0.2738,
+      "step": 4666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895779770154556e-05,
+      "loss": 0.2655,
+      "step": 4667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895726639199618e-05,
+      "loss": 0.2794,
+      "step": 4668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956734954500944e-05,
+      "loss": 0.278,
+      "step": 4669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956203389067442e-05,
+      "loss": 0.2854,
+      "step": 4670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955671695703267e-05,
+      "loss": 0.2897,
+      "step": 4671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955139874416012e-05,
+      "loss": 0.2541,
+      "step": 4672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954607925213277e-05,
+      "loss": 0.2762,
+      "step": 4673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895407584810266e-05,
+      "loss": 0.3007,
+      "step": 4674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953543643091756e-05,
+      "loss": 0.317,
+      "step": 4675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895301131018817e-05,
+      "loss": 0.3042,
+      "step": 4676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895247884939951e-05,
+      "loss": 0.2763,
+      "step": 4677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951946260733377e-05,
+      "loss": 0.2861,
+      "step": 4678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895141354419738e-05,
+      "loss": 0.2914,
+      "step": 4679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950880699799126e-05,
+      "loss": 0.3124,
+      "step": 4680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895034772754623e-05,
+      "loss": 0.2678,
+      "step": 4681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949814627446304e-05,
+      "loss": 0.2967,
+      "step": 4682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949281399506962e-05,
+      "loss": 0.2864,
+      "step": 4683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894874804373582e-05,
+      "loss": 0.3294,
+      "step": 4684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948214560140497e-05,
+      "loss": 0.3038,
+      "step": 4685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947680948728614e-05,
+      "loss": 0.2886,
+      "step": 4686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894714720950779e-05,
+      "loss": 0.3204,
+      "step": 4687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946613342485654e-05,
+      "loss": 0.2961,
+      "step": 4688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946079347669827e-05,
+      "loss": 0.2682,
+      "step": 4689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945545225067943e-05,
+      "loss": 0.2763,
+      "step": 4690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945010974687626e-05,
+      "loss": 0.3272,
+      "step": 4691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944476596536505e-05,
+      "loss": 0.3138,
+      "step": 4692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894394209062222e-05,
+      "loss": 0.3086,
+      "step": 4693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89434074569524e-05,
+      "loss": 0.3129,
+      "step": 4694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894287269553469e-05,
+      "loss": 0.3097,
+      "step": 4695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8942337806376715e-05,
+      "loss": 0.3264,
+      "step": 4696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894180278948613e-05,
+      "loss": 0.2609,
+      "step": 4697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941267644870565e-05,
+      "loss": 0.2806,
+      "step": 4698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894073237253767e-05,
+      "loss": 0.3242,
+      "step": 4699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940196972495096e-05,
+      "loss": 0.2852,
+      "step": 4700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893966144475048e-05,
+      "loss": 0.2877,
+      "step": 4701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939125789311475e-05,
+      "loss": 0.2939,
+      "step": 4702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938590006185737e-05,
+      "loss": 0.3179,
+      "step": 4703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938054095380916e-05,
+      "loss": 0.3017,
+      "step": 4704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937518056904665e-05,
+      "loss": 0.2962,
+      "step": 4705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936981890764645e-05,
+      "loss": 0.2918,
+      "step": 4706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893644559696851e-05,
+      "loss": 0.2994,
+      "step": 4707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935909175523923e-05,
+      "loss": 0.2725,
+      "step": 4708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935372626438546e-05,
+      "loss": 0.2708,
+      "step": 4709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934835949720045e-05,
+      "loss": 0.2738,
+      "step": 4710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893429914537608e-05,
+      "loss": 0.2662,
+      "step": 4711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933762213414323e-05,
+      "loss": 0.285,
+      "step": 4712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933225153842446e-05,
+      "loss": 0.2883,
+      "step": 4713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932687966668118e-05,
+      "loss": 0.2654,
+      "step": 4714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893215065189901e-05,
+      "loss": 0.2933,
+      "step": 4715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931613209542797e-05,
+      "loss": 0.3132,
+      "step": 4716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931075639607163e-05,
+      "loss": 0.2868,
+      "step": 4717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930537942099777e-05,
+      "loss": 0.3001,
+      "step": 4718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930000117028325e-05,
+      "loss": 0.2844,
+      "step": 4719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892946216440049e-05,
+      "loss": 0.2692,
+      "step": 4720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892892408422395e-05,
+      "loss": 0.2979,
+      "step": 4721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928385876506396e-05,
+      "loss": 0.2841,
+      "step": 4722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927847541255517e-05,
+      "loss": 0.3185,
+      "step": 4723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927309078479e-05,
+      "loss": 0.2917,
+      "step": 4724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926770488184537e-05,
+      "loss": 0.2911,
+      "step": 4725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926231770379823e-05,
+      "loss": 0.2851,
+      "step": 4726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892569292507255e-05,
+      "loss": 0.2855,
+      "step": 4727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925153952270416e-05,
+      "loss": 0.277,
+      "step": 4728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924614851981117e-05,
+      "loss": 0.3002,
+      "step": 4729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892407562421236e-05,
+      "loss": 0.2761,
+      "step": 4730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923536268971844e-05,
+      "loss": 0.3368,
+      "step": 4731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892299678626727e-05,
+      "loss": 0.2833,
+      "step": 4732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892245717610635e-05,
+      "loss": 0.2979,
+      "step": 4733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921917438496786e-05,
+      "loss": 0.2859,
+      "step": 4734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921377573446288e-05,
+      "loss": 0.2805,
+      "step": 4735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920837580962573e-05,
+      "loss": 0.2654,
+      "step": 4736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920297461053353e-05,
+      "loss": 0.2981,
+      "step": 4737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891975721372634e-05,
+      "loss": 0.3122,
+      "step": 4738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919216838989252e-05,
+      "loss": 0.2984,
+      "step": 4739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918676336849807e-05,
+      "loss": 0.307,
+      "step": 4740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918135707315727e-05,
+      "loss": 0.3001,
+      "step": 4741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891759495039473e-05,
+      "loss": 0.3007,
+      "step": 4742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917054066094546e-05,
+      "loss": 0.294,
+      "step": 4743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916513054422903e-05,
+      "loss": 0.2904,
+      "step": 4744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891597191538752e-05,
+      "loss": 0.2934,
+      "step": 4745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915430648996133e-05,
+      "loss": 0.3146,
+      "step": 4746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914889255256473e-05,
+      "loss": 0.2937,
+      "step": 4747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914347734176268e-05,
+      "loss": 0.3032,
+      "step": 4748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913806085763263e-05,
+      "loss": 0.2805,
+      "step": 4749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913264310025184e-05,
+      "loss": 0.2999,
+      "step": 4750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912722406969776e-05,
+      "loss": 0.2999,
+      "step": 4751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912180376604777e-05,
+      "loss": 0.3139,
+      "step": 4752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911638218937936e-05,
+      "loss": 0.2842,
+      "step": 4753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891109593397699e-05,
+      "loss": 0.2945,
+      "step": 4754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910553521729686e-05,
+      "loss": 0.3057,
+      "step": 4755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910010982203774e-05,
+      "loss": 0.2743,
+      "step": 4756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909468315407e-05,
+      "loss": 0.2825,
+      "step": 4757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890892552134712e-05,
+      "loss": 0.2761,
+      "step": 4758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8908382600031886e-05,
+      "loss": 0.2832,
+      "step": 4759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907839551469046e-05,
+      "loss": 0.2615,
+      "step": 4760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890729637566637e-05,
+      "loss": 0.3348,
+      "step": 4761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906753072631606e-05,
+      "loss": 0.2831,
+      "step": 4762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906209642372525e-05,
+      "loss": 0.3554,
+      "step": 4763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890566608489688e-05,
+      "loss": 0.2697,
+      "step": 4764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905122400212435e-05,
+      "loss": 0.3199,
+      "step": 4765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890457858832696e-05,
+      "loss": 0.2695,
+      "step": 4766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904034649248223e-05,
+      "loss": 0.2775,
+      "step": 4767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890349058298399e-05,
+      "loss": 0.3098,
+      "step": 4768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890294638954204e-05,
+      "loss": 0.2968,
+      "step": 4769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902402068930138e-05,
+      "loss": 0.2774,
+      "step": 4770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901857621156066e-05,
+      "loss": 0.3208,
+      "step": 4771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901313046227593e-05,
+      "loss": 0.2654,
+      "step": 4772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900768344152505e-05,
+      "loss": 0.2849,
+      "step": 4773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900223514938578e-05,
+      "loss": 0.2768,
+      "step": 4774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899678558593595e-05,
+      "loss": 0.2962,
+      "step": 4775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899133475125344e-05,
+      "loss": 0.2966,
+      "step": 4776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898588264541606e-05,
+      "loss": 0.2642,
+      "step": 4777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898042926850172e-05,
+      "loss": 0.2985,
+      "step": 4778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897497462058833e-05,
+      "loss": 0.2696,
+      "step": 4779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896951870175373e-05,
+      "loss": 0.2781,
+      "step": 4780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896406151207593e-05,
+      "loss": 0.3007,
+      "step": 4781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895860305163284e-05,
+      "loss": 0.2869,
+      "step": 4782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895314332050244e-05,
+      "loss": 0.28,
+      "step": 4783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894768231876272e-05,
+      "loss": 0.2866,
+      "step": 4784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889422200464917e-05,
+      "loss": 0.2808,
+      "step": 4785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889367565037674e-05,
+      "loss": 0.3108,
+      "step": 4786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889312916906678e-05,
+      "loss": 0.2928,
+      "step": 4787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88925825607271e-05,
+      "loss": 0.2904,
+      "step": 4788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892035825365516e-05,
+      "loss": 0.2754,
+      "step": 4789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891488962989824e-05,
+      "loss": 0.3092,
+      "step": 4790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890941973607843e-05,
+      "loss": 0.3269,
+      "step": 4791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890394857227387e-05,
+      "loss": 0.2688,
+      "step": 4792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889847613856268e-05,
+      "loss": 0.2942,
+      "step": 4793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889300243502303e-05,
+      "loss": 0.3021,
+      "step": 4794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888875274617331e-05,
+      "loss": 0.301,
+      "step": 4795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888820512187711e-05,
+      "loss": 0.272,
+      "step": 4796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888765737062153e-05,
+      "loss": 0.2999,
+      "step": 4797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888710949241439e-05,
+      "loss": 0.2789,
+      "step": 4798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886561487263512e-05,
+      "loss": 0.2795,
+      "step": 4799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888601335517673e-05,
+      "loss": 0.2908,
+      "step": 4800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885465096161873e-05,
+      "loss": 0.3022,
+      "step": 4801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888491671022677e-05,
+      "loss": 0.2766,
+      "step": 4802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884368197379254e-05,
+      "loss": 0.2685,
+      "step": 4803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883819557627155e-05,
+      "loss": 0.3006,
+      "step": 4804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883270790978323e-05,
+      "loss": 0.2842,
+      "step": 4805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882721897440588e-05,
+      "loss": 0.2565,
+      "step": 4806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888217287702179e-05,
+      "loss": 0.2976,
+      "step": 4807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881623729729775e-05,
+      "loss": 0.2947,
+      "step": 4808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881074455572385e-05,
+      "loss": 0.3054,
+      "step": 4809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880525054557465e-05,
+      "loss": 0.2817,
+      "step": 4810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887997552669286e-05,
+      "loss": 0.2619,
+      "step": 4811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879425871986425e-05,
+      "loss": 0.2906,
+      "step": 4812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887887609044601e-05,
+      "loss": 0.2955,
+      "step": 4813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878326182079466e-05,
+      "loss": 0.2837,
+      "step": 4814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887777614689465e-05,
+      "loss": 0.3053,
+      "step": 4815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887722598489942e-05,
+      "loss": 0.2575,
+      "step": 4816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876675696101627e-05,
+      "loss": 0.3196,
+      "step": 4817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876125280509142e-05,
+      "loss": 0.2779,
+      "step": 4818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875574738129814e-05,
+      "loss": 0.2627,
+      "step": 4819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875024068971525e-05,
+      "loss": 0.2544,
+      "step": 4820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874473273042124e-05,
+      "loss": 0.313,
+      "step": 4821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873922350349486e-05,
+      "loss": 0.3005,
+      "step": 4822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887337130090148e-05,
+      "loss": 0.2742,
+      "step": 4823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887282012470598e-05,
+      "loss": 0.2805,
+      "step": 4824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872268821770855e-05,
+      "loss": 0.3137,
+      "step": 4825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871717392103976e-05,
+      "loss": 0.2906,
+      "step": 4826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8871165835713227e-05,
+      "loss": 0.2951,
+      "step": 4827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870614152606484e-05,
+      "loss": 0.3082,
+      "step": 4828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887006234279163e-05,
+      "loss": 0.3465,
+      "step": 4829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8869510406276538e-05,
+      "loss": 0.295,
+      "step": 4830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88689583430691e-05,
+      "loss": 0.3077,
+      "step": 4831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88684061531772e-05,
+      "loss": 0.3014,
+      "step": 4832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8867853836608727e-05,
+      "loss": 0.2782,
+      "step": 4833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8867301393371568e-05,
+      "loss": 0.288,
+      "step": 4834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886674882347361e-05,
+      "loss": 0.2877,
+      "step": 4835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866196126922752e-05,
+      "loss": 0.2873,
+      "step": 4836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865643303726894e-05,
+      "loss": 0.2738,
+      "step": 4837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886509035389392e-05,
+      "loss": 0.2823,
+      "step": 4838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864537277431734e-05,
+      "loss": 0.2833,
+      "step": 4839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863984074348236e-05,
+      "loss": 0.2642,
+      "step": 4840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886343074465133e-05,
+      "loss": 0.2709,
+      "step": 4841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862877288348917e-05,
+      "loss": 0.2589,
+      "step": 4842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88623237054489e-05,
+      "loss": 0.3227,
+      "step": 4843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861769995959196e-05,
+      "loss": 0.3078,
+      "step": 4844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861216159887706e-05,
+      "loss": 0.2762,
+      "step": 4845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860662197242342e-05,
+      "loss": 0.284,
+      "step": 4846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886010810803102e-05,
+      "loss": 0.3303,
+      "step": 4847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859553892261653e-05,
+      "loss": 0.2898,
+      "step": 4848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858999549942156e-05,
+      "loss": 0.3136,
+      "step": 4849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858445081080445e-05,
+      "loss": 0.2957,
+      "step": 4850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857890485684447e-05,
+      "loss": 0.2722,
+      "step": 4851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885733576376208e-05,
+      "loss": 0.2901,
+      "step": 4852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856780915321266e-05,
+      "loss": 0.2909,
+      "step": 4853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856225940369933e-05,
+      "loss": 0.2829,
+      "step": 4854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855670838916008e-05,
+      "loss": 0.2853,
+      "step": 4855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885511561096742e-05,
+      "loss": 0.2831,
+      "step": 4856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 0.27,
+      "step": 4857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885400477561798e-05,
+      "loss": 0.3073,
+      "step": 4858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853449168232997e-05,
+      "loss": 0.2753,
+      "step": 4859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885289343438508e-05,
+      "loss": 0.2769,
+      "step": 4860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852337574082174e-05,
+      "loss": 0.2908,
+      "step": 4861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885178158733222e-05,
+      "loss": 0.3102,
+      "step": 4862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851225474143154e-05,
+      "loss": 0.2934,
+      "step": 4863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850669234522922e-05,
+      "loss": 0.2737,
+      "step": 4864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850112868479473e-05,
+      "loss": 0.3059,
+      "step": 4865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884955637602075e-05,
+      "loss": 0.3044,
+      "step": 4866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848999757154704e-05,
+      "loss": 0.2895,
+      "step": 4867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848443011889278e-05,
+      "loss": 0.3099,
+      "step": 4868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847886140232438e-05,
+      "loss": 0.2927,
+      "step": 4869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884732914219213e-05,
+      "loss": 0.2605,
+      "step": 4870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846772017776312e-05,
+      "loss": 0.2713,
+      "step": 4871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884621476699294e-05,
+      "loss": 0.2584,
+      "step": 4872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845657389849977e-05,
+      "loss": 0.2846,
+      "step": 4873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884509988635538e-05,
+      "loss": 0.2875,
+      "step": 4874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884454225651712e-05,
+      "loss": 0.3056,
+      "step": 4875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843984500343152e-05,
+      "loss": 0.3136,
+      "step": 4876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843426617841455e-05,
+      "loss": 0.3015,
+      "step": 4877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842868609019987e-05,
+      "loss": 0.2418,
+      "step": 4878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842310473886726e-05,
+      "loss": 0.3196,
+      "step": 4879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884175221244964e-05,
+      "loss": 0.3009,
+      "step": 4880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841193824716705e-05,
+      "loss": 0.3142,
+      "step": 4881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840635310695896e-05,
+      "loss": 0.2754,
+      "step": 4882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840076670395194e-05,
+      "loss": 0.2925,
+      "step": 4883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8839517903822574e-05,
+      "loss": 0.2883,
+      "step": 4884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883895901098602e-05,
+      "loss": 0.2697,
+      "step": 4885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838399991893518e-05,
+      "loss": 0.2735,
+      "step": 4886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837840846553046e-05,
+      "loss": 0.2977,
+      "step": 4887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837281574972597e-05,
+      "loss": 0.3129,
+      "step": 4888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836722177160155e-05,
+      "loss": 0.2833,
+      "step": 4889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836162653123718e-05,
+      "loss": 0.2898,
+      "step": 4890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835603002871267e-05,
+      "loss": 0.2835,
+      "step": 4891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835043226410805e-05,
+      "loss": 0.2985,
+      "step": 4892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834483323750325e-05,
+      "loss": 0.302,
+      "step": 4893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833923294897826e-05,
+      "loss": 0.2945,
+      "step": 4894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833363139861305e-05,
+      "loss": 0.2855,
+      "step": 4895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832802858648763e-05,
+      "loss": 0.3174,
+      "step": 4896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883224245126821e-05,
+      "loss": 0.3238,
+      "step": 4897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883168191772764e-05,
+      "loss": 0.2663,
+      "step": 4898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883112125803507e-05,
+      "loss": 0.3024,
+      "step": 4899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88305604721985e-05,
+      "loss": 0.2718,
+      "step": 4900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829999560225943e-05,
+      "loss": 0.2837,
+      "step": 4901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829438522125417e-05,
+      "loss": 0.2973,
+      "step": 4902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8828877357904924e-05,
+      "loss": 0.2501,
+      "step": 4903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882831606757249e-05,
+      "loss": 0.2686,
+      "step": 4904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882775465113613e-05,
+      "loss": 0.2917,
+      "step": 4905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827193108603863e-05,
+      "loss": 0.2579,
+      "step": 4906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826631439983707e-05,
+      "loss": 0.26,
+      "step": 4907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882606964528369e-05,
+      "loss": 0.2729,
+      "step": 4908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882550772451183e-05,
+      "loss": 0.282,
+      "step": 4909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882494567767616e-05,
+      "loss": 0.2936,
+      "step": 4910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824383504784707e-05,
+      "loss": 0.2906,
+      "step": 4911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823821205845498e-05,
+      "loss": 0.3072,
+      "step": 4912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823258780866567e-05,
+      "loss": 0.2749,
+      "step": 4913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822696229855947e-05,
+      "loss": 0.2924,
+      "step": 4914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822133552821675e-05,
+      "loss": 0.2989,
+      "step": 4915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882157074977179e-05,
+      "loss": 0.2999,
+      "step": 4916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8821007820714326e-05,
+      "loss": 0.2688,
+      "step": 4917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820444765657324e-05,
+      "loss": 0.2988,
+      "step": 4918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881988158460883e-05,
+      "loss": 0.2914,
+      "step": 4919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819318277576892e-05,
+      "loss": 0.2837,
+      "step": 4920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818754844569546e-05,
+      "loss": 0.2888,
+      "step": 4921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818191285594845e-05,
+      "loss": 0.306,
+      "step": 4922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8817627600660843e-05,
+      "loss": 0.2952,
+      "step": 4923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881706378977559e-05,
+      "loss": 0.2942,
+      "step": 4924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816499852947134e-05,
+      "loss": 0.2883,
+      "step": 4925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815935790183538e-05,
+      "loss": 0.2934,
+      "step": 4926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815371601492855e-05,
+      "loss": 0.2968,
+      "step": 4927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881480728688314e-05,
+      "loss": 0.3211,
+      "step": 4928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881424284636246e-05,
+      "loss": 0.2857,
+      "step": 4929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813678279938876e-05,
+      "loss": 0.2923,
+      "step": 4930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813113587620454e-05,
+      "loss": 0.3061,
+      "step": 4931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812548769415254e-05,
+      "loss": 0.2687,
+      "step": 4932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881198382533135e-05,
+      "loss": 0.3106,
+      "step": 4933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811418755376806e-05,
+      "loss": 0.275,
+      "step": 4934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88108535595597e-05,
+      "loss": 0.3022,
+      "step": 4935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810288237888098e-05,
+      "loss": 0.281,
+      "step": 4936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809722790370086e-05,
+      "loss": 0.2787,
+      "step": 4937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880915721701373e-05,
+      "loss": 0.2839,
+      "step": 4938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880859151782711e-05,
+      "loss": 0.2951,
+      "step": 4939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808025692818312e-05,
+      "loss": 0.3026,
+      "step": 4940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807459741995417e-05,
+      "loss": 0.2594,
+      "step": 4941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806893665366503e-05,
+      "loss": 0.301,
+      "step": 4942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806327462939664e-05,
+      "loss": 0.3058,
+      "step": 4943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805761134722984e-05,
+      "loss": 0.2891,
+      "step": 4944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880519468072455e-05,
+      "loss": 0.3128,
+      "step": 4945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880462810095246e-05,
+      "loss": 0.2634,
+      "step": 4946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804061395414795e-05,
+      "loss": 0.2831,
+      "step": 4947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803494564119664e-05,
+      "loss": 0.2899,
+      "step": 4948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802927607075157e-05,
+      "loss": 0.2884,
+      "step": 4949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802360524289368e-05,
+      "loss": 0.3322,
+      "step": 4950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801793315770408e-05,
+      "loss": 0.3209,
+      "step": 4951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801225981526364e-05,
+      "loss": 0.3162,
+      "step": 4952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8800658521565358e-05,
+      "loss": 0.2868,
+      "step": 4953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880009093589548e-05,
+      "loss": 0.2874,
+      "step": 4954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799523224524848e-05,
+      "loss": 0.2644,
+      "step": 4955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798955387461565e-05,
+      "loss": 0.2493,
+      "step": 4956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798387424713742e-05,
+      "loss": 0.2835,
+      "step": 4957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797819336289495e-05,
+      "loss": 0.2674,
+      "step": 4958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797251122196937e-05,
+      "loss": 0.2536,
+      "step": 4959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796682782444188e-05,
+      "loss": 0.2769,
+      "step": 4960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879611431703936e-05,
+      "loss": 0.3018,
+      "step": 4961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795545725990573e-05,
+      "loss": 0.3098,
+      "step": 4962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794977009305957e-05,
+      "loss": 0.2716,
+      "step": 4963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794408166993626e-05,
+      "loss": 0.3132,
+      "step": 4964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879383919906171e-05,
+      "loss": 0.2763,
+      "step": 4965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793270105518336e-05,
+      "loss": 0.306,
+      "step": 4966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792700886371635e-05,
+      "loss": 0.2817,
+      "step": 4967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792131541629737e-05,
+      "loss": 0.2795,
+      "step": 4968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791562071300768e-05,
+      "loss": 0.3115,
+      "step": 4969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879099247539287e-05,
+      "loss": 0.3043,
+      "step": 4970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790422753914176e-05,
+      "loss": 0.3085,
+      "step": 4971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878985290687282e-05,
+      "loss": 0.3072,
+      "step": 4972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789282934276952e-05,
+      "loss": 0.2807,
+      "step": 4973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788712836134704e-05,
+      "loss": 0.299,
+      "step": 4974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788142612454225e-05,
+      "loss": 0.2758,
+      "step": 4975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787572263243653e-05,
+      "loss": 0.2778,
+      "step": 4976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787001788511147e-05,
+      "loss": 0.2765,
+      "step": 4977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786431188264845e-05,
+      "loss": 0.3112,
+      "step": 4978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785860462512898e-05,
+      "loss": 0.2841,
+      "step": 4979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785289611263465e-05,
+      "loss": 0.3048,
+      "step": 4980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784718634524696e-05,
+      "loss": 0.2974,
+      "step": 4981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784147532304748e-05,
+      "loss": 0.3266,
+      "step": 4982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783576304611776e-05,
+      "loss": 0.2952,
+      "step": 4983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878300495145394e-05,
+      "loss": 0.3172,
+      "step": 4984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782433472839404e-05,
+      "loss": 0.3049,
+      "step": 4985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781861868776328e-05,
+      "loss": 0.3169,
+      "step": 4986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878129013927288e-05,
+      "loss": 0.2955,
+      "step": 4987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878071828433722e-05,
+      "loss": 0.3051,
+      "step": 4988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780146303977525e-05,
+      "loss": 0.2804,
+      "step": 4989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779574198201964e-05,
+      "loss": 0.2825,
+      "step": 4990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779001967018705e-05,
+      "loss": 0.2936,
+      "step": 4991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778429610435922e-05,
+      "loss": 0.2757,
+      "step": 4992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877785712846179e-05,
+      "loss": 0.2957,
+      "step": 4993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877728452110449e-05,
+      "loss": 0.3023,
+      "step": 4994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776711788372198e-05,
+      "loss": 0.2905,
+      "step": 4995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776138930273096e-05,
+      "loss": 0.2951,
+      "step": 4996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775565946815368e-05,
+      "loss": 0.2788,
+      "step": 4997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87749928380072e-05,
+      "loss": 0.2809,
+      "step": 4998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877441960385677e-05,
+      "loss": 0.3082,
+      "step": 4999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773846244372277e-05,
+      "loss": 0.2622,
+      "step": 5000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773272759561903e-05,
+      "loss": 0.28,
+      "step": 5001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772699149433842e-05,
+      "loss": 0.2936,
+      "step": 5002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877212541399629e-05,
+      "loss": 0.3072,
+      "step": 5003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771551553257436e-05,
+      "loss": 0.3023,
+      "step": 5004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770977567225484e-05,
+      "loss": 0.2683,
+      "step": 5005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770403455908627e-05,
+      "loss": 0.2953,
+      "step": 5006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769829219315072e-05,
+      "loss": 0.2984,
+      "step": 5007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769254857453015e-05,
+      "loss": 0.2723,
+      "step": 5008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768680370330666e-05,
+      "loss": 0.2812,
+      "step": 5009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768105757956226e-05,
+      "loss": 0.2807,
+      "step": 5010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876753102033791e-05,
+      "loss": 0.2833,
+      "step": 5011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8766956157483914e-05,
+      "loss": 0.2841,
+      "step": 5012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8766381169402465e-05,
+      "loss": 0.269,
+      "step": 5013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765806056101762e-05,
+      "loss": 0.2638,
+      "step": 5014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876523081759003e-05,
+      "loss": 0.3216,
+      "step": 5015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764655453875487e-05,
+      "loss": 0.313,
+      "step": 5016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764079964966342e-05,
+      "loss": 0.2895,
+      "step": 5017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763504350870823e-05,
+      "loss": 0.2598,
+      "step": 5018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762928611597147e-05,
+      "loss": 0.2673,
+      "step": 5019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762352747153543e-05,
+      "loss": 0.2687,
+      "step": 5020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876177675754823e-05,
+      "loss": 0.2994,
+      "step": 5021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761200642789447e-05,
+      "loss": 0.3268,
+      "step": 5022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876062440288541e-05,
+      "loss": 0.3062,
+      "step": 5023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8760048037844356e-05,
+      "loss": 0.2817,
+      "step": 5024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875947154767452e-05,
+      "loss": 0.3074,
+      "step": 5025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758894932384132e-05,
+      "loss": 0.2568,
+      "step": 5026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758318191981433e-05,
+      "loss": 0.3107,
+      "step": 5027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757741326474657e-05,
+      "loss": 0.3314,
+      "step": 5028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757164335872045e-05,
+      "loss": 0.2999,
+      "step": 5029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875658722018184e-05,
+      "loss": 0.2924,
+      "step": 5030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756009979412287e-05,
+      "loss": 0.2726,
+      "step": 5031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755432613571627e-05,
+      "loss": 0.2865,
+      "step": 5032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754855122668114e-05,
+      "loss": 0.2835,
+      "step": 5033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754277506709984e-05,
+      "loss": 0.2936,
+      "step": 5034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753699765705503e-05,
+      "loss": 0.2913,
+      "step": 5035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753121899662915e-05,
+      "loss": 0.2724,
+      "step": 5036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752543908590475e-05,
+      "loss": 0.2799,
+      "step": 5037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875196579249644e-05,
+      "loss": 0.2932,
+      "step": 5038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751387551389066e-05,
+      "loss": 0.2574,
+      "step": 5039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750809185276617e-05,
+      "loss": 0.2871,
+      "step": 5040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750230694167347e-05,
+      "loss": 0.2974,
+      "step": 5041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749652078069532e-05,
+      "loss": 0.308,
+      "step": 5042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874907333699142e-05,
+      "loss": 0.2856,
+      "step": 5043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874849447094129e-05,
+      "loss": 0.2956,
+      "step": 5044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874791547992741e-05,
+      "loss": 0.2758,
+      "step": 5045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747336363958048e-05,
+      "loss": 0.2974,
+      "step": 5046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874675712304147e-05,
+      "loss": 0.3079,
+      "step": 5047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874617775718596e-05,
+      "loss": 0.2691,
+      "step": 5048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745598266399783e-05,
+      "loss": 0.2594,
+      "step": 5049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874501865069123e-05,
+      "loss": 0.3037,
+      "step": 5050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8744438910068567e-05,
+      "loss": 0.3094,
+      "step": 5051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743859044540085e-05,
+      "loss": 0.2997,
+      "step": 5052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874327905411406e-05,
+      "loss": 0.302,
+      "step": 5053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8742698938798783e-05,
+      "loss": 0.2584,
+      "step": 5054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8742118698602533e-05,
+      "loss": 0.3193,
+      "step": 5055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741538333533603e-05,
+      "loss": 0.2977,
+      "step": 5056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874095784360028e-05,
+      "loss": 0.2719,
+      "step": 5057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874037722881086e-05,
+      "loss": 0.3353,
+      "step": 5058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739796489173635e-05,
+      "loss": 0.282,
+      "step": 5059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739215624696895e-05,
+      "loss": 0.2759,
+      "step": 5060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738634635388945e-05,
+      "loss": 0.306,
+      "step": 5061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873805352125808e-05,
+      "loss": 0.3058,
+      "step": 5062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87374722823126e-05,
+      "loss": 0.2955,
+      "step": 5063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873689091856081e-05,
+      "loss": 0.2969,
+      "step": 5064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736309430011007e-05,
+      "loss": 0.2899,
+      "step": 5065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873572781667151e-05,
+      "loss": 0.3065,
+      "step": 5066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735146078550612e-05,
+      "loss": 0.2595,
+      "step": 5067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734564215656635e-05,
+      "loss": 0.2654,
+      "step": 5068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733982227997886e-05,
+      "loss": 0.2793,
+      "step": 5069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733400115582676e-05,
+      "loss": 0.2653,
+      "step": 5070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732817878419317e-05,
+      "loss": 0.2825,
+      "step": 5071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732235516516137e-05,
+      "loss": 0.3016,
+      "step": 5072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731653029881442e-05,
+      "loss": 0.2643,
+      "step": 5073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731070418523562e-05,
+      "loss": 0.2882,
+      "step": 5074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730487682450814e-05,
+      "loss": 0.2841,
+      "step": 5075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872990482167152e-05,
+      "loss": 0.2745,
+      "step": 5076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872932183619401e-05,
+      "loss": 0.3144,
+      "step": 5077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728738726026606e-05,
+      "loss": 0.2942,
+      "step": 5078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872815549117765e-05,
+      "loss": 0.3147,
+      "step": 5079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727572131655453e-05,
+      "loss": 0.3025,
+      "step": 5080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726988647468366e-05,
+      "loss": 0.2765,
+      "step": 5081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872640503862471e-05,
+      "loss": 0.2914,
+      "step": 5082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725821305132832e-05,
+      "loss": 0.2765,
+      "step": 5083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725237447001065e-05,
+      "loss": 0.3075,
+      "step": 5084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724653464237747e-05,
+      "loss": 0.3209,
+      "step": 5085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872406935685122e-05,
+      "loss": 0.3208,
+      "step": 5086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872348512484983e-05,
+      "loss": 0.2932,
+      "step": 5087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722900768241925e-05,
+      "loss": 0.2937,
+      "step": 5088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722316287035846e-05,
+      "loss": 0.2615,
+      "step": 5089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721731681239945e-05,
+      "loss": 0.2971,
+      "step": 5090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872114695086257e-05,
+      "loss": 0.2783,
+      "step": 5091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872056209591208e-05,
+      "loss": 0.2741,
+      "step": 5092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719977116396815e-05,
+      "loss": 0.2683,
+      "step": 5093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719392012325146e-05,
+      "loss": 0.3036,
+      "step": 5094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871880678370542e-05,
+      "loss": 0.3121,
+      "step": 5095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8718221430546007e-05,
+      "loss": 0.2739,
+      "step": 5096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717635952855258e-05,
+      "loss": 0.2682,
+      "step": 5097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871705035064154e-05,
+      "loss": 0.2703,
+      "step": 5098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871646462391322e-05,
+      "loss": 0.3183,
+      "step": 5099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871587877267866e-05,
+      "loss": 0.2957,
+      "step": 5100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871529279694623e-05,
+      "loss": 0.3021,
+      "step": 5101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714706696724307e-05,
+      "loss": 0.3019,
+      "step": 5102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714120472021252e-05,
+      "loss": 0.311,
+      "step": 5103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713534122845442e-05,
+      "loss": 0.2731,
+      "step": 5104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712947649205255e-05,
+      "loss": 0.3021,
+      "step": 5105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871236105110907e-05,
+      "loss": 0.3068,
+      "step": 5106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711774328565257e-05,
+      "loss": 0.2721,
+      "step": 5107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711187481582208e-05,
+      "loss": 0.2691,
+      "step": 5108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710600510168297e-05,
+      "loss": 0.2826,
+      "step": 5109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710013414331913e-05,
+      "loss": 0.2764,
+      "step": 5110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870942619408144e-05,
+      "loss": 0.2695,
+      "step": 5111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708838849425267e-05,
+      "loss": 0.2653,
+      "step": 5112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870825138037178e-05,
+      "loss": 0.2644,
+      "step": 5113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870766378692938e-05,
+      "loss": 0.2783,
+      "step": 5114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870707606910645e-05,
+      "loss": 0.2941,
+      "step": 5115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870648822691139e-05,
+      "loss": 0.3262,
+      "step": 5116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705900260352595e-05,
+      "loss": 0.3003,
+      "step": 5117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705312169438463e-05,
+      "loss": 0.3026,
+      "step": 5118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704723954177394e-05,
+      "loss": 0.2665,
+      "step": 5119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704135614577798e-05,
+      "loss": 0.2809,
+      "step": 5120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703547150648066e-05,
+      "loss": 0.2669,
+      "step": 5121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870295856239661e-05,
+      "loss": 0.268,
+      "step": 5122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870236984983184e-05,
+      "loss": 0.2649,
+      "step": 5123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870178101296216e-05,
+      "loss": 0.2603,
+      "step": 5124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701192051795988e-05,
+      "loss": 0.2818,
+      "step": 5125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870060296634173e-05,
+      "loss": 0.2875,
+      "step": 5126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700013756607802e-05,
+      "loss": 0.3114,
+      "step": 5127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869942442260262e-05,
+      "loss": 0.3083,
+      "step": 5128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698834964334604e-05,
+      "loss": 0.2946,
+      "step": 5129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698245381812176e-05,
+      "loss": 0.293,
+      "step": 5130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869765567504375e-05,
+      "loss": 0.3111,
+      "step": 5131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697065844037756e-05,
+      "loss": 0.2766,
+      "step": 5132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696475888802616e-05,
+      "loss": 0.3225,
+      "step": 5133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695885809346758e-05,
+      "loss": 0.2679,
+      "step": 5134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695295605678614e-05,
+      "loss": 0.3256,
+      "step": 5135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869470527780661e-05,
+      "loss": 0.3013,
+      "step": 5136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869411482573918e-05,
+      "loss": 0.3215,
+      "step": 5137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8693524249484754e-05,
+      "loss": 0.3051,
+      "step": 5138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692933549051773e-05,
+      "loss": 0.2735,
+      "step": 5139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692342724448674e-05,
+      "loss": 0.3052,
+      "step": 5140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691751775683895e-05,
+      "loss": 0.2606,
+      "step": 5141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691160702765878e-05,
+      "loss": 0.254,
+      "step": 5142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690569505703063e-05,
+      "loss": 0.2795,
+      "step": 5143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86899781845039e-05,
+      "loss": 0.2982,
+      "step": 5144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689386739176833e-05,
+      "loss": 0.2939,
+      "step": 5145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688795169730304e-05,
+      "loss": 0.2814,
+      "step": 5146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868820347617277e-05,
+      "loss": 0.2515,
+      "step": 5147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687611658512684e-05,
+      "loss": 0.3145,
+      "step": 5148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86870197167585e-05,
+      "loss": 0.2998,
+      "step": 5149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686427650918665e-05,
+      "loss": 0.2883,
+      "step": 5150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685835461001645e-05,
+      "loss": 0.2945,
+      "step": 5151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685243147015893e-05,
+      "loss": 0.3013,
+      "step": 5152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684650708969874e-05,
+      "loss": 0.2819,
+      "step": 5153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684058146872045e-05,
+      "loss": 0.2984,
+      "step": 5154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683465460730876e-05,
+      "loss": 0.2992,
+      "step": 5155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868287265055483e-05,
+      "loss": 0.2948,
+      "step": 5156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682279716352374e-05,
+      "loss": 0.2841,
+      "step": 5157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681686658131982e-05,
+      "loss": 0.3005,
+      "step": 5158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681093475902118e-05,
+      "loss": 0.3034,
+      "step": 5159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868050016967126e-05,
+      "loss": 0.2582,
+      "step": 5160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867990673944788e-05,
+      "loss": 0.248,
+      "step": 5161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867931318524046e-05,
+      "loss": 0.2999,
+      "step": 5162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678719507057473e-05,
+      "loss": 0.3095,
+      "step": 5163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86781257049074e-05,
+      "loss": 0.288,
+      "step": 5164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677531778798724e-05,
+      "loss": 0.2789,
+      "step": 5165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867693772873993e-05,
+      "loss": 0.3159,
+      "step": 5166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86763435547395e-05,
+      "loss": 0.2762,
+      "step": 5167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867574925680592e-05,
+      "loss": 0.3149,
+      "step": 5168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867515483494769e-05,
+      "loss": 0.2843,
+      "step": 5169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674560289173284e-05,
+      "loss": 0.2925,
+      "step": 5170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867396561949121e-05,
+      "loss": 0.2984,
+      "step": 5171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867337082590995e-05,
+      "loss": 0.2742,
+      "step": 5172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672775908438008e-05,
+      "loss": 0.2984,
+      "step": 5173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672180867083876e-05,
+      "loss": 0.3302,
+      "step": 5174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671585701856062e-05,
+      "loss": 0.3157,
+      "step": 5175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867099041276306e-05,
+      "loss": 0.313,
+      "step": 5176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670394999813375e-05,
+      "loss": 0.2904,
+      "step": 5177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669799463015513e-05,
+      "loss": 0.3183,
+      "step": 5178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866920380237798e-05,
+      "loss": 0.29,
+      "step": 5179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668608017909283e-05,
+      "loss": 0.2981,
+      "step": 5180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668012109617933e-05,
+      "loss": 0.2926,
+      "step": 5181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667416077512443e-05,
+      "loss": 0.2937,
+      "step": 5182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666819921601328e-05,
+      "loss": 0.2763,
+      "step": 5183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86662236418931e-05,
+      "loss": 0.2754,
+      "step": 5184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866562723839628e-05,
+      "loss": 0.2472,
+      "step": 5185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665030711119386e-05,
+      "loss": 0.2953,
+      "step": 5186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664434060070933e-05,
+      "loss": 0.2887,
+      "step": 5187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663837285259453e-05,
+      "loss": 0.3292,
+      "step": 5188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663240386693462e-05,
+      "loss": 0.31,
+      "step": 5189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8662643364381493e-05,
+      "loss": 0.3045,
+      "step": 5190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866204621833207e-05,
+      "loss": 0.2549,
+      "step": 5191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8661448948553724e-05,
+      "loss": 0.315,
+      "step": 5192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8660851555054984e-05,
+      "loss": 0.2695,
+      "step": 5193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.2821,
+      "step": 5194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659656396930467e-05,
+      "loss": 0.2921,
+      "step": 5195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659058632321758e-05,
+      "loss": 0.2965,
+      "step": 5196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86584607440268e-05,
+      "loss": 0.2857,
+      "step": 5197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657862732054136e-05,
+      "loss": 0.2881,
+      "step": 5198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86572645964123e-05,
+      "loss": 0.2777,
+      "step": 5199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656666337109847e-05,
+      "loss": 0.2952,
+      "step": 5200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656067954155316e-05,
+      "loss": 0.2926,
+      "step": 5201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655469447557254e-05,
+      "loss": 0.286,
+      "step": 5202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865487081732421e-05,
+      "loss": 0.2895,
+      "step": 5203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865427206346474e-05,
+      "loss": 0.3083,
+      "step": 5204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653673185987385e-05,
+      "loss": 0.3103,
+      "step": 5205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865307418490071e-05,
+      "loss": 0.2821,
+      "step": 5206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652475060213275e-05,
+      "loss": 0.309,
+      "step": 5207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651875811933625e-05,
+      "loss": 0.2985,
+      "step": 5208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865127644007033e-05,
+      "loss": 0.2719,
+      "step": 5209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650676944631944e-05,
+      "loss": 0.3105,
+      "step": 5210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650077325627034e-05,
+      "loss": 0.3105,
+      "step": 5211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649477583064164e-05,
+      "loss": 0.2977,
+      "step": 5212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648877716951903e-05,
+      "loss": 0.2638,
+      "step": 5213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648277727298815e-05,
+      "loss": 0.276,
+      "step": 5214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647677614113478e-05,
+      "loss": 0.288,
+      "step": 5215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647077377404454e-05,
+      "loss": 0.2703,
+      "step": 5216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646477017180324e-05,
+      "loss": 0.3179,
+      "step": 5217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645876533449663e-05,
+      "loss": 0.2593,
+      "step": 5218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645275926221042e-05,
+      "loss": 0.2944,
+      "step": 5219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864467519550305e-05,
+      "loss": 0.3055,
+      "step": 5220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644074341304258e-05,
+      "loss": 0.2958,
+      "step": 5221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643473363633258e-05,
+      "loss": 0.2674,
+      "step": 5222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642872262498628e-05,
+      "loss": 0.289,
+      "step": 5223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642271037908956e-05,
+      "loss": 0.2818,
+      "step": 5224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641669689872828e-05,
+      "loss": 0.2989,
+      "step": 5225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641068218398835e-05,
+      "loss": 0.3206,
+      "step": 5226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864046662349557e-05,
+      "loss": 0.2917,
+      "step": 5227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639864905171628e-05,
+      "loss": 0.2993,
+      "step": 5228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639263063435597e-05,
+      "loss": 0.2592,
+      "step": 5229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863866109829608e-05,
+      "loss": 0.2885,
+      "step": 5230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638059009761673e-05,
+      "loss": 0.2906,
+      "step": 5231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637456797840972e-05,
+      "loss": 0.2825,
+      "step": 5232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863685446254259e-05,
+      "loss": 0.2791,
+      "step": 5233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863625200387512e-05,
+      "loss": 0.2984,
+      "step": 5234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635649421847174e-05,
+      "loss": 0.2893,
+      "step": 5235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635046716467358e-05,
+      "loss": 0.2841,
+      "step": 5236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634443887744278e-05,
+      "loss": 0.305,
+      "step": 5237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633840935686547e-05,
+      "loss": 0.2981,
+      "step": 5238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633237860302776e-05,
+      "loss": 0.2881,
+      "step": 5239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632634661601584e-05,
+      "loss": 0.2717,
+      "step": 5240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632031339591584e-05,
+      "loss": 0.2949,
+      "step": 5241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631427894281393e-05,
+      "loss": 0.2933,
+      "step": 5242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630824325679632e-05,
+      "loss": 0.2775,
+      "step": 5243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630220633794925e-05,
+      "loss": 0.2764,
+      "step": 5244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629616818635886e-05,
+      "loss": 0.2998,
+      "step": 5245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862901288021115e-05,
+      "loss": 0.281,
+      "step": 5246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628408818529338e-05,
+      "loss": 0.3119,
+      "step": 5247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627804633599082e-05,
+      "loss": 0.2953,
+      "step": 5248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627200325429005e-05,
+      "loss": 0.2772,
+      "step": 5249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862659589402775e-05,
+      "loss": 0.2787,
+      "step": 5250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625991339403943e-05,
+      "loss": 0.2703,
+      "step": 5251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862538666156622e-05,
+      "loss": 0.2891,
+      "step": 5252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862478186052322e-05,
+      "loss": 0.3019,
+      "step": 5253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624176936283584e-05,
+      "loss": 0.2792,
+      "step": 5254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623571888855948e-05,
+      "loss": 0.2886,
+      "step": 5255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862296671824896e-05,
+      "loss": 0.2524,
+      "step": 5256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862236142447126e-05,
+      "loss": 0.3058,
+      "step": 5257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621756007531493e-05,
+      "loss": 0.2634,
+      "step": 5258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862115046743831e-05,
+      "loss": 0.2823,
+      "step": 5259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862054480420036e-05,
+      "loss": 0.3308,
+      "step": 5260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619939017826294e-05,
+      "loss": 0.2644,
+      "step": 5261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8619333108324763e-05,
+      "loss": 0.2874,
+      "step": 5262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861872707570443e-05,
+      "loss": 0.2956,
+      "step": 5263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618120919973938e-05,
+      "loss": 0.2927,
+      "step": 5264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617514641141956e-05,
+      "loss": 0.2693,
+      "step": 5265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616908239217144e-05,
+      "loss": 0.2717,
+      "step": 5266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616301714208155e-05,
+      "loss": 0.2732,
+      "step": 5267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615695066123663e-05,
+      "loss": 0.279,
+      "step": 5268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615088294972327e-05,
+      "loss": 0.3295,
+      "step": 5269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8614481400762815e-05,
+      "loss": 0.2843,
+      "step": 5270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613874383503796e-05,
+      "loss": 0.2882,
+      "step": 5271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613267243203942e-05,
+      "loss": 0.2498,
+      "step": 5272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612659979871925e-05,
+      "loss": 0.2841,
+      "step": 5273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861205259351642e-05,
+      "loss": 0.2931,
+      "step": 5274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86114450841461e-05,
+      "loss": 0.3064,
+      "step": 5275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610837451769647e-05,
+      "loss": 0.2783,
+      "step": 5276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861022969639574e-05,
+      "loss": 0.3079,
+      "step": 5277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609621818033053e-05,
+      "loss": 0.2919,
+      "step": 5278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609013816690275e-05,
+      "loss": 0.2853,
+      "step": 5279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860840569237609e-05,
+      "loss": 0.2594,
+      "step": 5280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607797445099188e-05,
+      "loss": 0.3025,
+      "step": 5281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860718907486825e-05,
+      "loss": 0.3189,
+      "step": 5282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860658058169197e-05,
+      "loss": 0.2772,
+      "step": 5283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8605971965579042e-05,
+      "loss": 0.3029,
+      "step": 5284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860536322653815e-05,
+      "loss": 0.2666,
+      "step": 5285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604754364578003e-05,
+      "loss": 0.3017,
+      "step": 5286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860414537970729e-05,
+      "loss": 0.2745,
+      "step": 5287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603536271934707e-05,
+      "loss": 0.3295,
+      "step": 5288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860292704126896e-05,
+      "loss": 0.3187,
+      "step": 5289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860231768771875e-05,
+      "loss": 0.2955,
+      "step": 5290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601708211292782e-05,
+      "loss": 0.2829,
+      "step": 5291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601098611999758e-05,
+      "loss": 0.2795,
+      "step": 5292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8600488889848392e-05,
+      "loss": 0.2952,
+      "step": 5293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599879044847383e-05,
+      "loss": 0.2729,
+      "step": 5294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599269077005454e-05,
+      "loss": 0.2575,
+      "step": 5295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598658986331307e-05,
+      "loss": 0.3276,
+      "step": 5296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598048772833666e-05,
+      "loss": 0.3,
+      "step": 5297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859743843652124e-05,
+      "loss": 0.2961,
+      "step": 5298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859682797740275e-05,
+      "loss": 0.2741,
+      "step": 5299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596217395486917e-05,
+      "loss": 0.3015,
+      "step": 5300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595606690782462e-05,
+      "loss": 0.2952,
+      "step": 5301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594995863298108e-05,
+      "loss": 0.2628,
+      "step": 5302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859438491304258e-05,
+      "loss": 0.3087,
+      "step": 5303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85937738400246e-05,
+      "loss": 0.3062,
+      "step": 5304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593162644252903e-05,
+      "loss": 0.2745,
+      "step": 5305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8592551325736218e-05,
+      "loss": 0.3442,
+      "step": 5306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859193988448328e-05,
+      "loss": 0.2941,
+      "step": 5307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591328320502816e-05,
+      "loss": 0.3024,
+      "step": 5308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590716633803565e-05,
+      "loss": 0.2684,
+      "step": 5309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590104824394265e-05,
+      "loss": 0.2893,
+      "step": 5310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589492892283653e-05,
+      "loss": 0.3017,
+      "step": 5311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588880837480475e-05,
+      "loss": 0.2881,
+      "step": 5312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858826865999347e-05,
+      "loss": 0.2822,
+      "step": 5313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587656359831378e-05,
+      "loss": 0.3107,
+      "step": 5314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858704393700295e-05,
+      "loss": 0.2695,
+      "step": 5315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8586431391516935e-05,
+      "loss": 0.2811,
+      "step": 5316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858581872338208e-05,
+      "loss": 0.2861,
+      "step": 5317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585205932607137e-05,
+      "loss": 0.2947,
+      "step": 5318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584593019200864e-05,
+      "loss": 0.3027,
+      "step": 5319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583979983172005e-05,
+      "loss": 0.2904,
+      "step": 5320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583366824529324e-05,
+      "loss": 0.2928,
+      "step": 5321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582753543281578e-05,
+      "loss": 0.3257,
+      "step": 5322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582140139437533e-05,
+      "loss": 0.2867,
+      "step": 5323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858152661300594e-05,
+      "loss": 0.294,
+      "step": 5324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858091296399557e-05,
+      "loss": 0.282,
+      "step": 5325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580299192415185e-05,
+      "loss": 0.3243,
+      "step": 5326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579685298273552e-05,
+      "loss": 0.2697,
+      "step": 5327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579071281579444e-05,
+      "loss": 0.267,
+      "step": 5328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578457142341628e-05,
+      "loss": 0.3165,
+      "step": 5329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857784288056888e-05,
+      "loss": 0.2788,
+      "step": 5330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857722849626997e-05,
+      "loss": 0.2946,
+      "step": 5331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8576613989453673e-05,
+      "loss": 0.2834,
+      "step": 5332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857599936012877e-05,
+      "loss": 0.2829,
+      "step": 5333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575384608304038e-05,
+      "loss": 0.2857,
+      "step": 5334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857476973398826e-05,
+      "loss": 0.2641,
+      "step": 5335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857415473719022e-05,
+      "loss": 0.2683,
+      "step": 5336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85735396179187e-05,
+      "loss": 0.3083,
+      "step": 5337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857292437618249e-05,
+      "loss": 0.2776,
+      "step": 5338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857230901199037e-05,
+      "loss": 0.2625,
+      "step": 5339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857169352535114e-05,
+      "loss": 0.29,
+      "step": 5340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571077916273584e-05,
+      "loss": 0.3206,
+      "step": 5341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570462184766503e-05,
+      "loss": 0.2848,
+      "step": 5342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569846330838684e-05,
+      "loss": 0.2718,
+      "step": 5343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856923035449893e-05,
+      "loss": 0.3067,
+      "step": 5344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568614255756034e-05,
+      "loss": 0.3079,
+      "step": 5345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85679980346188e-05,
+      "loss": 0.2447,
+      "step": 5346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856738169109603e-05,
+      "loss": 0.3078,
+      "step": 5347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566765225196527e-05,
+      "loss": 0.2897,
+      "step": 5348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566148636929098e-05,
+      "loss": 0.2805,
+      "step": 5349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565531926302548e-05,
+      "loss": 0.2952,
+      "step": 5350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564915093325687e-05,
+      "loss": 0.2941,
+      "step": 5351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564298138007328e-05,
+      "loss": 0.3137,
+      "step": 5352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563681060356283e-05,
+      "loss": 0.2473,
+      "step": 5353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563063860381364e-05,
+      "loss": 0.3043,
+      "step": 5354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562446538091388e-05,
+      "loss": 0.2823,
+      "step": 5355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8561829093495173e-05,
+      "loss": 0.284,
+      "step": 5356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856121152660154e-05,
+      "loss": 0.298,
+      "step": 5357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560593837419304e-05,
+      "loss": 0.2662,
+      "step": 5358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85599760259573e-05,
+      "loss": 0.262,
+      "step": 5359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8559358092224343e-05,
+      "loss": 0.2901,
+      "step": 5360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558740036229265e-05,
+      "loss": 0.2984,
+      "step": 5361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855812185798089e-05,
+      "loss": 0.2869,
+      "step": 5362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855750355748805e-05,
+      "loss": 0.2907,
+      "step": 5363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556885134759578e-05,
+      "loss": 0.2855,
+      "step": 5364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855626658980431e-05,
+      "loss": 0.2915,
+      "step": 5365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855564792263107e-05,
+      "loss": 0.299,
+      "step": 5366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555029133248708e-05,
+      "loss": 0.3126,
+      "step": 5367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554410221666056e-05,
+      "loss": 0.2776,
+      "step": 5368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553791187891956e-05,
+      "loss": 0.2749,
+      "step": 5369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855317203193525e-05,
+      "loss": 0.2721,
+      "step": 5370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552552753804786e-05,
+      "loss": 0.3119,
+      "step": 5371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551933353509405e-05,
+      "loss": 0.2984,
+      "step": 5372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551313831057952e-05,
+      "loss": 0.2862,
+      "step": 5373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8550694186459286e-05,
+      "loss": 0.2627,
+      "step": 5374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855007441972225e-05,
+      "loss": 0.2681,
+      "step": 5375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549454530855697e-05,
+      "loss": 0.288,
+      "step": 5376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548834519868485e-05,
+      "loss": 0.2925,
+      "step": 5377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854821438676947e-05,
+      "loss": 0.2752,
+      "step": 5378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854759413156751e-05,
+      "loss": 0.249,
+      "step": 5379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854697375427146e-05,
+      "loss": 0.2788,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546353254890182e-05,
+      "loss": 0.2998,
+      "step": 5381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545732633432548e-05,
+      "loss": 0.2861,
+      "step": 5382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545111889907414e-05,
+      "loss": 0.2666,
+      "step": 5383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8544491024323653e-05,
+      "loss": 0.3117,
+      "step": 5384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543870036690125e-05,
+      "loss": 0.2868,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854324892701571e-05,
+      "loss": 0.3181,
+      "step": 5386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542627695309273e-05,
+      "loss": 0.2906,
+      "step": 5387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542006341579693e-05,
+      "loss": 0.2795,
+      "step": 5388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854138486583584e-05,
+      "loss": 0.2925,
+      "step": 5389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540763268086594e-05,
+      "loss": 0.2832,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540141548340837e-05,
+      "loss": 0.3465,
+      "step": 5391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539519706607444e-05,
+      "loss": 0.2722,
+      "step": 5392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85388977428953e-05,
+      "loss": 0.268,
+      "step": 5393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853827565721329e-05,
+      "loss": 0.2704,
+      "step": 5394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537653449570298e-05,
+      "loss": 0.3066,
+      "step": 5395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853703111997521e-05,
+      "loss": 0.306,
+      "step": 5396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853640866843692e-05,
+      "loss": 0.2952,
+      "step": 5397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8535786094964318e-05,
+      "loss": 0.2767,
+      "step": 5398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8535163399566292e-05,
+      "loss": 0.2595,
+      "step": 5399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534540582251745e-05,
+      "loss": 0.3215,
+      "step": 5400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853391764302957e-05,
+      "loss": 0.2988,
+      "step": 5401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533294581908658e-05,
+      "loss": 0.2791,
+      "step": 5402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532671398897915e-05,
+      "loss": 0.2765,
+      "step": 5403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532048094006247e-05,
+      "loss": 0.3244,
+      "step": 5404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853142466724255e-05,
+      "loss": 0.324,
+      "step": 5405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853080111861573e-05,
+      "loss": 0.2959,
+      "step": 5406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530177448134695e-05,
+      "loss": 0.2901,
+      "step": 5407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852955365580836e-05,
+      "loss": 0.2675,
+      "step": 5408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528929741645624e-05,
+      "loss": 0.2976,
+      "step": 5409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528305705655404e-05,
+      "loss": 0.2618,
+      "step": 5410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527681547846616e-05,
+      "loss": 0.2924,
+      "step": 5411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527057268228175e-05,
+      "loss": 0.2863,
+      "step": 5412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852643286680899e-05,
+      "loss": 0.3041,
+      "step": 5413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525808343597993e-05,
+      "loss": 0.2933,
+      "step": 5414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525183698604098e-05,
+      "loss": 0.3067,
+      "step": 5415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524558931836223e-05,
+      "loss": 0.2921,
+      "step": 5416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85239340433033e-05,
+      "loss": 0.3064,
+      "step": 5417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852330903301425e-05,
+      "loss": 0.2708,
+      "step": 5418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522683900978004e-05,
+      "loss": 0.307,
+      "step": 5419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522058647203494e-05,
+      "loss": 0.2884,
+      "step": 5420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852143327169964e-05,
+      "loss": 0.2928,
+      "step": 5421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520807774475386e-05,
+      "loss": 0.2914,
+      "step": 5422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852018215553966e-05,
+      "loss": 0.2585,
+      "step": 5423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519556414901404e-05,
+      "loss": 0.3142,
+      "step": 5424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851893055256955e-05,
+      "loss": 0.2936,
+      "step": 5425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8518304568553044e-05,
+      "loss": 0.2618,
+      "step": 5426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517678462860826e-05,
+      "loss": 0.2866,
+      "step": 5427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517052235501832e-05,
+      "loss": 0.2851,
+      "step": 5428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516425886485017e-05,
+      "loss": 0.2985,
+      "step": 5429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515799415819323e-05,
+      "loss": 0.2803,
+      "step": 5430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515172823513697e-05,
+      "loss": 0.322,
+      "step": 5431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514546109577094e-05,
+      "loss": 0.2936,
+      "step": 5432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513919274018463e-05,
+      "loss": 0.3036,
+      "step": 5433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513292316846758e-05,
+      "loss": 0.2974,
+      "step": 5434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512665238070933e-05,
+      "loss": 0.2831,
+      "step": 5435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512038037699948e-05,
+      "loss": 0.2783,
+      "step": 5436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511410715742757e-05,
+      "loss": 0.295,
+      "step": 5437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851078327220833e-05,
+      "loss": 0.2646,
+      "step": 5438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851015570710562e-05,
+      "loss": 0.2875,
+      "step": 5439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509528020443597e-05,
+      "loss": 0.3134,
+      "step": 5440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508900212231225e-05,
+      "loss": 0.2938,
+      "step": 5441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508272282477472e-05,
+      "loss": 0.2787,
+      "step": 5442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507644231191303e-05,
+      "loss": 0.2747,
+      "step": 5443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85070160583817e-05,
+      "loss": 0.2938,
+      "step": 5444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506387764057624e-05,
+      "loss": 0.2975,
+      "step": 5445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505759348228055e-05,
+      "loss": 0.281,
+      "step": 5446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505130810901966e-05,
+      "loss": 0.2769,
+      "step": 5447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504502152088344e-05,
+      "loss": 0.2844,
+      "step": 5448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850387337179616e-05,
+      "loss": 0.2665,
+      "step": 5449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503244470034398e-05,
+      "loss": 0.2922,
+      "step": 5450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850261544681204e-05,
+      "loss": 0.283,
+      "step": 5451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501986302138074e-05,
+      "loss": 0.2972,
+      "step": 5452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501357036021483e-05,
+      "loss": 0.2923,
+      "step": 5453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500727648471258e-05,
+      "loss": 0.2851,
+      "step": 5454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850009813949639e-05,
+      "loss": 0.2899,
+      "step": 5455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849946850910587e-05,
+      "loss": 0.3285,
+      "step": 5456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849883875730869e-05,
+      "loss": 0.3146,
+      "step": 5457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498208884113848e-05,
+      "loss": 0.2685,
+      "step": 5458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849757888953034e-05,
+      "loss": 0.2996,
+      "step": 5459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849694877356716e-05,
+      "loss": 0.2549,
+      "step": 5460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849631853623332e-05,
+      "loss": 0.2744,
+      "step": 5461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849568817753781e-05,
+      "loss": 0.279,
+      "step": 5462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495057697489644e-05,
+      "loss": 0.2791,
+      "step": 5463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849442709609782e-05,
+      "loss": 0.2987,
+      "step": 5464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493796373371352e-05,
+      "loss": 0.297,
+      "step": 5465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493165529319243e-05,
+      "loss": 0.3081,
+      "step": 5466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849253456395051e-05,
+      "loss": 0.2869,
+      "step": 5467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849190347727416e-05,
+      "loss": 0.3028,
+      "step": 5468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491272269299215e-05,
+      "loss": 0.2806,
+      "step": 5469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849064094003468e-05,
+      "loss": 0.2812,
+      "step": 5470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8490009489489585e-05,
+      "loss": 0.2795,
+      "step": 5471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489377917672944e-05,
+      "loss": 0.2781,
+      "step": 5472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488746224593778e-05,
+      "loss": 0.3064,
+      "step": 5473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488114410261108e-05,
+      "loss": 0.2866,
+      "step": 5474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8487482474683965e-05,
+      "loss": 0.2735,
+      "step": 5475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848685041787137e-05,
+      "loss": 0.3082,
+      "step": 5476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486218239832353e-05,
+      "loss": 0.2765,
+      "step": 5477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485585940575946e-05,
+      "loss": 0.2834,
+      "step": 5478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848495352011118e-05,
+      "loss": 0.2842,
+      "step": 5479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484320978447087e-05,
+      "loss": 0.2822,
+      "step": 5480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483688315592703e-05,
+      "loss": 0.2807,
+      "step": 5481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483055531557064e-05,
+      "loss": 0.3098,
+      "step": 5482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848242262634921e-05,
+      "loss": 0.2781,
+      "step": 5483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481789599978183e-05,
+      "loss": 0.2905,
+      "step": 5484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848115645245302e-05,
+      "loss": 0.3003,
+      "step": 5485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848052318378277e-05,
+      "loss": 0.2758,
+      "step": 5486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479889793976476e-05,
+      "loss": 0.264,
+      "step": 5487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847925628304319e-05,
+      "loss": 0.3444,
+      "step": 5488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478622650991952e-05,
+      "loss": 0.2643,
+      "step": 5489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477988897831823e-05,
+      "loss": 0.3046,
+      "step": 5490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477355023571847e-05,
+      "loss": 0.2685,
+      "step": 5491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476721028221084e-05,
+      "loss": 0.2846,
+      "step": 5492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476086911788588e-05,
+      "loss": 0.2907,
+      "step": 5493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475452674283415e-05,
+      "loss": 0.3179,
+      "step": 5494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474818315714625e-05,
+      "loss": 0.2906,
+      "step": 5495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474183836091284e-05,
+      "loss": 0.288,
+      "step": 5496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473549235422453e-05,
+      "loss": 0.3238,
+      "step": 5497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847291451371719e-05,
+      "loss": 0.2772,
+      "step": 5498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472279670984568e-05,
+      "loss": 0.2877,
+      "step": 5499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471644707233654e-05,
+      "loss": 0.2796,
+      "step": 5500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471009622473516e-05,
+      "loss": 0.2806,
+      "step": 5501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470374416713228e-05,
+      "loss": 0.2998,
+      "step": 5502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469739089961862e-05,
+      "loss": 0.257,
+      "step": 5503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469103642228492e-05,
+      "loss": 0.2935,
+      "step": 5504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468468073522196e-05,
+      "loss": 0.2735,
+      "step": 5505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467832383852056e-05,
+      "loss": 0.3076,
+      "step": 5506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467196573227146e-05,
+      "loss": 0.2947,
+      "step": 5507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846656064165655e-05,
+      "loss": 0.2773,
+      "step": 5508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465924589149353e-05,
+      "loss": 0.2879,
+      "step": 5509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846528841571464e-05,
+      "loss": 0.3016,
+      "step": 5510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464652121361493e-05,
+      "loss": 0.301,
+      "step": 5511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464015706099014e-05,
+      "loss": 0.3078,
+      "step": 5512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846337916993628e-05,
+      "loss": 0.2671,
+      "step": 5513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846274251288239e-05,
+      "loss": 0.3087,
+      "step": 5514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462105734946437e-05,
+      "loss": 0.3091,
+      "step": 5515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461468836137513e-05,
+      "loss": 0.2829,
+      "step": 5516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846083181646472e-05,
+      "loss": 0.2768,
+      "step": 5517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8460194675937157e-05,
+      "loss": 0.3052,
+      "step": 5518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459557414563926e-05,
+      "loss": 0.2961,
+      "step": 5519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458920032354123e-05,
+      "loss": 0.3084,
+      "step": 5520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845828252931686e-05,
+      "loss": 0.3252,
+      "step": 5521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8457644905461237e-05,
+      "loss": 0.2912,
+      "step": 5522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845700716079637e-05,
+      "loss": 0.291,
+      "step": 5523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845636929533136e-05,
+      "loss": 0.2974,
+      "step": 5524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845573130907532e-05,
+      "loss": 0.2941,
+      "step": 5525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455093202037368e-05,
+      "loss": 0.2661,
+      "step": 5526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454454974226615e-05,
+      "loss": 0.2886,
+      "step": 5527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453816625652175e-05,
+      "loss": 0.3072,
+      "step": 5528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453178156323174e-05,
+      "loss": 0.2841,
+      "step": 5529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8452539566248725e-05,
+      "loss": 0.2871,
+      "step": 5530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 0.2634,
+      "step": 5531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8451262023899973e-05,
+      "loss": 0.273,
+      "step": 5532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450623071643922e-05,
+      "loss": 0.3176,
+      "step": 5533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449983998678924e-05,
+      "loss": 0.3025,
+      "step": 5534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449344805014103e-05,
+      "loss": 0.2965,
+      "step": 5535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448705490658596e-05,
+      "loss": 0.3107,
+      "step": 5536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448066055621524e-05,
+      "loss": 0.2757,
+      "step": 5537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447426499912032e-05,
+      "loss": 0.2683,
+      "step": 5538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446786823539248e-05,
+      "loss": 0.3001,
+      "step": 5539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446147026512312e-05,
+      "loss": 0.2874,
+      "step": 5540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445507108840367e-05,
+      "loss": 0.3117,
+      "step": 5541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444867070532545e-05,
+      "loss": 0.2762,
+      "step": 5542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444226911597996e-05,
+      "loss": 0.2861,
+      "step": 5543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443586632045857e-05,
+      "loss": 0.3101,
+      "step": 5544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442946231885283e-05,
+      "loss": 0.3196,
+      "step": 5545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844230571112541e-05,
+      "loss": 0.2444,
+      "step": 5546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84416650697754e-05,
+      "loss": 0.2674,
+      "step": 5547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844102430784439e-05,
+      "loss": 0.304,
+      "step": 5548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844038342534155e-05,
+      "loss": 0.2625,
+      "step": 5549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439742422276015e-05,
+      "loss": 0.2903,
+      "step": 5550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439101298656953e-05,
+      "loss": 0.2652,
+      "step": 5551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438460054493523e-05,
+      "loss": 0.292,
+      "step": 5552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437818689794878e-05,
+      "loss": 0.2796,
+      "step": 5553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.843717720457018e-05,
+      "loss": 0.297,
+      "step": 5554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8436535598828598e-05,
+      "loss": 0.304,
+      "step": 5555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843589387257929e-05,
+      "loss": 0.3126,
+      "step": 5556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435252025831427e-05,
+      "loss": 0.2755,
+      "step": 5557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434610058594177e-05,
+      "loss": 0.3221,
+      "step": 5558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433967970876706e-05,
+      "loss": 0.2926,
+      "step": 5559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433325762688185e-05,
+      "loss": 0.2905,
+      "step": 5560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432683434037796e-05,
+      "loss": 0.2714,
+      "step": 5561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432040984934704e-05,
+      "loss": 0.2879,
+      "step": 5562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843139841538809e-05,
+      "loss": 0.2579,
+      "step": 5563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430755725407134e-05,
+      "loss": 0.2692,
+      "step": 5564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430112915001013e-05,
+      "loss": 0.2865,
+      "step": 5565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429469984178917e-05,
+      "loss": 0.2869,
+      "step": 5566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428826932950013e-05,
+      "loss": 0.2892,
+      "step": 5567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428183761323507e-05,
+      "loss": 0.2449,
+      "step": 5568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427540469308568e-05,
+      "loss": 0.3086,
+      "step": 5569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426897056914397e-05,
+      "loss": 0.2819,
+      "step": 5570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426253524150176e-05,
+      "loss": 0.2525,
+      "step": 5571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425609871025105e-05,
+      "loss": 0.2659,
+      "step": 5572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8424966097548374e-05,
+      "loss": 0.312,
+      "step": 5573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8424322203729177e-05,
+      "loss": 0.2748,
+      "step": 5574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423678189576715e-05,
+      "loss": 0.262,
+      "step": 5575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423034055100184e-05,
+      "loss": 0.29,
+      "step": 5576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422389800308785e-05,
+      "loss": 0.3182,
+      "step": 5577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421745425211725e-05,
+      "loss": 0.3042,
+      "step": 5578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421100929818206e-05,
+      "loss": 0.3301,
+      "step": 5579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842045631413743e-05,
+      "loss": 0.2766,
+      "step": 5580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419811578178608e-05,
+      "loss": 0.3111,
+      "step": 5581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841916672195095e-05,
+      "loss": 0.2749,
+      "step": 5582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418521745463665e-05,
+      "loss": 0.2877,
+      "step": 5583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841787664872597e-05,
+      "loss": 0.2578,
+      "step": 5584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417231431747074e-05,
+      "loss": 0.2888,
+      "step": 5585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84165860945362e-05,
+      "loss": 0.3065,
+      "step": 5586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415940637102557e-05,
+      "loss": 0.2759,
+      "step": 5587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415295059455373e-05,
+      "loss": 0.2729,
+      "step": 5588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414649361603866e-05,
+      "loss": 0.2752,
+      "step": 5589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841400354355726e-05,
+      "loss": 0.3068,
+      "step": 5590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841335760532478e-05,
+      "loss": 0.2787,
+      "step": 5591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841271154691565e-05,
+      "loss": 0.2506,
+      "step": 5592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84120653683391e-05,
+      "loss": 0.2916,
+      "step": 5593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411419069604363e-05,
+      "loss": 0.2752,
+      "step": 5594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410772650720665e-05,
+      "loss": 0.2891,
+      "step": 5595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410126111697245e-05,
+      "loss": 0.275,
+      "step": 5596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840947945254333e-05,
+      "loss": 0.2741,
+      "step": 5597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840883267326817e-05,
+      "loss": 0.2722,
+      "step": 5598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408185773880993e-05,
+      "loss": 0.2741,
+      "step": 5599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407538754391043e-05,
+      "loss": 0.3074,
+      "step": 5600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840689161480756e-05,
+      "loss": 0.3122,
+      "step": 5601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406244355139794e-05,
+      "loss": 0.2924,
+      "step": 5602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840559697539698e-05,
+      "loss": 0.2827,
+      "step": 5603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404949475588374e-05,
+      "loss": 0.2964,
+      "step": 5604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404301855723223e-05,
+      "loss": 0.2738,
+      "step": 5605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403654115810775e-05,
+      "loss": 0.2766,
+      "step": 5606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403006255860284e-05,
+      "loss": 0.2938,
+      "step": 5607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402358275881003e-05,
+      "loss": 0.2664,
+      "step": 5608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840171017588219e-05,
+      "loss": 0.2758,
+      "step": 5609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401061955873102e-05,
+      "loss": 0.2842,
+      "step": 5610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400413615862995e-05,
+      "loss": 0.2957,
+      "step": 5611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399765155861134e-05,
+      "loss": 0.2926,
+      "step": 5612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839911657587678e-05,
+      "loss": 0.3184,
+      "step": 5613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398467875919196e-05,
+      "loss": 0.3086,
+      "step": 5614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839781905599765e-05,
+      "loss": 0.2707,
+      "step": 5615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839717011612141e-05,
+      "loss": 0.2769,
+      "step": 5616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8396521056299742e-05,
+      "loss": 0.2931,
+      "step": 5617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395871876541924e-05,
+      "loss": 0.2807,
+      "step": 5618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839522257685722e-05,
+      "loss": 0.2604,
+      "step": 5619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839457315725491e-05,
+      "loss": 0.286,
+      "step": 5620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393923617744272e-05,
+      "loss": 0.2575,
+      "step": 5621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393273958334583e-05,
+      "loss": 0.2617,
+      "step": 5622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392624179035118e-05,
+      "loss": 0.2696,
+      "step": 5623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839197427985516e-05,
+      "loss": 0.2935,
+      "step": 5624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391324260804e-05,
+      "loss": 0.2731,
+      "step": 5625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390674121890913e-05,
+      "loss": 0.2813,
+      "step": 5626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839002386312519e-05,
+      "loss": 0.2933,
+      "step": 5627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389373484516117e-05,
+      "loss": 0.3028,
+      "step": 5628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388722986072988e-05,
+      "loss": 0.3059,
+      "step": 5629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838807236780509e-05,
+      "loss": 0.2903,
+      "step": 5630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387421629721725e-05,
+      "loss": 0.295,
+      "step": 5631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838677077183218e-05,
+      "loss": 0.3086,
+      "step": 5632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838611979414575e-05,
+      "loss": 0.2669,
+      "step": 5633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838546869667174e-05,
+      "loss": 0.271,
+      "step": 5634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838481747941945e-05,
+      "loss": 0.2893,
+      "step": 5635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384166142398178e-05,
+      "loss": 0.2514,
+      "step": 5636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838351468561723e-05,
+      "loss": 0.3134,
+      "step": 5637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8382863109085913e-05,
+      "loss": 0.2858,
+      "step": 5638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838221141281353e-05,
+      "loss": 0.2799,
+      "step": 5639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838155959680939e-05,
+      "loss": 0.2892,
+      "step": 5640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380907661082806e-05,
+      "loss": 0.2678,
+      "step": 5641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380255605643093e-05,
+      "loss": 0.2782,
+      "step": 5642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379603430499563e-05,
+      "loss": 0.3025,
+      "step": 5643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8378951135661526e-05,
+      "loss": 0.2888,
+      "step": 5644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837829872113831e-05,
+      "loss": 0.3182,
+      "step": 5645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377646186939224e-05,
+      "loss": 0.2802,
+      "step": 5646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376993533073593e-05,
+      "loss": 0.3404,
+      "step": 5647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837634075955074e-05,
+      "loss": 0.2999,
+      "step": 5648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375687866379988e-05,
+      "loss": 0.2961,
+      "step": 5649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375034853570665e-05,
+      "loss": 0.2587,
+      "step": 5650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8374381721132097e-05,
+      "loss": 0.2948,
+      "step": 5651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373728469073612e-05,
+      "loss": 0.2637,
+      "step": 5652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373075097404544e-05,
+      "loss": 0.2859,
+      "step": 5653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372421606134224e-05,
+      "loss": 0.3089,
+      "step": 5654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371767995271986e-05,
+      "loss": 0.3109,
+      "step": 5655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837111426482717e-05,
+      "loss": 0.2835,
+      "step": 5656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837046041480911e-05,
+      "loss": 0.3031,
+      "step": 5657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836980644522715e-05,
+      "loss": 0.3166,
+      "step": 5658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836915235609062e-05,
+      "loss": 0.2957,
+      "step": 5659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8368498147408876e-05,
+      "loss": 0.2728,
+      "step": 5660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836784381919126e-05,
+      "loss": 0.2839,
+      "step": 5661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836718937144711e-05,
+      "loss": 0.2424,
+      "step": 5662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366534804185786e-05,
+      "loss": 0.243,
+      "step": 5663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365880117416627e-05,
+      "loss": 0.296,
+      "step": 5664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365225311148995e-05,
+      "loss": 0.2964,
+      "step": 5665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364570385392237e-05,
+      "loss": 0.2832,
+      "step": 5666
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363915340155706e-05,
+      "loss": 0.281,
+      "step": 5667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363260175448765e-05,
+      "loss": 0.2531,
+      "step": 5668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362604891280766e-05,
+      "loss": 0.2926,
+      "step": 5669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361949487661074e-05,
+      "loss": 0.2819,
+      "step": 5670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836129396459905e-05,
+      "loss": 0.3042,
+      "step": 5671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360638322104057e-05,
+      "loss": 0.2532,
+      "step": 5672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359982560185458e-05,
+      "loss": 0.305,
+      "step": 5673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835932667885262e-05,
+      "loss": 0.2948,
+      "step": 5674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358670678114915e-05,
+      "loss": 0.2638,
+      "step": 5675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358014557981715e-05,
+      "loss": 0.263,
+      "step": 5676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357358318462382e-05,
+      "loss": 0.3214,
+      "step": 5677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83567019595663e-05,
+      "loss": 0.2898,
+      "step": 5678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356045481302845e-05,
+      "loss": 0.2811,
+      "step": 5679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355388883681384e-05,
+      "loss": 0.341,
+      "step": 5680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354732166711303e-05,
+      "loss": 0.3052,
+      "step": 5681
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354075330401985e-05,
+      "loss": 0.2967,
+      "step": 5682
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353418374762804e-05,
+      "loss": 0.2678,
+      "step": 5683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8352761299803153e-05,
+      "loss": 0.2353,
+      "step": 5684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8352104105532412e-05,
+      "loss": 0.3071,
+      "step": 5685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351446791959967e-05,
+      "loss": 0.2681,
+      "step": 5686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350789359095214e-05,
+      "loss": 0.303,
+      "step": 5687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350131806947537e-05,
+      "loss": 0.2825,
+      "step": 5688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349474135526332e-05,
+      "loss": 0.2656,
+      "step": 5689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834881634484099e-05,
+      "loss": 0.2766,
+      "step": 5690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834815843490091e-05,
+      "loss": 0.2683,
+      "step": 5691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347500405715494e-05,
+      "loss": 0.3008,
+      "step": 5692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346842257294132e-05,
+      "loss": 0.2983,
+      "step": 5693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346183989646225e-05,
+      "loss": 0.2753,
+      "step": 5694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8345525602781186e-05,
+      "loss": 0.2896,
+      "step": 5695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344867096708416e-05,
+      "loss": 0.2898,
+      "step": 5696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834420847143731e-05,
+      "loss": 0.2944,
+      "step": 5697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343549726977288e-05,
+      "loss": 0.2919,
+      "step": 5698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834289086333776e-05,
+      "loss": 0.2758,
+      "step": 5699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342231880528127e-05,
+      "loss": 0.2682,
+      "step": 5700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834157277855781e-05,
+      "loss": 0.287,
+      "step": 5701
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834091355743622e-05,
+      "loss": 0.2978,
+      "step": 5702
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340254217172775e-05,
+      "loss": 0.2531,
+      "step": 5703
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339594757776894e-05,
+      "loss": 0.2817,
+      "step": 5704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338935179257997e-05,
+      "loss": 0.2758,
+      "step": 5705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83382754816255e-05,
+      "loss": 0.297,
+      "step": 5706
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337615664888835e-05,
+      "loss": 0.2958,
+      "step": 5707
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336955729057424e-05,
+      "loss": 0.3033,
+      "step": 5708
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336295674140684e-05,
+      "loss": 0.2769,
+      "step": 5709
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8335635500148056e-05,
+      "loss": 0.2906,
+      "step": 5710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334975207088962e-05,
+      "loss": 0.2858,
+      "step": 5711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833431479497284e-05,
+      "loss": 0.2632,
+      "step": 5712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333654263809114e-05,
+      "loss": 0.2829,
+      "step": 5713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833299361360723e-05,
+      "loss": 0.2783,
+      "step": 5714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833233284437662e-05,
+      "loss": 0.2707,
+      "step": 5715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833167195612672e-05,
+      "loss": 0.3071,
+      "step": 5716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833101094886697e-05,
+      "loss": 0.2785,
+      "step": 5717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8330349822606818e-05,
+      "loss": 0.2897,
+      "step": 5718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83296885773557e-05,
+      "loss": 0.2783,
+      "step": 5719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832902721312307e-05,
+      "loss": 0.2734,
+      "step": 5720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328365729918364e-05,
+      "loss": 0.2928,
+      "step": 5721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832770412775104e-05,
+      "loss": 0.2891,
+      "step": 5722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832704240663054e-05,
+      "loss": 0.2709,
+      "step": 5723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832638056656633e-05,
+      "loss": 0.2751,
+      "step": 5724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832571860756785e-05,
+      "loss": 0.2529,
+      "step": 5725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325056529644557e-05,
+      "loss": 0.282,
+      "step": 5726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324394332805913e-05,
+      "loss": 0.2932,
+      "step": 5727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323732017061374e-05,
+      "loss": 0.2486,
+      "step": 5728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323069582420403e-05,
+      "loss": 0.2663,
+      "step": 5729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832240702889246e-05,
+      "loss": 0.2854,
+      "step": 5730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832174435648701e-05,
+      "loss": 0.2744,
+      "step": 5731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832108156521352e-05,
+      "loss": 0.2451,
+      "step": 5732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320418655081452e-05,
+      "loss": 0.2734,
+      "step": 5733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831975562610028e-05,
+      "loss": 0.2895,
+      "step": 5734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319092478279478e-05,
+      "loss": 0.2715,
+      "step": 5735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831842921162851e-05,
+      "loss": 0.3013,
+      "step": 5736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831776582615685e-05,
+      "loss": 0.329,
+      "step": 5737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317102321873984e-05,
+      "loss": 0.2609,
+      "step": 5738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831643869878938e-05,
+      "loss": 0.289,
+      "step": 5739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315774956912525e-05,
+      "loss": 0.2627,
+      "step": 5740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315111096252887e-05,
+      "loss": 0.289,
+      "step": 5741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314447116819966e-05,
+      "loss": 0.2669,
+      "step": 5742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313783018623233e-05,
+      "loss": 0.3145,
+      "step": 5743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313118801672177e-05,
+      "loss": 0.3042,
+      "step": 5744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831245446597629e-05,
+      "loss": 0.2667,
+      "step": 5745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311790011545055e-05,
+      "loss": 0.2793,
+      "step": 5746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831112543838797e-05,
+      "loss": 0.3393,
+      "step": 5747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831046074651452e-05,
+      "loss": 0.2918,
+      "step": 5748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309795935934203e-05,
+      "loss": 0.2622,
+      "step": 5749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309131006656522e-05,
+      "loss": 0.2781,
+      "step": 5750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308465958690965e-05,
+      "loss": 0.257,
+      "step": 5751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307800792047036e-05,
+      "loss": 0.2844,
+      "step": 5752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307135506734237e-05,
+      "loss": 0.2652,
+      "step": 5753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830647010276207e-05,
+      "loss": 0.2711,
+      "step": 5754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305804580140036e-05,
+      "loss": 0.2755,
+      "step": 5755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305138938877647e-05,
+      "loss": 0.2747,
+      "step": 5756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304473178984408e-05,
+      "loss": 0.2865,
+      "step": 5757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830380730046983e-05,
+      "loss": 0.2849,
+      "step": 5758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830314130334342e-05,
+      "loss": 0.2872,
+      "step": 5759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83024751876147e-05,
+      "loss": 0.2831,
+      "step": 5760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301808953293176e-05,
+      "loss": 0.2536,
+      "step": 5761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830114260038837e-05,
+      "loss": 0.2779,
+      "step": 5762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300476128909798e-05,
+      "loss": 0.2512,
+      "step": 5763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829980953886698e-05,
+      "loss": 0.3058,
+      "step": 5764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299142830269437e-05,
+      "loss": 0.2987,
+      "step": 5765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298476003126695e-05,
+      "loss": 0.2727,
+      "step": 5766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297809057448276e-05,
+      "loss": 0.2749,
+      "step": 5767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829714199324371e-05,
+      "loss": 0.2451,
+      "step": 5768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829647481052252e-05,
+      "loss": 0.2714,
+      "step": 5769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829580750929424e-05,
+      "loss": 0.2746,
+      "step": 5770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295140089568404e-05,
+      "loss": 0.3001,
+      "step": 5771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294472551354537e-05,
+      "loss": 0.2846,
+      "step": 5772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293804894662184e-05,
+      "loss": 0.2559,
+      "step": 5773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829313711950088e-05,
+      "loss": 0.2875,
+      "step": 5774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292469225880153e-05,
+      "loss": 0.3453,
+      "step": 5775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8291801213809557e-05,
+      "loss": 0.3219,
+      "step": 5776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8291133083298628e-05,
+      "loss": 0.28,
+      "step": 5777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290464834356908e-05,
+      "loss": 0.3081,
+      "step": 5778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828979646699394e-05,
+      "loss": 0.2824,
+      "step": 5779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828912798121928e-05,
+      "loss": 0.2842,
+      "step": 5780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828845937704247e-05,
+      "loss": 0.2851,
+      "step": 5781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287790654473062e-05,
+      "loss": 0.2949,
+      "step": 5782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828712181352061e-05,
+      "loss": 0.2906,
+      "step": 5783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8286452854194662e-05,
+      "loss": 0.2872,
+      "step": 5784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828578377650478e-05,
+      "loss": 0.2994,
+      "step": 5785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8285114580460515e-05,
+      "loss": 0.2874,
+      "step": 5786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8284445266071437e-05,
+      "loss": 0.279,
+      "step": 5787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828377583334709e-05,
+      "loss": 0.2703,
+      "step": 5788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828310628229705e-05,
+      "loss": 0.3013,
+      "step": 5789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828243661293087e-05,
+      "loss": 0.2932,
+      "step": 5790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8281766825258128e-05,
+      "loss": 0.2897,
+      "step": 5791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828109691928838e-05,
+      "loss": 0.2842,
+      "step": 5792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280426895031204e-05,
+      "loss": 0.2823,
+      "step": 5793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279756752496163e-05,
+      "loss": 0.295,
+      "step": 5794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279086491692832e-05,
+      "loss": 0.3083,
+      "step": 5795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8278416112630786e-05,
+      "loss": 0.2552,
+      "step": 5796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82777456153196e-05,
+      "loss": 0.2828,
+      "step": 5797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827707499976885e-05,
+      "loss": 0.3221,
+      "step": 5798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276404265988123e-05,
+      "loss": 0.2913,
+      "step": 5799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275733413986986e-05,
+      "loss": 0.3144,
+      "step": 5800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275062443775035e-05,
+      "loss": 0.2913,
+      "step": 5801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274391355361842e-05,
+      "loss": 0.293,
+      "step": 5802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273720148757003e-05,
+      "loss": 0.299,
+      "step": 5803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82730488239701e-05,
+      "loss": 0.2882,
+      "step": 5804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8272377381010726e-05,
+      "loss": 0.2818,
+      "step": 5805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827170581988847e-05,
+      "loss": 0.2664,
+      "step": 5806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271034140612922e-05,
+      "loss": 0.2768,
+      "step": 5807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270362343193677e-05,
+      "loss": 0.2901,
+      "step": 5808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269690427640333e-05,
+      "loss": 0.3043,
+      "step": 5809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826901839396249e-05,
+      "loss": 0.2712,
+      "step": 5810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268346242169744e-05,
+      "loss": 0.3047,
+      "step": 5811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267673972271696e-05,
+      "loss": 0.2735,
+      "step": 5812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826700158427795e-05,
+      "loss": 0.2989,
+      "step": 5813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266329078198106e-05,
+      "loss": 0.2883,
+      "step": 5814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826565645404178e-05,
+      "loss": 0.2617,
+      "step": 5815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826498371181857e-05,
+      "loss": 0.2788,
+      "step": 5816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264310851538086e-05,
+      "loss": 0.2773,
+      "step": 5817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8263637873209945e-05,
+      "loss": 0.3066,
+      "step": 5818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262964776843758e-05,
+      "loss": 0.2896,
+      "step": 5819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826229156244914e-05,
+      "loss": 0.306,
+      "step": 5820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261618230035704e-05,
+      "loss": 0.3162,
+      "step": 5821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260944779613065e-05,
+      "loss": 0.2864,
+      "step": 5822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260271211190853e-05,
+      "loss": 0.273,
+      "step": 5823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8259597524778682e-05,
+      "loss": 0.3193,
+      "step": 5824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825892372038618e-05,
+      "loss": 0.2928,
+      "step": 5825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258249798022962e-05,
+      "loss": 0.2721,
+      "step": 5826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257575757698664e-05,
+      "loss": 0.2824,
+      "step": 5827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825690159942291e-05,
+      "loss": 0.2758,
+      "step": 5828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825622732320533e-05,
+      "loss": 0.2696,
+      "step": 5829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255552929055556e-05,
+      "loss": 0.2788,
+      "step": 5830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254878416983217e-05,
+      "loss": 0.2921,
+      "step": 5831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254203786997956e-05,
+      "loss": 0.2686,
+      "step": 5832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253529039109402e-05,
+      "loss": 0.2973,
+      "step": 5833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252854173327197e-05,
+      "loss": 0.2809,
+      "step": 5834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825217918966098e-05,
+      "loss": 0.2852,
+      "step": 5835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251504088120392e-05,
+      "loss": 0.2974,
+      "step": 5836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8250828868715074e-05,
+      "loss": 0.3013,
+      "step": 5837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8250153531454677e-05,
+      "loss": 0.2924,
+      "step": 5838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249478076348845e-05,
+      "loss": 0.2708,
+      "step": 5839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824880250340722e-05,
+      "loss": 0.2968,
+      "step": 5840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824812681263946e-05,
+      "loss": 0.3007,
+      "step": 5841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824745100405521e-05,
+      "loss": 0.3338,
+      "step": 5842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246775077664134e-05,
+      "loss": 0.2869,
+      "step": 5843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246099033475872e-05,
+      "loss": 0.2979,
+      "step": 5844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245422871500094e-05,
+      "loss": 0.2835,
+      "step": 5845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824474659174645e-05,
+      "loss": 0.3008,
+      "step": 5846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244070194224602e-05,
+      "loss": 0.2704,
+      "step": 5847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243393678944212e-05,
+      "loss": 0.3067,
+      "step": 5848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242717045914946e-05,
+      "loss": 0.3391,
+      "step": 5849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242040295146464e-05,
+      "loss": 0.3,
+      "step": 5850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824136342664844e-05,
+      "loss": 0.2719,
+      "step": 5851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240686440430534e-05,
+      "loss": 0.2721,
+      "step": 5852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824000933650242e-05,
+      "loss": 0.295,
+      "step": 5853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823933211487377e-05,
+      "loss": 0.2953,
+      "step": 5854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238654775554256e-05,
+      "loss": 0.2901,
+      "step": 5855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237977318553558e-05,
+      "loss": 0.2882,
+      "step": 5856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823729974388135e-05,
+      "loss": 0.2755,
+      "step": 5857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236622051547302e-05,
+      "loss": 0.3138,
+      "step": 5858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823594424156111e-05,
+      "loss": 0.2699,
+      "step": 5859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823526631393244e-05,
+      "loss": 0.2948,
+      "step": 5860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823458826867099e-05,
+      "loss": 0.2762,
+      "step": 5861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823391010578644e-05,
+      "loss": 0.2727,
+      "step": 5862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8233231825288468e-05,
+      "loss": 0.2687,
+      "step": 5863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232553427186773e-05,
+      "loss": 0.3143,
+      "step": 5864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231874911491046e-05,
+      "loss": 0.2672,
+      "step": 5865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231196278210968e-05,
+      "loss": 0.2919,
+      "step": 5866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230517527356246e-05,
+      "loss": 0.2822,
+      "step": 5867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 0.2866,
+      "step": 5868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229159672961628e-05,
+      "loss": 0.2698,
+      "step": 5869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822848056944113e-05,
+      "loss": 0.2793,
+      "step": 5870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227801348384774e-05,
+      "loss": 0.2582,
+      "step": 5871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227122009802262e-05,
+      "loss": 0.2837,
+      "step": 5872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226442553703297e-05,
+      "loss": 0.2935,
+      "step": 5873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822576298009758e-05,
+      "loss": 0.2747,
+      "step": 5874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225083288994827e-05,
+      "loss": 0.2906,
+      "step": 5875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822440348040474e-05,
+      "loss": 0.3037,
+      "step": 5876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223723554337032e-05,
+      "loss": 0.2961,
+      "step": 5877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223043510801412e-05,
+      "loss": 0.2806,
+      "step": 5878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222363349807598e-05,
+      "loss": 0.306,
+      "step": 5879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221683071365303e-05,
+      "loss": 0.2701,
+      "step": 5880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822100267548424e-05,
+      "loss": 0.2729,
+      "step": 5881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220322162174144e-05,
+      "loss": 0.2724,
+      "step": 5882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219641531444713e-05,
+      "loss": 0.3225,
+      "step": 5883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218960783305685e-05,
+      "loss": 0.3158,
+      "step": 5884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821827991776678e-05,
+      "loss": 0.28,
+      "step": 5885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821759893483772e-05,
+      "loss": 0.3125,
+      "step": 5886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216917834528235e-05,
+      "loss": 0.2763,
+      "step": 5887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216236616848053e-05,
+      "loss": 0.2347,
+      "step": 5888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215555281806905e-05,
+      "loss": 0.2718,
+      "step": 5889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214873829414523e-05,
+      "loss": 0.2578,
+      "step": 5890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821419225968064e-05,
+      "loss": 0.2823,
+      "step": 5891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8213510572614996e-05,
+      "loss": 0.3085,
+      "step": 5892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212828768227326e-05,
+      "loss": 0.3005,
+      "step": 5893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212146846527364e-05,
+      "loss": 0.2857,
+      "step": 5894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211464807524855e-05,
+      "loss": 0.2975,
+      "step": 5895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821078265122954e-05,
+      "loss": 0.3279,
+      "step": 5896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210100377651165e-05,
+      "loss": 0.3,
+      "step": 5897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209417986799477e-05,
+      "loss": 0.2876,
+      "step": 5898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208735478684217e-05,
+      "loss": 0.2937,
+      "step": 5899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208052853315138e-05,
+      "loss": 0.3061,
+      "step": 5900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207370110701988e-05,
+      "loss": 0.3116,
+      "step": 5901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206687250854524e-05,
+      "loss": 0.3081,
+      "step": 5902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206004273782495e-05,
+      "loss": 0.2701,
+      "step": 5903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820532117949566e-05,
+      "loss": 0.3141,
+      "step": 5904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204637968003777e-05,
+      "loss": 0.2824,
+      "step": 5905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203954639316604e-05,
+      "loss": 0.2851,
+      "step": 5906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82032711934439e-05,
+      "loss": 0.2571,
+      "step": 5907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202587630395427e-05,
+      "loss": 0.2835,
+      "step": 5908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820190395018095e-05,
+      "loss": 0.2661,
+      "step": 5909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820122015281024e-05,
+      "loss": 0.2883,
+      "step": 5910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200536238293053e-05,
+      "loss": 0.285,
+      "step": 5911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819985220663917e-05,
+      "loss": 0.2722,
+      "step": 5912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199168057858353e-05,
+      "loss": 0.2676,
+      "step": 5913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819848379196038e-05,
+      "loss": 0.2553,
+      "step": 5914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197799408955023e-05,
+      "loss": 0.2596,
+      "step": 5915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197114908852055e-05,
+      "loss": 0.2661,
+      "step": 5916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196430291661256e-05,
+      "loss": 0.2621,
+      "step": 5917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819574555739241e-05,
+      "loss": 0.281,
+      "step": 5918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195060706055286e-05,
+      "loss": 0.3042,
+      "step": 5919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819437573765968e-05,
+      "loss": 0.2957,
+      "step": 5920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193690652215367e-05,
+      "loss": 0.2962,
+      "step": 5921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193005449732134e-05,
+      "loss": 0.2843,
+      "step": 5922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819232013021977e-05,
+      "loss": 0.3035,
+      "step": 5923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191634693688064e-05,
+      "loss": 0.2735,
+      "step": 5924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819094914014681e-05,
+      "loss": 0.2954,
+      "step": 5925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190263469605793e-05,
+      "loss": 0.2919,
+      "step": 5926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818957768207481e-05,
+      "loss": 0.2923,
+      "step": 5927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818889177756366e-05,
+      "loss": 0.2789,
+      "step": 5928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8188205756082138e-05,
+      "loss": 0.2844,
+      "step": 5929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818751961764004e-05,
+      "loss": 0.2884,
+      "step": 5930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186833362247176e-05,
+      "loss": 0.2399,
+      "step": 5931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818614698991334e-05,
+      "loss": 0.2922,
+      "step": 5932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8185460500648342e-05,
+      "loss": 0.2663,
+      "step": 5933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8184773894461983e-05,
+      "loss": 0.2822,
+      "step": 5934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8184087171364074e-05,
+      "loss": 0.2919,
+      "step": 5935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818340033136442e-05,
+      "loss": 0.3124,
+      "step": 5936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182713374472836e-05,
+      "loss": 0.2961,
+      "step": 5937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182026300699133e-05,
+      "loss": 0.2573,
+      "step": 5938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181339110053126e-05,
+      "loss": 0.2453,
+      "step": 5939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8180651802544625e-05,
+      "loss": 0.3223,
+      "step": 5940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179964378183457e-05,
+      "loss": 0.2567,
+      "step": 5941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179276836979437e-05,
+      "loss": 0.2949,
+      "step": 5942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178589178942384e-05,
+      "loss": 0.2457,
+      "step": 5943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177901404082123e-05,
+      "loss": 0.3141,
+      "step": 5944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177213512408477e-05,
+      "loss": 0.2604,
+      "step": 5945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8176525503931274e-05,
+      "loss": 0.2582,
+      "step": 5946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175837378660337e-05,
+      "loss": 0.2889,
+      "step": 5947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175149136605495e-05,
+      "loss": 0.3084,
+      "step": 5948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174460777776587e-05,
+      "loss": 0.2772,
+      "step": 5949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817377230218344e-05,
+      "loss": 0.2779,
+      "step": 5950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173083709835887e-05,
+      "loss": 0.2835,
+      "step": 5951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8172395000743765e-05,
+      "loss": 0.2804,
+      "step": 5952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171706174916912e-05,
+      "loss": 0.2752,
+      "step": 5953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817101723236517e-05,
+      "loss": 0.2725,
+      "step": 5954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170328173098373e-05,
+      "loss": 0.2912,
+      "step": 5955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816963899712637e-05,
+      "loss": 0.2678,
+      "step": 5956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168949704459002e-05,
+      "loss": 0.267,
+      "step": 5957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168260295106114e-05,
+      "loss": 0.2739,
+      "step": 5958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167570769077557e-05,
+      "loss": 0.2681,
+      "step": 5959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816688112638318e-05,
+      "loss": 0.2785,
+      "step": 5960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8166191367032828e-05,
+      "loss": 0.2704,
+      "step": 5961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816550149103636e-05,
+      "loss": 0.2898,
+      "step": 5962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164811498403628e-05,
+      "loss": 0.2982,
+      "step": 5963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816412138914449e-05,
+      "loss": 0.2896,
+      "step": 5964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81634311632688e-05,
+      "loss": 0.2844,
+      "step": 5965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162740820786416e-05,
+      "loss": 0.2839,
+      "step": 5966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162050361707206e-05,
+      "loss": 0.2869,
+      "step": 5967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161359786041025e-05,
+      "loss": 0.2863,
+      "step": 5968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816066909379774e-05,
+      "loss": 0.2699,
+      "step": 5969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159978284987215e-05,
+      "loss": 0.2868,
+      "step": 5970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159287359619325e-05,
+      "loss": 0.2843,
+      "step": 5971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815859631770393e-05,
+      "loss": 0.3093,
+      "step": 5972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157905159250906e-05,
+      "loss": 0.2909,
+      "step": 5973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157213884270123e-05,
+      "loss": 0.2946,
+      "step": 5974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815652249277146e-05,
+      "loss": 0.2628,
+      "step": 5975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815583098476479e-05,
+      "loss": 0.2689,
+      "step": 5976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155139360259984e-05,
+      "loss": 0.2674,
+      "step": 5977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815444761926693e-05,
+      "loss": 0.292,
+      "step": 5978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153755761795503e-05,
+      "loss": 0.3278,
+      "step": 5979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153063787855593e-05,
+      "loss": 0.2616,
+      "step": 5980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815237169745708e-05,
+      "loss": 0.3046,
+      "step": 5981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151679490609845e-05,
+      "loss": 0.288,
+      "step": 5982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150987167323784e-05,
+      "loss": 0.2589,
+      "step": 5983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150294727608778e-05,
+      "loss": 0.2991,
+      "step": 5984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149602171474724e-05,
+      "loss": 0.2944,
+      "step": 5985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814890949893151e-05,
+      "loss": 0.2914,
+      "step": 5986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148216709989034e-05,
+      "loss": 0.2808,
+      "step": 5987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147523804657195e-05,
+      "loss": 0.2868,
+      "step": 5988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146830782945878e-05,
+      "loss": 0.2591,
+      "step": 5989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146137644864997e-05,
+      "loss": 0.283,
+      "step": 5990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814544439042444e-05,
+      "loss": 0.2886,
+      "step": 5991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814475101963412e-05,
+      "loss": 0.2826,
+      "step": 5992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144057532503932e-05,
+      "loss": 0.3143,
+      "step": 5993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814336392904379e-05,
+      "loss": 0.279,
+      "step": 5994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142670209263596e-05,
+      "loss": 0.2837,
+      "step": 5995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141976373173258e-05,
+      "loss": 0.2787,
+      "step": 5996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141282420782694e-05,
+      "loss": 0.3091,
+      "step": 5997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814058835210181e-05,
+      "loss": 0.2975,
+      "step": 5998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813989416714052e-05,
+      "loss": 0.3101,
+      "step": 5999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139199865908742e-05,
+      "loss": 0.2742,
+      "step": 6000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8138505448416396e-05,
+      "loss": 0.2928,
+      "step": 6001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81378109146734e-05,
+      "loss": 0.298,
+      "step": 6002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137116264689668e-05,
+      "loss": 0.3382,
+      "step": 6003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813642149847513e-05,
+      "loss": 0.3021,
+      "step": 6004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135726616039705e-05,
+      "loss": 0.2675,
+      "step": 6005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135031617393326e-05,
+      "loss": 0.3164,
+      "step": 6006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813433650254591e-05,
+      "loss": 0.2943,
+      "step": 6007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133641271507396e-05,
+      "loss": 0.2736,
+      "step": 6008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813294592428771e-05,
+      "loss": 0.2606,
+      "step": 6009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132250460896782e-05,
+      "loss": 0.2736,
+      "step": 6010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813155488134455e-05,
+      "loss": 0.3019,
+      "step": 6011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813085918564095e-05,
+      "loss": 0.2771,
+      "step": 6012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130163373795918e-05,
+      "loss": 0.2787,
+      "step": 6013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812946744581939e-05,
+      "loss": 0.2821,
+      "step": 6014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812877140172131e-05,
+      "loss": 0.2785,
+      "step": 6015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812807524151162e-05,
+      "loss": 0.26,
+      "step": 6016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127378965200262e-05,
+      "loss": 0.2934,
+      "step": 6017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8126682572797185e-05,
+      "loss": 0.2757,
+      "step": 6018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125986064312336e-05,
+      "loss": 0.2892,
+      "step": 6019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812528943975566e-05,
+      "loss": 0.2585,
+      "step": 6020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124592699137113e-05,
+      "loss": 0.3244,
+      "step": 6021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812389584246664e-05,
+      "loss": 0.2719,
+      "step": 6022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81231988697542e-05,
+      "loss": 0.2952,
+      "step": 6023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122501781009752e-05,
+      "loss": 0.2807,
+      "step": 6024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121804576243247e-05,
+      "loss": 0.2781,
+      "step": 6025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121107255464646e-05,
+      "loss": 0.2967,
+      "step": 6026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120409818683908e-05,
+      "loss": 0.274,
+      "step": 6027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119712265911e-05,
+      "loss": 0.3119,
+      "step": 6028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119014597155882e-05,
+      "loss": 0.3107,
+      "step": 6029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118316812428517e-05,
+      "loss": 0.2755,
+      "step": 6030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811761891173888e-05,
+      "loss": 0.2713,
+      "step": 6031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811692089509693e-05,
+      "loss": 0.2672,
+      "step": 6032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116222762512647e-05,
+      "loss": 0.2725,
+      "step": 6033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115524513995996e-05,
+      "loss": 0.3105,
+      "step": 6034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114826149556958e-05,
+      "loss": 0.2614,
+      "step": 6035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114127669205496e-05,
+      "loss": 0.3004,
+      "step": 6036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81134290729516e-05,
+      "loss": 0.2769,
+      "step": 6037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112730360805247e-05,
+      "loss": 0.2944,
+      "step": 6038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811203153277641e-05,
+      "loss": 0.2865,
+      "step": 6039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811133258887508e-05,
+      "loss": 0.2643,
+      "step": 6040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811063352911123e-05,
+      "loss": 0.2601,
+      "step": 6041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8109934353494856e-05,
+      "loss": 0.2824,
+      "step": 6042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810923506203594e-05,
+      "loss": 0.2918,
+      "step": 6043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810853565474447e-05,
+      "loss": 0.2887,
+      "step": 6044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810783613163044e-05,
+      "loss": 0.2958,
+      "step": 6045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107136492703836e-05,
+      "loss": 0.3107,
+      "step": 6046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810643673797466e-05,
+      "loss": 0.2615,
+      "step": 6047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105736867452896e-05,
+      "loss": 0.2793,
+      "step": 6048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8105036881148553e-05,
+      "loss": 0.271,
+      "step": 6049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104336779071622e-05,
+      "loss": 0.3025,
+      "step": 6050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8103636561232107e-05,
+      "loss": 0.3138,
+      "step": 6051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102936227640007e-05,
+      "loss": 0.306,
+      "step": 6052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810223577830533e-05,
+      "loss": 0.2927,
+      "step": 6053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101535213238076e-05,
+      "loss": 0.2699,
+      "step": 6054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100834532448256e-05,
+      "loss": 0.266,
+      "step": 6055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100133735945874e-05,
+      "loss": 0.2736,
+      "step": 6056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099432823740946e-05,
+      "loss": 0.3372,
+      "step": 6057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098731795843475e-05,
+      "loss": 0.2866,
+      "step": 6058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098030652263485e-05,
+      "loss": 0.2885,
+      "step": 6059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097329393010988e-05,
+      "loss": 0.2904,
+      "step": 6060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8096628018095997e-05,
+      "loss": 0.2826,
+      "step": 6061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809592652752853e-05,
+      "loss": 0.2791,
+      "step": 6062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095224921318616e-05,
+      "loss": 0.2709,
+      "step": 6063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8094523199476266e-05,
+      "loss": 0.2701,
+      "step": 6064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809382136201151e-05,
+      "loss": 0.2751,
+      "step": 6065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809311940893437e-05,
+      "loss": 0.2734,
+      "step": 6066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092417340254876e-05,
+      "loss": 0.2829,
+      "step": 6067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809171515598305e-05,
+      "loss": 0.2749,
+      "step": 6068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8091012856128932e-05,
+      "loss": 0.2676,
+      "step": 6069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090310440702544e-05,
+      "loss": 0.2681,
+      "step": 6070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089607909713924e-05,
+      "loss": 0.3145,
+      "step": 6071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088905263173102e-05,
+      "loss": 0.2732,
+      "step": 6072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088202501090126e-05,
+      "loss": 0.294,
+      "step": 6073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087499623475023e-05,
+      "loss": 0.2773,
+      "step": 6074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086796630337837e-05,
+      "loss": 0.2917,
+      "step": 6075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808609352168861e-05,
+      "loss": 0.3262,
+      "step": 6076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8085390297537383e-05,
+      "loss": 0.2769,
+      "step": 6077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084686957894207e-05,
+      "loss": 0.2803,
+      "step": 6078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808398350276912e-05,
+      "loss": 0.2524,
+      "step": 6079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808327993217218e-05,
+      "loss": 0.2863,
+      "step": 6080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8082576246113425e-05,
+      "loss": 0.3161,
+      "step": 6081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081872444602916e-05,
+      "loss": 0.2743,
+      "step": 6082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081168527650702e-05,
+      "loss": 0.2817,
+      "step": 6083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080464495266838e-05,
+      "loss": 0.2559,
+      "step": 6084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079760347461382e-05,
+      "loss": 0.2516,
+      "step": 6085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807905608424439e-05,
+      "loss": 0.3012,
+      "step": 6086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8078351705625922e-05,
+      "loss": 0.263,
+      "step": 6087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807764721161604e-05,
+      "loss": 0.2785,
+      "step": 6088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076942602224807e-05,
+      "loss": 0.2746,
+      "step": 6089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807623787746229e-05,
+      "loss": 0.2712,
+      "step": 6090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075533037338554e-05,
+      "loss": 0.2674,
+      "step": 6091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8074828081863663e-05,
+      "loss": 0.2915,
+      "step": 6092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807412301104769e-05,
+      "loss": 0.2738,
+      "step": 6093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073417824900707e-05,
+      "loss": 0.2892,
+      "step": 6094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072712523432785e-05,
+      "loss": 0.2961,
+      "step": 6095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072007106654003e-05,
+      "loss": 0.2705,
+      "step": 6096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807130157457443e-05,
+      "loss": 0.285,
+      "step": 6097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070595927204147e-05,
+      "loss": 0.2803,
+      "step": 6098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069890164553234e-05,
+      "loss": 0.3083,
+      "step": 6099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069184286631774e-05,
+      "loss": 0.2927,
+      "step": 6100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068478293449846e-05,
+      "loss": 0.266,
+      "step": 6101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067772185017536e-05,
+      "loss": 0.2863,
+      "step": 6102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067065961344932e-05,
+      "loss": 0.2993,
+      "step": 6103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806635962244212e-05,
+      "loss": 0.2974,
+      "step": 6104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065653168319186e-05,
+      "loss": 0.3075,
+      "step": 6105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806494659898623e-05,
+      "loss": 0.3088,
+      "step": 6106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064239914453333e-05,
+      "loss": 0.2958,
+      "step": 6107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80635331147306e-05,
+      "loss": 0.28,
+      "step": 6108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062826199828117e-05,
+      "loss": 0.2862,
+      "step": 6109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062119169755994e-05,
+      "loss": 0.2485,
+      "step": 6110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806141202452432e-05,
+      "loss": 0.2764,
+      "step": 6111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060704764143196e-05,
+      "loss": 0.268,
+      "step": 6112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8059997388622733e-05,
+      "loss": 0.288,
+      "step": 6113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8059289897973026e-05,
+      "loss": 0.3085,
+      "step": 6114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058582292204186e-05,
+      "loss": 0.3139,
+      "step": 6115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057874571326316e-05,
+      "loss": 0.3092,
+      "step": 6116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057166735349533e-05,
+      "loss": 0.2733,
+      "step": 6117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056458784283938e-05,
+      "loss": 0.2895,
+      "step": 6118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055750718139653e-05,
+      "loss": 0.2856,
+      "step": 6119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055042536926783e-05,
+      "loss": 0.2997,
+      "step": 6120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054334240655447e-05,
+      "loss": 0.2817,
+      "step": 6121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053625829335767e-05,
+      "loss": 0.2829,
+      "step": 6122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805291730297786e-05,
+      "loss": 0.2753,
+      "step": 6123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052208661591843e-05,
+      "loss": 0.2881,
+      "step": 6124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8051499905187836e-05,
+      "loss": 0.2796,
+      "step": 6125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050791033775973e-05,
+      "loss": 0.2988,
+      "step": 6126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805008204736637e-05,
+      "loss": 0.2844,
+      "step": 6127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804937294596916e-05,
+      "loss": 0.2606,
+      "step": 6128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048663729594466e-05,
+      "loss": 0.305,
+      "step": 6129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047954398252427e-05,
+      "loss": 0.303,
+      "step": 6130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047244951953166e-05,
+      "loss": 0.2806,
+      "step": 6131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804653539070682e-05,
+      "loss": 0.2939,
+      "step": 6132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804582571452353e-05,
+      "loss": 0.2796,
+      "step": 6133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8045115923413427e-05,
+      "loss": 0.2578,
+      "step": 6134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044406017386648e-05,
+      "loss": 0.3017,
+      "step": 6135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043695996453338e-05,
+      "loss": 0.2779,
+      "step": 6136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042985860623637e-05,
+      "loss": 0.2845,
+      "step": 6137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804227560990769e-05,
+      "loss": 0.3173,
+      "step": 6138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8041565244315643e-05,
+      "loss": 0.261,
+      "step": 6139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040854763857635e-05,
+      "loss": 0.2877,
+      "step": 6140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040144168543826e-05,
+      "loss": 0.3031,
+      "step": 6141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803943345838436e-05,
+      "loss": 0.302,
+      "step": 6142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803872263338939e-05,
+      "loss": 0.286,
+      "step": 6143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038011693569065e-05,
+      "loss": 0.2574,
+      "step": 6144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037300638933545e-05,
+      "loss": 0.2889,
+      "step": 6145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036589469492988e-05,
+      "loss": 0.2973,
+      "step": 6146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803587818525755e-05,
+      "loss": 0.2839,
+      "step": 6147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035166786237388e-05,
+      "loss": 0.2633,
+      "step": 6148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803445527244267e-05,
+      "loss": 0.3125,
+      "step": 6149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033743643883554e-05,
+      "loss": 0.2693,
+      "step": 6150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803303190057021e-05,
+      "loss": 0.3213,
+      "step": 6151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80323200425128e-05,
+      "loss": 0.3083,
+      "step": 6152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031608069721495e-05,
+      "loss": 0.2778,
+      "step": 6153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803089598220646e-05,
+      "loss": 0.2697,
+      "step": 6154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803018377997787e-05,
+      "loss": 0.2769,
+      "step": 6155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8029471463045904e-05,
+      "loss": 0.3189,
+      "step": 6156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028759031420725e-05,
+      "loss": 0.2637,
+      "step": 6157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028046485112516e-05,
+      "loss": 0.2668,
+      "step": 6158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8027333824131456e-05,
+      "loss": 0.307,
+      "step": 6159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8026621048487725e-05,
+      "loss": 0.2871,
+      "step": 6160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025908158191502e-05,
+      "loss": 0.2849,
+      "step": 6161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025195153252964e-05,
+      "loss": 0.2645,
+      "step": 6162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802448203368231e-05,
+      "loss": 0.2724,
+      "step": 6163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023768799489714e-05,
+      "loss": 0.2809,
+      "step": 6164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023055450685368e-05,
+      "loss": 0.2513,
+      "step": 6165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8022341987279463e-05,
+      "loss": 0.28,
+      "step": 6166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021628409282187e-05,
+      "loss": 0.2922,
+      "step": 6167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020914716703735e-05,
+      "loss": 0.2881,
+      "step": 6168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80202009095543e-05,
+      "loss": 0.3029,
+      "step": 6169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019486987844078e-05,
+      "loss": 0.2861,
+      "step": 6170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801877295158327e-05,
+      "loss": 0.2877,
+      "step": 6171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8018058800782073e-05,
+      "loss": 0.2789,
+      "step": 6172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017344535450686e-05,
+      "loss": 0.3101,
+      "step": 6173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016630155599314e-05,
+      "loss": 0.3117,
+      "step": 6174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015915661238157e-05,
+      "loss": 0.2937,
+      "step": 6175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015201052377427e-05,
+      "loss": 0.2753,
+      "step": 6176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014486329027333e-05,
+      "loss": 0.2769,
+      "step": 6177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013771491198076e-05,
+      "loss": 0.2696,
+      "step": 6178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801305653889987e-05,
+      "loss": 0.2419,
+      "step": 6179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801234147214293e-05,
+      "loss": 0.2713,
+      "step": 6180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011626290937467e-05,
+      "loss": 0.3289,
+      "step": 6181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80109109952937e-05,
+      "loss": 0.2749,
+      "step": 6182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8010195585221845e-05,
+      "loss": 0.2678,
+      "step": 6183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009480060732113e-05,
+      "loss": 0.2924,
+      "step": 6184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008764421834738e-05,
+      "loss": 0.301,
+      "step": 6185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008048668539936e-05,
+      "loss": 0.2951,
+      "step": 6186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800733280085793e-05,
+      "loss": 0.2969,
+      "step": 6187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8006616818798946e-05,
+      "loss": 0.2914,
+      "step": 6188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005900722373208e-05,
+      "loss": 0.2902,
+      "step": 6189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005184511590954e-05,
+      "loss": 0.2697,
+      "step": 6190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8004468186462403e-05,
+      "loss": 0.305,
+      "step": 6191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003751746997795e-05,
+      "loss": 0.2871,
+      "step": 6192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003035193207364e-05,
+      "loss": 0.2759,
+      "step": 6193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800231852510134e-05,
+      "loss": 0.2921,
+      "step": 6194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800160174268996e-05,
+      "loss": 0.2478,
+      "step": 6195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8000884845983465e-05,
+      "loss": 0.3159,
+      "step": 6196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.80001678349921e-05,
+      "loss": 0.2572,
+      "step": 6197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999450709726097e-05,
+      "loss": 0.304,
+      "step": 6198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998733470195705e-05,
+      "loss": 0.2888,
+      "step": 6199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799801611641117e-05,
+      "loss": 0.2864,
+      "step": 6200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7997298648382738e-05,
+      "loss": 0.2697,
+      "step": 6201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996581066120653e-05,
+      "loss": 0.2791,
+      "step": 6202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799586336963517e-05,
+      "loss": 0.2591,
+      "step": 6203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7995145558936537e-05,
+      "loss": 0.2717,
+      "step": 6204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 0.2842,
+      "step": 6205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7993709594940846e-05,
+      "loss": 0.2786,
+      "step": 6206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992991441664298e-05,
+      "loss": 0.2585,
+      "step": 6207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992273174215622e-05,
+      "loss": 0.2714,
+      "step": 6208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799155479260508e-05,
+      "loss": 0.2553,
+      "step": 6209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990836296842936e-05,
+      "loss": 0.2883,
+      "step": 6210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990117686939453e-05,
+      "loss": 0.282,
+      "step": 6211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989398962904888e-05,
+      "loss": 0.3095,
+      "step": 6212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988680124749516e-05,
+      "loss": 0.2799,
+      "step": 6213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79879611724836e-05,
+      "loss": 0.29,
+      "step": 6214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987242106117414e-05,
+      "loss": 0.2976,
+      "step": 6215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798652292566122e-05,
+      "loss": 0.2774,
+      "step": 6216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985803631125304e-05,
+      "loss": 0.291,
+      "step": 6217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985084222519934e-05,
+      "loss": 0.2561,
+      "step": 6218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798436469985538e-05,
+      "loss": 0.2965,
+      "step": 6219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983645063141927e-05,
+      "loss": 0.2719,
+      "step": 6220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982925312389856e-05,
+      "loss": 0.2405,
+      "step": 6221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798220544760944e-05,
+      "loss": 0.2595,
+      "step": 6222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981485468810967e-05,
+      "loss": 0.2746,
+      "step": 6223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980765376004722e-05,
+      "loss": 0.2538,
+      "step": 6224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798004516920099e-05,
+      "loss": 0.2564,
+      "step": 6225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979324848410055e-05,
+      "loss": 0.2979,
+      "step": 6226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978604413642213e-05,
+      "loss": 0.2645,
+      "step": 6227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977883864907744e-05,
+      "loss": 0.281,
+      "step": 6228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977163202216953e-05,
+      "loss": 0.2677,
+      "step": 6229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797644242558012e-05,
+      "loss": 0.2799,
+      "step": 6230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975721535007557e-05,
+      "loss": 0.2865,
+      "step": 6231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797500053050955e-05,
+      "loss": 0.2605,
+      "step": 6232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974279412096397e-05,
+      "loss": 0.2555,
+      "step": 6233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79735581797784e-05,
+      "loss": 0.2816,
+      "step": 6234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797283683356587e-05,
+      "loss": 0.292,
+      "step": 6235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972115373469096e-05,
+      "loss": 0.2848,
+      "step": 6236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971393799498396e-05,
+      "loss": 0.2925,
+      "step": 6237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970672111664073e-05,
+      "loss": 0.2748,
+      "step": 6238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969950309976432e-05,
+      "loss": 0.2765,
+      "step": 6239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969228394445785e-05,
+      "loss": 0.259,
+      "step": 6240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968506365082448e-05,
+      "loss": 0.2978,
+      "step": 6241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796778422189673e-05,
+      "loss": 0.2865,
+      "step": 6242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7967061964898947e-05,
+      "loss": 0.2881,
+      "step": 6243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966339594099413e-05,
+      "loss": 0.3005,
+      "step": 6244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965617109508454e-05,
+      "loss": 0.2702,
+      "step": 6245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964894511136382e-05,
+      "loss": 0.2781,
+      "step": 6246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964171798993525e-05,
+      "loss": 0.2721,
+      "step": 6247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79634489730902e-05,
+      "loss": 0.2464,
+      "step": 6248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796272603343674e-05,
+      "loss": 0.2717,
+      "step": 6249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796200298004346e-05,
+      "loss": 0.2712,
+      "step": 6250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79612798129207e-05,
+      "loss": 0.2775,
+      "step": 6251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796055653207878e-05,
+      "loss": 0.2663,
+      "step": 6252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795983313752804e-05,
+      "loss": 0.2933,
+      "step": 6253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795910962927881e-05,
+      "loss": 0.3026,
+      "step": 6254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795838600734142e-05,
+      "loss": 0.2766,
+      "step": 6255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795766227172621e-05,
+      "loss": 0.2844,
+      "step": 6256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956938422443518e-05,
+      "loss": 0.293,
+      "step": 6257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956214459503686e-05,
+      "loss": 0.235,
+      "step": 6258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955490382917046e-05,
+      "loss": 0.2821,
+      "step": 6259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954766192693948e-05,
+      "loss": 0.276,
+      "step": 6260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795404188884474e-05,
+      "loss": 0.2869,
+      "step": 6261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795331747137976e-05,
+      "loss": 0.2897,
+      "step": 6262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795259294030936e-05,
+      "loss": 0.26,
+      "step": 6263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951868295643886e-05,
+      "loss": 0.2512,
+      "step": 6264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951143537393693e-05,
+      "loss": 0.2748,
+      "step": 6265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950418665569128e-05,
+      "loss": 0.2798,
+      "step": 6266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949693680180555e-05,
+      "loss": 0.2776,
+      "step": 6267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948968581238318e-05,
+      "loss": 0.2646,
+      "step": 6268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794824336875278e-05,
+      "loss": 0.2826,
+      "step": 6269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79475180427343e-05,
+      "loss": 0.2589,
+      "step": 6270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794679260319324e-05,
+      "loss": 0.2952,
+      "step": 6271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946067050139957e-05,
+      "loss": 0.256,
+      "step": 6272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945341383584818e-05,
+      "loss": 0.2938,
+      "step": 6273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794461560353819e-05,
+      "loss": 0.2875,
+      "step": 6274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794388971001044e-05,
+      "loss": 0.2637,
+      "step": 6275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943163703011932e-05,
+      "loss": 0.2867,
+      "step": 6276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7942437582553043e-05,
+      "loss": 0.2909,
+      "step": 6277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794171134864414e-05,
+      "loss": 0.2758,
+      "step": 6278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940985001295597e-05,
+      "loss": 0.2735,
+      "step": 6279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794025854051779e-05,
+      "loss": 0.2528,
+      "step": 6280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939531966321097e-05,
+      "loss": 0.3134,
+      "step": 6281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7938805278715895e-05,
+      "loss": 0.2699,
+      "step": 6282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7938078477712562e-05,
+      "loss": 0.252,
+      "step": 6283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937351563321486e-05,
+      "loss": 0.2978,
+      "step": 6284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936624535553043e-05,
+      "loss": 0.2689,
+      "step": 6285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935897394417626e-05,
+      "loss": 0.2918,
+      "step": 6286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935170139925613e-05,
+      "loss": 0.3219,
+      "step": 6287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934442772087398e-05,
+      "loss": 0.2544,
+      "step": 6288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933715290913363e-05,
+      "loss": 0.2854,
+      "step": 6289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793298769641391e-05,
+      "loss": 0.2625,
+      "step": 6290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932259988599426e-05,
+      "loss": 0.2798,
+      "step": 6291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931532167480305e-05,
+      "loss": 0.2556,
+      "step": 6292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793080423306695e-05,
+      "loss": 0.3234,
+      "step": 6293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930076185369746e-05,
+      "loss": 0.29,
+      "step": 6294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929348024399105e-05,
+      "loss": 0.3297,
+      "step": 6295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792861975016542e-05,
+      "loss": 0.2824,
+      "step": 6296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927891362679098e-05,
+      "loss": 0.284,
+      "step": 6297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927162861950545e-05,
+      "loss": 0.2898,
+      "step": 6298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792643424799016e-05,
+      "loss": 0.2418,
+      "step": 6299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925705520808356e-05,
+      "loss": 0.2816,
+      "step": 6300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924976680415542e-05,
+      "loss": 0.3047,
+      "step": 6301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924247726822125e-05,
+      "loss": 0.2501,
+      "step": 6302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923518660038523e-05,
+      "loss": 0.2911,
+      "step": 6303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922789480075144e-05,
+      "loss": 0.2491,
+      "step": 6304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922060186942407e-05,
+      "loss": 0.2786,
+      "step": 6305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792133078065073e-05,
+      "loss": 0.2674,
+      "step": 6306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792060126121053e-05,
+      "loss": 0.2728,
+      "step": 6307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791987162863223e-05,
+      "loss": 0.2787,
+      "step": 6308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919141882926244e-05,
+      "loss": 0.2804,
+      "step": 6309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918412024103008e-05,
+      "loss": 0.3017,
+      "step": 6310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791768205217294e-05,
+      "loss": 0.302,
+      "step": 6311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791695196714647e-05,
+      "loss": 0.2984,
+      "step": 6312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791622176903402e-05,
+      "loss": 0.3155,
+      "step": 6313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7915491457846025e-05,
+      "loss": 0.286,
+      "step": 6314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914761033592923e-05,
+      "loss": 0.2655,
+      "step": 6315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914030496285132e-05,
+      "loss": 0.2869,
+      "step": 6316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913299845933103e-05,
+      "loss": 0.2756,
+      "step": 6317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912569082547264e-05,
+      "loss": 0.2757,
+      "step": 6318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7911838206138055e-05,
+      "loss": 0.3,
+      "step": 6319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791110721671592e-05,
+      "loss": 0.3142,
+      "step": 6320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791037611429129e-05,
+      "loss": 0.2625,
+      "step": 6321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909644898874615e-05,
+      "loss": 0.2865,
+      "step": 6322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790891357047634e-05,
+      "loss": 0.2661,
+      "step": 6323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7908182129106913e-05,
+      "loss": 0.2855,
+      "step": 6324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907450574776776e-05,
+      "loss": 0.3085,
+      "step": 6325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906718907496383e-05,
+      "loss": 0.329,
+      "step": 6326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905987127276186e-05,
+      "loss": 0.29,
+      "step": 6327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905255234126632e-05,
+      "loss": 0.2929,
+      "step": 6328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790452322805818e-05,
+      "loss": 0.2695,
+      "step": 6329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903791109081287e-05,
+      "loss": 0.2879,
+      "step": 6330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903058877206412e-05,
+      "loss": 0.3019,
+      "step": 6331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902326532444007e-05,
+      "loss": 0.3241,
+      "step": 6332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901594074804537e-05,
+      "loss": 0.2786,
+      "step": 6333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900861504298463e-05,
+      "loss": 0.2696,
+      "step": 6334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900128820936253e-05,
+      "loss": 0.2844,
+      "step": 6335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899396024728367e-05,
+      "loss": 0.2649,
+      "step": 6336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789866311568528e-05,
+      "loss": 0.2898,
+      "step": 6337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897930093817453e-05,
+      "loss": 0.2776,
+      "step": 6338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789719695913536e-05,
+      "loss": 0.3014,
+      "step": 6339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789646371164947e-05,
+      "loss": 0.2952,
+      "step": 6340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7895730351370267e-05,
+      "loss": 0.2575,
+      "step": 6341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894996878308212e-05,
+      "loss": 0.2718,
+      "step": 6342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894263292473793e-05,
+      "loss": 0.2853,
+      "step": 6343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789352959387748e-05,
+      "loss": 0.299,
+      "step": 6344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892795782529764e-05,
+      "loss": 0.2913,
+      "step": 6345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892061858441117e-05,
+      "loss": 0.2742,
+      "step": 6346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7891327821622023e-05,
+      "loss": 0.2648,
+      "step": 6347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890593672082973e-05,
+      "loss": 0.3159,
+      "step": 6348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889859409834453e-05,
+      "loss": 0.2635,
+      "step": 6349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889125034886942e-05,
+      "loss": 0.2716,
+      "step": 6350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888390547250944e-05,
+      "loss": 0.2571,
+      "step": 6351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887655946936938e-05,
+      "loss": 0.277,
+      "step": 6352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788692123395542e-05,
+      "loss": 0.3235,
+      "step": 6353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788618640831689e-05,
+      "loss": 0.275,
+      "step": 6354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788545147003184e-05,
+      "loss": 0.2756,
+      "step": 6355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884716419110773e-05,
+      "loss": 0.284,
+      "step": 6356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883981255564177e-05,
+      "loss": 0.2856,
+      "step": 6357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883245979402565e-05,
+      "loss": 0.2944,
+      "step": 6358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882510590636434e-05,
+      "loss": 0.2797,
+      "step": 6359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7881775089276287e-05,
+      "loss": 0.2848,
+      "step": 6360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7881039475332637e-05,
+      "loss": 0.2865,
+      "step": 6361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7880303748815984e-05,
+      "loss": 0.2801,
+      "step": 6362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787956790973684e-05,
+      "loss": 0.2658,
+      "step": 6363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7878831958105716e-05,
+      "loss": 0.2753,
+      "step": 6364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787809589393312e-05,
+      "loss": 0.2527,
+      "step": 6365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787735971722958e-05,
+      "loss": 0.3023,
+      "step": 6366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7876623428005593e-05,
+      "loss": 0.2836,
+      "step": 6367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7875887026271686e-05,
+      "loss": 0.2828,
+      "step": 6368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787515051203838e-05,
+      "loss": 0.2582,
+      "step": 6369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787441388531619e-05,
+      "loss": 0.2929,
+      "step": 6370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787367714611564e-05,
+      "loss": 0.281,
+      "step": 6371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872940294447252e-05,
+      "loss": 0.2812,
+      "step": 6372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872203330321555e-05,
+      "loss": 0.2755,
+      "step": 6373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7871466253749072e-05,
+      "loss": 0.2589,
+      "step": 6374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870729064740335e-05,
+      "loss": 0.3002,
+      "step": 6375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786999176330587e-05,
+      "loss": 0.2828,
+      "step": 6376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786925434945621e-05,
+      "loss": 0.2622,
+      "step": 6377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868516823201893e-05,
+      "loss": 0.2711,
+      "step": 6378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7867779184553447e-05,
+      "loss": 0.3212,
+      "step": 6379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786704143352141e-05,
+      "loss": 0.3068,
+      "step": 6380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866303570116322e-05,
+      "loss": 0.2607,
+      "step": 6381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865565594348723e-05,
+      "loss": 0.2681,
+      "step": 6382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864827506229152e-05,
+      "loss": 0.2656,
+      "step": 6383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864089305768154e-05,
+      "loss": 0.2989,
+      "step": 6384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863350992976272e-05,
+      "loss": 0.2943,
+      "step": 6385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862612567864054e-05,
+      "loss": 0.2671,
+      "step": 6386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861874030442044e-05,
+      "loss": 0.2999,
+      "step": 6387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861135380720795e-05,
+      "loss": 0.2909,
+      "step": 6388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860396618710858e-05,
+      "loss": 0.2953,
+      "step": 6389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785965774442278e-05,
+      "loss": 0.2788,
+      "step": 6390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858918757867122e-05,
+      "loss": 0.2837,
+      "step": 6391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785817965905444e-05,
+      "loss": 0.2994,
+      "step": 6392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785744044799528e-05,
+      "loss": 0.2533,
+      "step": 6393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7856701124700215e-05,
+      "loss": 0.3002,
+      "step": 6394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7855961689179796e-05,
+      "loss": 0.3022,
+      "step": 6395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7855222141444594e-05,
+      "loss": 0.2818,
+      "step": 6396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7854482481505163e-05,
+      "loss": 0.3083,
+      "step": 6397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785374270937207e-05,
+      "loss": 0.2674,
+      "step": 6398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785300282505589e-05,
+      "loss": 0.3005,
+      "step": 6399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852262828567186e-05,
+      "loss": 0.2857,
+      "step": 6400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7851522719916524e-05,
+      "loss": 0.2763,
+      "step": 6401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850782499114485e-05,
+      "loss": 0.3031,
+      "step": 6402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850042166171637e-05,
+      "loss": 0.2737,
+      "step": 6403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784930172109855e-05,
+      "loss": 0.2867,
+      "step": 6404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7848561163905812e-05,
+      "loss": 0.2813,
+      "step": 6405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847820494603995e-05,
+      "loss": 0.279,
+      "step": 6406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847079713203676e-05,
+      "loss": 0.2607,
+      "step": 6407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846338819715436e-05,
+      "loss": 0.2988,
+      "step": 6408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845597814149864e-05,
+      "loss": 0.2693,
+      "step": 6409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844856696517543e-05,
+      "loss": 0.2999,
+      "step": 6410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844115466829058e-05,
+      "loss": 0.2943,
+      "step": 6411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843374125094994e-05,
+      "loss": 0.2863,
+      "step": 6412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7842632671325942e-05,
+      "loss": 0.3193,
+      "step": 6413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841891105532494e-05,
+      "loss": 0.3104,
+      "step": 6414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841149427725243e-05,
+      "loss": 0.2748,
+      "step": 6415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7840407637914785e-05,
+      "loss": 0.259,
+      "step": 6416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783966573611171e-05,
+      "loss": 0.3149,
+      "step": 6417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783892372232662e-05,
+      "loss": 0.2929,
+      "step": 6418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7838181596570115e-05,
+      "loss": 0.2568,
+      "step": 6419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783743935885279e-05,
+      "loss": 0.2692,
+      "step": 6420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783669700918525e-05,
+      "loss": 0.3007,
+      "step": 6421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835954547578097e-05,
+      "loss": 0.2663,
+      "step": 6422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835211974041945e-05,
+      "loss": 0.2842,
+      "step": 6423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834469288587387e-05,
+      "loss": 0.301,
+      "step": 6424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833726491225044e-05,
+      "loss": 0.2685,
+      "step": 6425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832983581965517e-05,
+      "loss": 0.3325,
+      "step": 6426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783224056081942e-05,
+      "loss": 0.2787,
+      "step": 6427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831497427797373e-05,
+      "loss": 0.2585,
+      "step": 6428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830754182909985e-05,
+      "loss": 0.3081,
+      "step": 6429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783001082616787e-05,
+      "loss": 0.281,
+      "step": 6430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829267357581656e-05,
+      "loss": 0.2848,
+      "step": 6431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782852377716195e-05,
+      "loss": 0.2723,
+      "step": 6432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827780084919384e-05,
+      "loss": 0.2844,
+      "step": 6433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827036280864574e-05,
+      "loss": 0.2799,
+      "step": 6434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826292365008148e-05,
+      "loss": 0.2799,
+      "step": 6435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782554833736073e-05,
+      "loss": 0.3138,
+      "step": 6436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824804197932948e-05,
+      "loss": 0.2871,
+      "step": 6437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824059946735436e-05,
+      "loss": 0.2642,
+      "step": 6438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823315583778818e-05,
+      "loss": 0.2738,
+      "step": 6439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782257110907373e-05,
+      "loss": 0.2863,
+      "step": 6440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7821826522630805e-05,
+      "loss": 0.2768,
+      "step": 6441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782108182446068e-05,
+      "loss": 0.2887,
+      "step": 6442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7820337014573992e-05,
+      "loss": 0.3029,
+      "step": 6443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781959209298138e-05,
+      "loss": 0.2809,
+      "step": 6444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818847059693486e-05,
+      "loss": 0.2947,
+      "step": 6445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818101914720943e-05,
+      "loss": 0.2826,
+      "step": 6446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781735665807441e-05,
+      "loss": 0.3008,
+      "step": 6447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781661128976452e-05,
+      "loss": 0.2905,
+      "step": 6448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815865809801923e-05,
+      "loss": 0.2908,
+      "step": 6449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815120218197272e-05,
+      "loss": 0.2636,
+      "step": 6450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781437451496121e-05,
+      "loss": 0.323,
+      "step": 6451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813628700104395e-05,
+      "loss": 0.301,
+      "step": 6452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812882773637476e-05,
+      "loss": 0.2937,
+      "step": 6453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781213673557111e-05,
+      "loss": 0.2675,
+      "step": 6454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781139058591595e-05,
+      "loss": 0.2354,
+      "step": 6455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781064432468266e-05,
+      "loss": 0.268,
+      "step": 6456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7809897951881893e-05,
+      "loss": 0.2864,
+      "step": 6457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7809151467524316e-05,
+      "loss": 0.2757,
+      "step": 6458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7808404871620587e-05,
+      "loss": 0.2885,
+      "step": 6459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807658164181373e-05,
+      "loss": 0.2856,
+      "step": 6460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780691134521734e-05,
+      "loss": 0.2714,
+      "step": 6461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7806164414739156e-05,
+      "loss": 0.2706,
+      "step": 6462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780541737275749e-05,
+      "loss": 0.2814,
+      "step": 6463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780467021928301e-05,
+      "loss": 0.265,
+      "step": 6464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7803922954326395e-05,
+      "loss": 0.2926,
+      "step": 6465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780317557789831e-05,
+      "loss": 0.2912,
+      "step": 6466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802428090009435e-05,
+      "loss": 0.2736,
+      "step": 6467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780168049067045e-05,
+      "loss": 0.2921,
+      "step": 6468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780093277989203e-05,
+      "loss": 0.2704,
+      "step": 6469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780018495768485e-05,
+      "loss": 0.2862,
+      "step": 6470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799437024059607e-05,
+      "loss": 0.2673,
+      "step": 6471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798688979026972e-05,
+      "loss": 0.2702,
+      "step": 6472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7797940822597636e-05,
+      "loss": 0.3097,
+      "step": 6473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779719255478228e-05,
+      "loss": 0.2759,
+      "step": 6474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7796444175591595e-05,
+      "loss": 0.289,
+      "step": 6475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7795695685036276e-05,
+      "loss": 0.269,
+      "step": 6476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794947083127007e-05,
+      "loss": 0.2733,
+      "step": 6477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794198369874488e-05,
+      "loss": 0.2824,
+      "step": 6478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793449545289405e-05,
+      "loss": 0.2873,
+      "step": 6479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779270060938246e-05,
+      "loss": 0.2802,
+      "step": 6480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779195156216435e-05,
+      "loss": 0.2761,
+      "step": 6481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7791202403645774e-05,
+      "loss": 0.2432,
+      "step": 6482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790453133837436e-05,
+      "loss": 0.288,
+      "step": 6483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778970375275003e-05,
+      "loss": 0.2797,
+      "step": 6484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788954260394272e-05,
+      "loss": 0.3027,
+      "step": 6485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778820465678086e-05,
+      "loss": 0.3089,
+      "step": 6486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77874549419205e-05,
+      "loss": 0.268,
+      "step": 6487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778670511582391e-05,
+      "loss": 0.2789,
+      "step": 6488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778595517850179e-05,
+      "loss": 0.2777,
+      "step": 6489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785205129964857e-05,
+      "loss": 0.2982,
+      "step": 6490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784454970223826e-05,
+      "loss": 0.2882,
+      "step": 6491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778370469928941e-05,
+      "loss": 0.2538,
+      "step": 6492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782954317172324e-05,
+      "loss": 0.2656,
+      "step": 6493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778220382388329e-05,
+      "loss": 0.2672,
+      "step": 6494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778145321943303e-05,
+      "loss": 0.2748,
+      "step": 6495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7780702503832263e-05,
+      "loss": 0.2489,
+      "step": 6496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779951677091712e-05,
+      "loss": 0.308,
+      "step": 6497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779200739222103e-05,
+      "loss": 0.292,
+      "step": 6498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778449690234155e-05,
+      "loss": 0.304,
+      "step": 6499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777698530138608e-05,
+      "loss": 0.2952,
+      "step": 6500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776947258946186e-05,
+      "loss": 0.3044,
+      "step": 6501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776195876667615e-05,
+      "loss": 0.294,
+      "step": 6502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777544438331364e-05,
+      "loss": 0.2375,
+      "step": 6503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774692778894984e-05,
+      "loss": 0.2555,
+      "step": 6504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777394106342239e-05,
+      "loss": 0.2706,
+      "step": 6505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777318923690659e-05,
+      "loss": 0.2604,
+      "step": 6506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7772437299358324e-05,
+      "loss": 0.3011,
+      "step": 6507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771685250788336e-05,
+      "loss": 0.2915,
+      "step": 6508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777093309120737e-05,
+      "loss": 0.2447,
+      "step": 6509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7770180820626163e-05,
+      "loss": 0.2618,
+      "step": 6510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769428439055465e-05,
+      "loss": 0.3104,
+      "step": 6511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768675946506023e-05,
+      "loss": 0.2782,
+      "step": 6512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767923342988583e-05,
+      "loss": 0.3017,
+      "step": 6513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77671706285139e-05,
+      "loss": 0.2863,
+      "step": 6514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776641780309272e-05,
+      "loss": 0.2542,
+      "step": 6515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77656648667358e-05,
+      "loss": 0.3075,
+      "step": 6516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764911819453897e-05,
+      "loss": 0.3098,
+      "step": 6517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776415866125776e-05,
+      "loss": 0.3001,
+      "step": 6518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763405392158156e-05,
+      "loss": 0.3059,
+      "step": 6519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7762652012165837e-05,
+      "loss": 0.2995,
+      "step": 6520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776189852129157e-05,
+      "loss": 0.2744,
+      "step": 6521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761144919546115e-05,
+      "loss": 0.3114,
+      "step": 6522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760391206940234e-05,
+      "loss": 0.3093,
+      "step": 6523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77596373834847e-05,
+      "loss": 0.3124,
+      "step": 6524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775888344919028e-05,
+      "loss": 0.2873,
+      "step": 6525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775812940406773e-05,
+      "loss": 0.2866,
+      "step": 6526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757375248127838e-05,
+      "loss": 0.2595,
+      "step": 6527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775662098138137e-05,
+      "loss": 0.2559,
+      "step": 6528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77558666038391e-05,
+      "loss": 0.29,
+      "step": 6529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755112115511802e-05,
+      "loss": 0.2657,
+      "step": 6530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775435751641025e-05,
+      "loss": 0.296,
+      "step": 6531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775360280654523e-05,
+      "loss": 0.2637,
+      "step": 6532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775284798592752e-05,
+      "loss": 0.2965,
+      "step": 6533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752093054567895e-05,
+      "loss": 0.2463,
+      "step": 6534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775133801247715e-05,
+      "loss": 0.2824,
+      "step": 6535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7750582859666064e-05,
+      "loss": 0.3064,
+      "step": 6536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774982759614542e-05,
+      "loss": 0.2666,
+      "step": 6537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749072221926016e-05,
+      "loss": 0.2717,
+      "step": 6538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7748316737018632e-05,
+      "loss": 0.2927,
+      "step": 6539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747561141434064e-05,
+      "loss": 0.2671,
+      "step": 6540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746805435183107e-05,
+      "loss": 0.27,
+      "step": 6541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746049618276545e-05,
+      "loss": 0.2861,
+      "step": 6542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774529369072519e-05,
+      "loss": 0.2631,
+      "step": 6543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744537652539828e-05,
+      "loss": 0.2505,
+      "step": 6544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743781503731262e-05,
+      "loss": 0.2556,
+      "step": 6545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743025244310293e-05,
+      "loss": 0.2765,
+      "step": 6546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7742268874287723e-05,
+      "loss": 0.2679,
+      "step": 6547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7741512393674357e-05,
+      "loss": 0.285,
+      "step": 6548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740755802481e-05,
+      "loss": 0.2743,
+      "step": 6549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773999910071846e-05,
+      "loss": 0.2563,
+      "step": 6550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739242288397543e-05,
+      "loss": 0.2615,
+      "step": 6551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773848536552906e-05,
+      "loss": 0.2632,
+      "step": 6552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737728332123826e-05,
+      "loss": 0.3096,
+      "step": 6553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736971188192652e-05,
+      "loss": 0.2631,
+      "step": 6554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773621393374635e-05,
+      "loss": 0.282,
+      "step": 6555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735456568795746e-05,
+      "loss": 0.3036,
+      "step": 6556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773469909335165e-05,
+      "loss": 0.2406,
+      "step": 6557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733941507424884e-05,
+      "loss": 0.2682,
+      "step": 6558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733183811026266e-05,
+      "loss": 0.2636,
+      "step": 6559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732426004166627e-05,
+      "loss": 0.2983,
+      "step": 6560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773166808685679e-05,
+      "loss": 0.2989,
+      "step": 6561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773091005910757e-05,
+      "loss": 0.3155,
+      "step": 6562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7730151920929807e-05,
+      "loss": 0.283,
+      "step": 6563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729393672334327e-05,
+      "loss": 0.2821,
+      "step": 6564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772863531333196e-05,
+      "loss": 0.2836,
+      "step": 6565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727876843933533e-05,
+      "loss": 0.3061,
+      "step": 6566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727118264149892e-05,
+      "loss": 0.2876,
+      "step": 6567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772635957399186e-05,
+      "loss": 0.3156,
+      "step": 6568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725600773470286e-05,
+      "loss": 0.2629,
+      "step": 6569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724841862596e-05,
+      "loss": 0.2939,
+      "step": 6570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724082841379845e-05,
+      "loss": 0.262,
+      "step": 6571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723323709832664e-05,
+      "loss": 0.2512,
+      "step": 6572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77225644679653e-05,
+      "loss": 0.2798,
+      "step": 6573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721805115788595e-05,
+      "loss": 0.3065,
+      "step": 6574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77210456533134e-05,
+      "loss": 0.2998,
+      "step": 6575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772028608055056e-05,
+      "loss": 0.2879,
+      "step": 6576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771952639751093e-05,
+      "loss": 0.2926,
+      "step": 6577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771876660420535e-05,
+      "loss": 0.2963,
+      "step": 6578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718006700644687e-05,
+      "loss": 0.2997,
+      "step": 6579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717246686839785e-05,
+      "loss": 0.2639,
+      "step": 6580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716486562801508e-05,
+      "loss": 0.2887,
+      "step": 6581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715726328540707e-05,
+      "loss": 0.2702,
+      "step": 6582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771496598406824e-05,
+      "loss": 0.2973,
+      "step": 6583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7714205529394973e-05,
+      "loss": 0.263,
+      "step": 6584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771344496453177e-05,
+      "loss": 0.2998,
+      "step": 6585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7712684289489492e-05,
+      "loss": 0.2517,
+      "step": 6586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711923504279004e-05,
+      "loss": 0.2856,
+      "step": 6587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771116260891117e-05,
+      "loss": 0.2724,
+      "step": 6588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710401603396864e-05,
+      "loss": 0.3006,
+      "step": 6589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7709640487746955e-05,
+      "loss": 0.2745,
+      "step": 6590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708879261972313e-05,
+      "loss": 0.2656,
+      "step": 6591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708117926083812e-05,
+      "loss": 0.2921,
+      "step": 6592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707356480092327e-05,
+      "loss": 0.3143,
+      "step": 6593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706594924008738e-05,
+      "loss": 0.3014,
+      "step": 6594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770583325784392e-05,
+      "loss": 0.2969,
+      "step": 6595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705071481608746e-05,
+      "loss": 0.2646,
+      "step": 6596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704309595314112e-05,
+      "loss": 0.3056,
+      "step": 6597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703547598970888e-05,
+      "loss": 0.2484,
+      "step": 6598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702785492589962e-05,
+      "loss": 0.2701,
+      "step": 6599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770202327618222e-05,
+      "loss": 0.238,
+      "step": 6600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701260949758555e-05,
+      "loss": 0.2659,
+      "step": 6601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700498513329853e-05,
+      "loss": 0.2748,
+      "step": 6602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699735966907e-05,
+      "loss": 0.2628,
+      "step": 6603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769897331050089e-05,
+      "loss": 0.2704,
+      "step": 6604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698210544122422e-05,
+      "loss": 0.2573,
+      "step": 6605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697447667782483e-05,
+      "loss": 0.2484,
+      "step": 6606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7696684681491978e-05,
+      "loss": 0.2814,
+      "step": 6607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76959215852618e-05,
+      "loss": 0.282,
+      "step": 6608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7695158379102855e-05,
+      "loss": 0.2634,
+      "step": 6609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769439506302604e-05,
+      "loss": 0.2773,
+      "step": 6610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693631637042256e-05,
+      "loss": 0.2979,
+      "step": 6611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769286810116241e-05,
+      "loss": 0.282,
+      "step": 6612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692104455397415e-05,
+      "loss": 0.2776,
+      "step": 6613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7691340699758167e-05,
+      "loss": 0.2915,
+      "step": 6614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690576834255585e-05,
+      "loss": 0.3156,
+      "step": 6615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689812858900578e-05,
+      "loss": 0.2718,
+      "step": 6616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689048773704052e-05,
+      "loss": 0.2649,
+      "step": 6617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688284578676932e-05,
+      "loss": 0.3131,
+      "step": 6618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768752027383013e-05,
+      "loss": 0.2666,
+      "step": 6619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686755859174556e-05,
+      "loss": 0.2673,
+      "step": 6620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768599133472114e-05,
+      "loss": 0.3176,
+      "step": 6621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685226700480793e-05,
+      "loss": 0.2651,
+      "step": 6622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684461956464444e-05,
+      "loss": 0.2674,
+      "step": 6623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683697102683012e-05,
+      "loss": 0.3093,
+      "step": 6624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7682932139147426e-05,
+      "loss": 0.2694,
+      "step": 6625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768216706586861e-05,
+      "loss": 0.2674,
+      "step": 6626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681401882857493e-05,
+      "loss": 0.2716,
+      "step": 6627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680636590125005e-05,
+      "loss": 0.2875,
+      "step": 6628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679871187682075e-05,
+      "loss": 0.2902,
+      "step": 6629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679105675539647e-05,
+      "loss": 0.2743,
+      "step": 6630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678340053708637e-05,
+      "loss": 0.2792,
+      "step": 6631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677574322199997e-05,
+      "loss": 0.2696,
+      "step": 6632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767680848102466e-05,
+      "loss": 0.3025,
+      "step": 6633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767604253019356e-05,
+      "loss": 0.27,
+      "step": 6634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7675276469717646e-05,
+      "loss": 0.2555,
+      "step": 6635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674510299607855e-05,
+      "loss": 0.3155,
+      "step": 6636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767374401987513e-05,
+      "loss": 0.3196,
+      "step": 6637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672977630530423e-05,
+      "loss": 0.2842,
+      "step": 6638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672211131584678e-05,
+      "loss": 0.2666,
+      "step": 6639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767144452304884e-05,
+      "loss": 0.2718,
+      "step": 6640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670677804933865e-05,
+      "loss": 0.3049,
+      "step": 6641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76699109772507e-05,
+      "loss": 0.2636,
+      "step": 6642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7669144040010303e-05,
+      "loss": 0.2807,
+      "step": 6643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766837699322362e-05,
+      "loss": 0.2803,
+      "step": 6644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766760983690162e-05,
+      "loss": 0.2871,
+      "step": 6645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666842571055257e-05,
+      "loss": 0.2936,
+      "step": 6646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666075195695485e-05,
+      "loss": 0.2998,
+      "step": 6647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766530771083327e-05,
+      "loss": 0.2674,
+      "step": 6648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664540116479573e-05,
+      "loss": 0.2615,
+      "step": 6649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766377241264536e-05,
+      "loss": 0.2978,
+      "step": 6650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663004599341592e-05,
+      "loss": 0.2622,
+      "step": 6651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7662236676579246e-05,
+      "loss": 0.3115,
+      "step": 6652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7661468644369282e-05,
+      "loss": 0.3248,
+      "step": 6653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7660700502722675e-05,
+      "loss": 0.2871,
+      "step": 6654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659932251650396e-05,
+      "loss": 0.2753,
+      "step": 6655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765916389116342e-05,
+      "loss": 0.3006,
+      "step": 6656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765839542127272e-05,
+      "loss": 0.269,
+      "step": 6657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657626841989276e-05,
+      "loss": 0.2668,
+      "step": 6658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765685815332406e-05,
+      "loss": 0.2556,
+      "step": 6659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656089355288063e-05,
+      "loss": 0.3142,
+      "step": 6660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7655320447892254e-05,
+      "loss": 0.2704,
+      "step": 6661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654551431147625e-05,
+      "loss": 0.2874,
+      "step": 6662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653782305065158e-05,
+      "loss": 0.3197,
+      "step": 6663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653013069655837e-05,
+      "loss": 0.2715,
+      "step": 6664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652243724930652e-05,
+      "loss": 0.2693,
+      "step": 6665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651474270900595e-05,
+      "loss": 0.2715,
+      "step": 6666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765070470757665e-05,
+      "loss": 0.3182,
+      "step": 6667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649935034969817e-05,
+      "loss": 0.2885,
+      "step": 6668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649165253091084e-05,
+      "loss": 0.283,
+      "step": 6669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764839536195145e-05,
+      "loss": 0.268,
+      "step": 6670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647625361561914e-05,
+      "loss": 0.3091,
+      "step": 6671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764685525193347e-05,
+      "loss": 0.2747,
+      "step": 6672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646085033077117e-05,
+      "loss": 0.273,
+      "step": 6673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7645314705003862e-05,
+      "loss": 0.3011,
+      "step": 6674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644544267724706e-05,
+      "loss": 0.2934,
+      "step": 6675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643773721250653e-05,
+      "loss": 0.27,
+      "step": 6676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643003065592715e-05,
+      "loss": 0.2528,
+      "step": 6677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642232300761894e-05,
+      "loss": 0.2802,
+      "step": 6678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76414614267692e-05,
+      "loss": 0.2739,
+      "step": 6679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640690443625647e-05,
+      "loss": 0.3041,
+      "step": 6680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639919351342245e-05,
+      "loss": 0.2693,
+      "step": 6681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639148149930012e-05,
+      "loss": 0.2747,
+      "step": 6682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763837683939996e-05,
+      "loss": 0.2845,
+      "step": 6683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763760541976311e-05,
+      "loss": 0.2915,
+      "step": 6684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763683389103048e-05,
+      "loss": 0.2785,
+      "step": 6685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636062253213088e-05,
+      "loss": 0.28,
+      "step": 6686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7635290506321957e-05,
+      "loss": 0.2747,
+      "step": 6687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763451865036811e-05,
+      "loss": 0.2915,
+      "step": 6688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633746685362573e-05,
+      "loss": 0.273,
+      "step": 6689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632974611316374e-05,
+      "loss": 0.2854,
+      "step": 6690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632202428240545e-05,
+      "loss": 0.2765,
+      "step": 6691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631430136146107e-05,
+      "loss": 0.3024,
+      "step": 6692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76306577350441e-05,
+      "loss": 0.2593,
+      "step": 6693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629885224945548e-05,
+      "loss": 0.3149,
+      "step": 6694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629112605861495e-05,
+      "loss": 0.2566,
+      "step": 6695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628339877802973e-05,
+      "loss": 0.2879,
+      "step": 6696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762756704078101e-05,
+      "loss": 0.2788,
+      "step": 6697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762679409480667e-05,
+      "loss": 0.2969,
+      "step": 6698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626021039890967e-05,
+      "loss": 0.2907,
+      "step": 6699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625247876044954e-05,
+      "loss": 0.2878,
+      "step": 6700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624474603279683e-05,
+      "loss": 0.2414,
+      "step": 6701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762370122160619e-05,
+      "loss": 0.2735,
+      "step": 6702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762292773103552e-05,
+      "loss": 0.2987,
+      "step": 6703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622154131578728e-05,
+      "loss": 0.2932,
+      "step": 6704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621380423246862e-05,
+      "loss": 0.3024,
+      "step": 6705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762060660605097e-05,
+      "loss": 0.2837,
+      "step": 6706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7619832680002114e-05,
+      "loss": 0.3003,
+      "step": 6707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761905864511134e-05,
+      "loss": 0.2579,
+      "step": 6708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761828450138971e-05,
+      "loss": 0.2671,
+      "step": 6709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7617510248848277e-05,
+      "loss": 0.2621,
+      "step": 6710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616735887498106e-05,
+      "loss": 0.3007,
+      "step": 6711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7615961417350252e-05,
+      "loss": 0.3098,
+      "step": 6712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761518683841578e-05,
+      "loss": 0.3003,
+      "step": 6713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614412150705757e-05,
+      "loss": 0.2786,
+      "step": 6714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761363735423125e-05,
+      "loss": 0.2849,
+      "step": 6715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612862449003317e-05,
+      "loss": 0.279,
+      "step": 6716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612087435033033e-05,
+      "loss": 0.2809,
+      "step": 6717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761131231233147e-05,
+      "loss": 0.2721,
+      "step": 6718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610537080909696e-05,
+      "loss": 0.3203,
+      "step": 6719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760976174077879e-05,
+      "loss": 0.3446,
+      "step": 6720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7608986291949816e-05,
+      "loss": 0.2918,
+      "step": 6721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7608210734433862e-05,
+      "loss": 0.2509,
+      "step": 6722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607435068242e-05,
+      "loss": 0.3011,
+      "step": 6723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760665929338531e-05,
+      "loss": 0.2907,
+      "step": 6724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7605883409874876e-05,
+      "loss": 0.2861,
+      "step": 6725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760510741772178e-05,
+      "loss": 0.3163,
+      "step": 6726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7604331316937108e-05,
+      "loss": 0.2594,
+      "step": 6727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760355510753194e-05,
+      "loss": 0.2895,
+      "step": 6728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7602778789517364e-05,
+      "loss": 0.2548,
+      "step": 6729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7602002362904477e-05,
+      "loss": 0.2713,
+      "step": 6730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760122582770436e-05,
+      "loss": 0.269,
+      "step": 6731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760044918392811e-05,
+      "loss": 0.312,
+      "step": 6732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599672431586824e-05,
+      "loss": 0.2587,
+      "step": 6733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598895570691587e-05,
+      "loss": 0.2925,
+      "step": 6734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598118601253506e-05,
+      "loss": 0.2927,
+      "step": 6735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597341523283674e-05,
+      "loss": 0.2692,
+      "step": 6736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759656433679319e-05,
+      "loss": 0.2948,
+      "step": 6737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7595787041793157e-05,
+      "loss": 0.2904,
+      "step": 6738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759500963829468e-05,
+      "loss": 0.2658,
+      "step": 6739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594232126308858e-05,
+      "loss": 0.256,
+      "step": 6740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593454505846807e-05,
+      "loss": 0.3109,
+      "step": 6741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7592676776919622e-05,
+      "loss": 0.2574,
+      "step": 6742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759189893953842e-05,
+      "loss": 0.2593,
+      "step": 6743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759112099371431e-05,
+      "loss": 0.2845,
+      "step": 6744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590342939458406e-05,
+      "loss": 0.2551,
+      "step": 6745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589564776781817e-05,
+      "loss": 0.2678,
+      "step": 6746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758878650569566e-05,
+      "loss": 0.2657,
+      "step": 6747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588008126211057e-05,
+      "loss": 0.3106,
+      "step": 6748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587229638339116e-05,
+      "loss": 0.2917,
+      "step": 6749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7586451042090966e-05,
+      "loss": 0.2656,
+      "step": 6750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758567233747773e-05,
+      "loss": 0.3215,
+      "step": 6751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7584893524510522e-05,
+      "loss": 0.2915,
+      "step": 6752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758411460320047e-05,
+      "loss": 0.2668,
+      "step": 6753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583335573558705e-05,
+      "loss": 0.2569,
+      "step": 6754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582556435596352e-05,
+      "loss": 0.2762,
+      "step": 6755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7581777189324537e-05,
+      "loss": 0.2879,
+      "step": 6756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580997834754393e-05,
+      "loss": 0.2588,
+      "step": 6757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580218371897057e-05,
+      "loss": 0.2759,
+      "step": 6758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579438800763654e-05,
+      "loss": 0.2938,
+      "step": 6759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7578659121365323e-05,
+      "loss": 0.2893,
+      "step": 6760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577879333713203e-05,
+      "loss": 0.2947,
+      "step": 6761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757709943781843e-05,
+      "loss": 0.2945,
+      "step": 6762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7576319433692146e-05,
+      "loss": 0.2883,
+      "step": 6763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575539321345494e-05,
+      "loss": 0.2605,
+      "step": 6764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757475910078961e-05,
+      "loss": 0.286,
+      "step": 6765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757397877203565e-05,
+      "loss": 0.2858,
+      "step": 6766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7573198335094753e-05,
+      "loss": 0.2526,
+      "step": 6767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572417789978066e-05,
+      "loss": 0.2771,
+      "step": 6768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571637136696736e-05,
+      "loss": 0.2419,
+      "step": 6769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570856375261926e-05,
+      "loss": 0.2806,
+      "step": 6770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570075505684773e-05,
+      "loss": 0.2926,
+      "step": 6771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756929452797644e-05,
+      "loss": 0.2918,
+      "step": 6772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568513442148087e-05,
+      "loss": 0.3122,
+      "step": 6773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567732248210857e-05,
+      "loss": 0.3131,
+      "step": 6774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756695094617592e-05,
+      "loss": 0.2902,
+      "step": 6775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566169536054433e-05,
+      "loss": 0.2542,
+      "step": 6776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565388017857555e-05,
+      "loss": 0.2985,
+      "step": 6777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564606391596455e-05,
+      "loss": 0.2559,
+      "step": 6778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7563824657282292e-05,
+      "loss": 0.2959,
+      "step": 6779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7563042814926237e-05,
+      "loss": 0.2942,
+      "step": 6780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562260864539455e-05,
+      "loss": 0.2529,
+      "step": 6781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561478806133112e-05,
+      "loss": 0.3164,
+      "step": 6782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560696639718386e-05,
+      "loss": 0.2593,
+      "step": 6783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559914365306444e-05,
+      "loss": 0.2756,
+      "step": 6784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559131982908468e-05,
+      "loss": 0.2772,
+      "step": 6785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558349492535623e-05,
+      "loss": 0.2999,
+      "step": 6786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7557566894199092e-05,
+      "loss": 0.2991,
+      "step": 6787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556784187910052e-05,
+      "loss": 0.3265,
+      "step": 6788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556001373679683e-05,
+      "loss": 0.2614,
+      "step": 6789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755521845151917e-05,
+      "loss": 0.2848,
+      "step": 6790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7554435421439696e-05,
+      "loss": 0.2544,
+      "step": 6791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755365228345244e-05,
+      "loss": 0.2759,
+      "step": 6792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7552869037568592e-05,
+      "loss": 0.2789,
+      "step": 6793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755208568379934e-05,
+      "loss": 0.2715,
+      "step": 6794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7551302222155876e-05,
+      "loss": 0.2844,
+      "step": 6795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550518652649386e-05,
+      "loss": 0.3078,
+      "step": 6796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549734975291068e-05,
+      "loss": 0.305,
+      "step": 6797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754895119009211e-05,
+      "loss": 0.2672,
+      "step": 6798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548167297063713e-05,
+      "loss": 0.2743,
+      "step": 6799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547383296217072e-05,
+      "loss": 0.2863,
+      "step": 6800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546599187563384e-05,
+      "loss": 0.2894,
+      "step": 6801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754581497111385e-05,
+      "loss": 0.2732,
+      "step": 6802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7545030646879673e-05,
+      "loss": 0.2986,
+      "step": 6803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544246214872055e-05,
+      "loss": 0.2745,
+      "step": 6804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7543461675102205e-05,
+      "loss": 0.2681,
+      "step": 6805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754267702758132e-05,
+      "loss": 0.2678,
+      "step": 6806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541892272320622e-05,
+      "loss": 0.2675,
+      "step": 6807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541107409331307e-05,
+      "loss": 0.2805,
+      "step": 6808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7540322438624596e-05,
+      "loss": 0.2738,
+      "step": 6809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539537360211693e-05,
+      "loss": 0.2685,
+      "step": 6810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753875217410382e-05,
+      "loss": 0.2676,
+      "step": 6811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753796688031218e-05,
+      "loss": 0.282,
+      "step": 6812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7537181478848007e-05,
+      "loss": 0.2921,
+      "step": 6813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7536395969722512e-05,
+      "loss": 0.3016,
+      "step": 6814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7535610352946912e-05,
+      "loss": 0.2968,
+      "step": 6815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753482462853243e-05,
+      "loss": 0.2716,
+      "step": 6816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534038796490294e-05,
+      "loss": 0.2665,
+      "step": 6817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533252856831725e-05,
+      "loss": 0.2692,
+      "step": 6818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753246680956795e-05,
+      "loss": 0.2652,
+      "step": 6819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7531680654710195e-05,
+      "loss": 0.272,
+      "step": 6820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753089439226969e-05,
+      "loss": 0.2642,
+      "step": 6821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530108022257673e-05,
+      "loss": 0.2653,
+      "step": 6822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752932154468537e-05,
+      "loss": 0.254,
+      "step": 6823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7528534959564015e-05,
+      "loss": 0.2821,
+      "step": 6824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752774826690484e-05,
+      "loss": 0.2884,
+      "step": 6825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526961466719098e-05,
+      "loss": 0.2612,
+      "step": 6826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752617455901801e-05,
+      "loss": 0.2694,
+      "step": 6827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752538754381282e-05,
+      "loss": 0.3209,
+      "step": 6828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524600421114778e-05,
+      "loss": 0.2785,
+      "step": 6829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523813190935116e-05,
+      "loss": 0.269,
+      "step": 6830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523025853285087e-05,
+      "loss": 0.2593,
+      "step": 6831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7522238408175938e-05,
+      "loss": 0.2849,
+      "step": 6832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521450855618912e-05,
+      "loss": 0.2714,
+      "step": 6833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752066319562526e-05,
+      "loss": 0.2956,
+      "step": 6834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751987542820623e-05,
+      "loss": 0.2794,
+      "step": 6835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519087553373083e-05,
+      "loss": 0.3039,
+      "step": 6836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518299571137066e-05,
+      "loss": 0.2776,
+      "step": 6837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751751148150943e-05,
+      "loss": 0.2899,
+      "step": 6838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516723284501447e-05,
+      "loss": 0.2626,
+      "step": 6839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515934980124364e-05,
+      "loss": 0.285,
+      "step": 6840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515146568389445e-05,
+      "loss": 0.2495,
+      "step": 6841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7514358049307947e-05,
+      "loss": 0.2837,
+      "step": 6842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513569422891143e-05,
+      "loss": 0.2694,
+      "step": 6843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751278068915029e-05,
+      "loss": 0.2827,
+      "step": 6844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511991848096658e-05,
+      "loss": 0.274,
+      "step": 6845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751120289974151e-05,
+      "loss": 0.2694,
+      "step": 6846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7510413844096118e-05,
+      "loss": 0.2591,
+      "step": 6847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750962468117175e-05,
+      "loss": 0.2836,
+      "step": 6848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508835410979688e-05,
+      "loss": 0.2582,
+      "step": 6849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508046033531197e-05,
+      "loss": 0.2903,
+      "step": 6850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507256548837553e-05,
+      "loss": 0.2835,
+      "step": 6851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750646695691004e-05,
+      "loss": 0.278,
+      "step": 6852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750567725775993e-05,
+      "loss": 0.2712,
+      "step": 6853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504887451398505e-05,
+      "loss": 0.2983,
+      "step": 6854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504097537837044e-05,
+      "loss": 0.2801,
+      "step": 6855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503307517086833e-05,
+      "loss": 0.2705,
+      "step": 6856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750251738915916e-05,
+      "loss": 0.3237,
+      "step": 6857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7501727154065304e-05,
+      "loss": 0.2805,
+      "step": 6858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750093681181656e-05,
+      "loss": 0.2318,
+      "step": 6859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750014636242421e-05,
+      "loss": 0.2872,
+      "step": 6860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499355805899548e-05,
+      "loss": 0.2881,
+      "step": 6861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498565142253867e-05,
+      "loss": 0.2751,
+      "step": 6862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497774371498462e-05,
+      "loss": 0.2911,
+      "step": 6863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496983493644624e-05,
+      "loss": 0.2955,
+      "step": 6864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496192508703656e-05,
+      "loss": 0.2869,
+      "step": 6865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749540141668685e-05,
+      "loss": 0.2868,
+      "step": 6866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494610217605513e-05,
+      "loss": 0.2804,
+      "step": 6867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493818911470942e-05,
+      "loss": 0.2838,
+      "step": 6868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493027498294437e-05,
+      "loss": 0.2574,
+      "step": 6869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749223597808731e-05,
+      "loss": 0.2559,
+      "step": 6870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749144435086086e-05,
+      "loss": 0.2509,
+      "step": 6871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490652616626403e-05,
+      "loss": 0.2691,
+      "step": 6872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748986077539524e-05,
+      "loss": 0.2848,
+      "step": 6873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489068827178685e-05,
+      "loss": 0.2542,
+      "step": 6874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748827677198805e-05,
+      "loss": 0.2622,
+      "step": 6875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748748460983465e-05,
+      "loss": 0.3468,
+      "step": 6876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74866923407298e-05,
+      "loss": 0.2772,
+      "step": 6877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485899964684813e-05,
+      "loss": 0.2735,
+      "step": 6878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.2595,
+      "step": 6879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484314891819716e-05,
+      "loss": 0.305,
+      "step": 6880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483522195022244e-05,
+      "loss": 0.2743,
+      "step": 6881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748272939132992e-05,
+      "loss": 0.2477,
+      "step": 6882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748193648075407e-05,
+      "loss": 0.2927,
+      "step": 6883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481143463306018e-05,
+      "loss": 0.2936,
+      "step": 6884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7480350338997093e-05,
+      "loss": 0.2943,
+      "step": 6885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7479557107838624e-05,
+      "loss": 0.27,
+      "step": 6886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747876376984194e-05,
+      "loss": 0.3054,
+      "step": 6887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477970325018372e-05,
+      "loss": 0.2495,
+      "step": 6888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477176773379255e-05,
+      "loss": 0.2869,
+      "step": 6889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476383114935928e-05,
+      "loss": 0.2958,
+      "step": 6890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475589349699723e-05,
+      "loss": 0.2848,
+      "step": 6891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474795477681978e-05,
+      "loss": 0.2843,
+      "step": 6892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474001498894036e-05,
+      "loss": 0.3156,
+      "step": 6893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473207413347236e-05,
+      "loss": 0.29,
+      "step": 6894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747241322105292e-05,
+      "loss": 0.2767,
+      "step": 6895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471618922022434e-05,
+      "loss": 0.261,
+      "step": 6896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470824516267125e-05,
+      "loss": 0.2701,
+      "step": 6897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470030003798335e-05,
+      "loss": 0.2542,
+      "step": 6898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469235384627417e-05,
+      "loss": 0.2828,
+      "step": 6899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468440658765722e-05,
+      "loss": 0.2951,
+      "step": 6900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74676458262246e-05,
+      "loss": 0.294,
+      "step": 6901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466850887015403e-05,
+      "loss": 0.2333,
+      "step": 6902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466055841149486e-05,
+      "loss": 0.293,
+      "step": 6903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7465260688638214e-05,
+      "loss": 0.2749,
+      "step": 6904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464465429492934e-05,
+      "loss": 0.2669,
+      "step": 6905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746367006372501e-05,
+      "loss": 0.2562,
+      "step": 6906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462874591345804e-05,
+      "loss": 0.2796,
+      "step": 6907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462079012366677e-05,
+      "loss": 0.2694,
+      "step": 6908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461283326798994e-05,
+      "loss": 0.2633,
+      "step": 6909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746048753465412e-05,
+      "loss": 0.2878,
+      "step": 6910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745969163594342e-05,
+      "loss": 0.3019,
+      "step": 6911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458895630678265e-05,
+      "loss": 0.2852,
+      "step": 6912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458099518870027e-05,
+      "loss": 0.2575,
+      "step": 6913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457303300530073e-05,
+      "loss": 0.2694,
+      "step": 6914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745650697566978e-05,
+      "loss": 0.2806,
+      "step": 6915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455710544300523e-05,
+      "loss": 0.2729,
+      "step": 6916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454914006433676e-05,
+      "loss": 0.2832,
+      "step": 6917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454117362080614e-05,
+      "loss": 0.2439,
+      "step": 6918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453320611252726e-05,
+      "loss": 0.2603,
+      "step": 6919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7452523753961386e-05,
+      "loss": 0.3024,
+      "step": 6920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451726790217974e-05,
+      "loss": 0.2942,
+      "step": 6921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450929720033877e-05,
+      "loss": 0.3184,
+      "step": 6922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745013254342048e-05,
+      "loss": 0.2964,
+      "step": 6923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744933526038917e-05,
+      "loss": 0.2688,
+      "step": 6924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744853787095134e-05,
+      "loss": 0.2932,
+      "step": 6925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447740375118375e-05,
+      "loss": 0.3004,
+      "step": 6926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744694277290166e-05,
+      "loss": 0.2843,
+      "step": 6927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74461450643126e-05,
+      "loss": 0.2301,
+      "step": 6928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445347249362585e-05,
+      "loss": 0.2551,
+      "step": 6929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744454932806301e-05,
+      "loss": 0.2646,
+      "step": 6930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744375130042527e-05,
+      "loss": 0.2541,
+      "step": 6931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744295316646077e-05,
+      "loss": 0.2806,
+      "step": 6932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7442154926180907e-05,
+      "loss": 0.2936,
+      "step": 6933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441356579597082e-05,
+      "loss": 0.2869,
+      "step": 6934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74405581267207e-05,
+      "loss": 0.2616,
+      "step": 6935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743975956756317e-05,
+      "loss": 0.2517,
+      "step": 6936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438960902135894e-05,
+      "loss": 0.2703,
+      "step": 6937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438162130450278e-05,
+      "loss": 0.3055,
+      "step": 6938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743736325251774e-05,
+      "loss": 0.2599,
+      "step": 6939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743656426834968e-05,
+      "loss": 0.2799,
+      "step": 6940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435765177957524e-05,
+      "loss": 0.2623,
+      "step": 6941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434965981352673e-05,
+      "loss": 0.2965,
+      "step": 6942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434166678546548e-05,
+      "loss": 0.2644,
+      "step": 6943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743336726955057e-05,
+      "loss": 0.2476,
+      "step": 6944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743256775437616e-05,
+      "loss": 0.2656,
+      "step": 6945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431768133034724e-05,
+      "loss": 0.2644,
+      "step": 6946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7430968405537697e-05,
+      "loss": 0.2735,
+      "step": 6947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7430168571896494e-05,
+      "loss": 0.2737,
+      "step": 6948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742936863212255e-05,
+      "loss": 0.3027,
+      "step": 6949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428568586227285e-05,
+      "loss": 0.3121,
+      "step": 6950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742776843422212e-05,
+      "loss": 0.2767,
+      "step": 6951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426968176118502e-05,
+      "loss": 0.2638,
+      "step": 6952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742616781192785e-05,
+      "loss": 0.2765,
+      "step": 6953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7425367341661592e-05,
+      "loss": 0.2567,
+      "step": 6954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742456676533117e-05,
+      "loss": 0.259,
+      "step": 6955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742376608294802e-05,
+      "loss": 0.277,
+      "step": 6956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422965294523574e-05,
+      "loss": 0.2776,
+      "step": 6957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422164400069276e-05,
+      "loss": 0.3162,
+      "step": 6958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742136339959656e-05,
+      "loss": 0.2885,
+      "step": 6959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742056229311687e-05,
+      "loss": 0.2595,
+      "step": 6960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419761080641652e-05,
+      "loss": 0.2602,
+      "step": 6961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741895976218235e-05,
+      "loss": 0.2953,
+      "step": 6962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74181583377504e-05,
+      "loss": 0.2772,
+      "step": 6963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7417356807357263e-05,
+      "loss": 0.2788,
+      "step": 6964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7416555171014386e-05,
+      "loss": 0.245,
+      "step": 6965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741575342873321e-05,
+      "loss": 0.2947,
+      "step": 6966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414951580525195e-05,
+      "loss": 0.2735,
+      "step": 6967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414149626401794e-05,
+      "loss": 0.2566,
+      "step": 6968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7413347566374462e-05,
+      "loss": 0.2906,
+      "step": 6969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7412545400454656e-05,
+      "loss": 0.2654,
+      "step": 6970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411743128653828e-05,
+      "loss": 0.2821,
+      "step": 6971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741094075098345e-05,
+      "loss": 0.2473,
+      "step": 6972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410138267454968e-05,
+      "loss": 0.3105,
+      "step": 6973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7409335678079858e-05,
+      "loss": 0.2971,
+      "step": 6974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408532982869573e-05,
+      "loss": 0.2652,
+      "step": 6975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740773018183559e-05,
+      "loss": 0.2713,
+      "step": 6976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740692727498937e-05,
+      "loss": 0.2517,
+      "step": 6977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406124262342382e-05,
+      "loss": 0.2649,
+      "step": 6978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7405321143906095e-05,
+      "loss": 0.2567,
+      "step": 6979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404517919691985e-05,
+      "loss": 0.2979,
+      "step": 6980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740371458971152e-05,
+      "loss": 0.2669,
+      "step": 6981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740291115397618e-05,
+      "loss": 0.277,
+      "step": 6982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740210761249744e-05,
+      "loss": 0.2691,
+      "step": 6983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401303965286774e-05,
+      "loss": 0.2901,
+      "step": 6984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7400500212355663e-05,
+      "loss": 0.257,
+      "step": 6985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7399696353715593e-05,
+      "loss": 0.2744,
+      "step": 6986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398892389378036e-05,
+      "loss": 0.3158,
+      "step": 6987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398088319354487e-05,
+      "loss": 0.3047,
+      "step": 6988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397284143656425e-05,
+      "loss": 0.3261,
+      "step": 6989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739647986229534e-05,
+      "loss": 0.2977,
+      "step": 6990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7395675475282718e-05,
+      "loss": 0.2899,
+      "step": 6991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394870982630047e-05,
+      "loss": 0.2647,
+      "step": 6992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394066384348823e-05,
+      "loss": 0.2475,
+      "step": 6993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7393261680450533e-05,
+      "loss": 0.2884,
+      "step": 6994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739245687094668e-05,
+      "loss": 0.2649,
+      "step": 6995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391651955848753e-05,
+      "loss": 0.2725,
+      "step": 6996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390846935168252e-05,
+      "loss": 0.2766,
+      "step": 6997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390041808916675e-05,
+      "loss": 0.261,
+      "step": 6998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389236577105527e-05,
+      "loss": 0.2706,
+      "step": 6999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.73884312397463e-05,
+      "loss": 0.3122,
+      "step": 7000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738762579685051e-05,
+      "loss": 0.303,
+      "step": 7001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386820248429646e-05,
+      "loss": 0.2963,
+      "step": 7002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738601459449523e-05,
+      "loss": 0.2687,
+      "step": 7003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385208835058764e-05,
+      "loss": 0.2598,
+      "step": 7004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738440297013176e-05,
+      "loss": 0.2738,
+      "step": 7005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738359699972572e-05,
+      "loss": 0.2758,
+      "step": 7006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382790923852163e-05,
+      "loss": 0.2805,
+      "step": 7007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381984742522607e-05,
+      "loss": 0.2665,
+      "step": 7008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738117845574856e-05,
+      "loss": 0.2728,
+      "step": 7009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380372063541543e-05,
+      "loss": 0.2922,
+      "step": 7010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379565565913076e-05,
+      "loss": 0.2951,
+      "step": 7011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378758962874676e-05,
+      "loss": 0.278,
+      "step": 7012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377952254437863e-05,
+      "loss": 0.3076,
+      "step": 7013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377145440614165e-05,
+      "loss": 0.2449,
+      "step": 7014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376338521415107e-05,
+      "loss": 0.3143,
+      "step": 7015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375531496852205e-05,
+      "loss": 0.259,
+      "step": 7016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7374724366936998e-05,
+      "loss": 0.2893,
+      "step": 7017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373917131681012e-05,
+      "loss": 0.2731,
+      "step": 7018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373109791095775e-05,
+      "loss": 0.266,
+      "step": 7019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7372302345192816e-05,
+      "loss": 0.2879,
+      "step": 7020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737149479398368e-05,
+      "loss": 0.2623,
+      "step": 7021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370687137479892e-05,
+      "loss": 0.2459,
+      "step": 7022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736987937569299e-05,
+      "loss": 0.2569,
+      "step": 7023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7369071508634516e-05,
+      "loss": 0.3139,
+      "step": 7024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368263536316007e-05,
+      "loss": 0.2903,
+      "step": 7025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367455458749005e-05,
+      "loss": 0.2783,
+      "step": 7026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736664727594505e-05,
+      "loss": 0.2636,
+      "step": 7027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736583898791569e-05,
+      "loss": 0.2428,
+      "step": 7028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365030594672467e-05,
+      "loss": 0.2648,
+      "step": 7029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364222096226933e-05,
+      "loss": 0.2837,
+      "step": 7030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736341349259063e-05,
+      "loss": 0.2571,
+      "step": 7031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7362604783775117e-05,
+      "loss": 0.3014,
+      "step": 7032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361795969791935e-05,
+      "loss": 0.2861,
+      "step": 7033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736098705065264e-05,
+      "loss": 0.2775,
+      "step": 7034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7360178026368796e-05,
+      "loss": 0.2847,
+      "step": 7035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735936889695195e-05,
+      "loss": 0.2943,
+      "step": 7036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735855966241366e-05,
+      "loss": 0.2546,
+      "step": 7037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735775032276549e-05,
+      "loss": 0.2865,
+      "step": 7038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356940878018994e-05,
+      "loss": 0.2724,
+      "step": 7039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356131328185737e-05,
+      "loss": 0.2761,
+      "step": 7040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7355321673277288e-05,
+      "loss": 0.3068,
+      "step": 7041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354511913305203e-05,
+      "loss": 0.2881,
+      "step": 7042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353702048281053e-05,
+      "loss": 0.2836,
+      "step": 7043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352892078216407e-05,
+      "loss": 0.2449,
+      "step": 7044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352082003122834e-05,
+      "loss": 0.2883,
+      "step": 7045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351271823011904e-05,
+      "loss": 0.2968,
+      "step": 7046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350461537895194e-05,
+      "loss": 0.2871,
+      "step": 7047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349651147784275e-05,
+      "loss": 0.266,
+      "step": 7048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734884065269072e-05,
+      "loss": 0.2669,
+      "step": 7049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734803005262611e-05,
+      "loss": 0.2497,
+      "step": 7050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347219347602017e-05,
+      "loss": 0.2896,
+      "step": 7051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7346408537630032e-05,
+      "loss": 0.2809,
+      "step": 7052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345597622721727e-05,
+      "loss": 0.2565,
+      "step": 7053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344786602888692e-05,
+      "loss": 0.2491,
+      "step": 7054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343975478142506e-05,
+      "loss": 0.2997,
+      "step": 7055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734316424849476e-05,
+      "loss": 0.2807,
+      "step": 7056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734235291395704e-05,
+      "loss": 0.2836,
+      "step": 7057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341541474540934e-05,
+      "loss": 0.2646,
+      "step": 7058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734072993025803e-05,
+      "loss": 0.2765,
+      "step": 7059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339918281119928e-05,
+      "loss": 0.2985,
+      "step": 7060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339106527138213e-05,
+      "loss": 0.2811,
+      "step": 7061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338294668324486e-05,
+      "loss": 0.3034,
+      "step": 7062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337482704690343e-05,
+      "loss": 0.2832,
+      "step": 7063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336670636247376e-05,
+      "loss": 0.2662,
+      "step": 7064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335858463007196e-05,
+      "loss": 0.2671,
+      "step": 7065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335046184981393e-05,
+      "loss": 0.2937,
+      "step": 7066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7334233802181575e-05,
+      "loss": 0.2525,
+      "step": 7067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333421314619346e-05,
+      "loss": 0.2727,
+      "step": 7068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733260872230631e-05,
+      "loss": 0.2676,
+      "step": 7069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331796025254072e-05,
+      "loss": 0.2476,
+      "step": 7070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330983223474248e-05,
+      "loss": 0.3182,
+      "step": 7071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733017031697844e-05,
+      "loss": 0.2676,
+      "step": 7072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329357305778267e-05,
+      "loss": 0.2687,
+      "step": 7073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7328544189885338e-05,
+      "loss": 0.2728,
+      "step": 7074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327730969311263e-05,
+      "loss": 0.3107,
+      "step": 7075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326917644067666e-05,
+      "loss": 0.2788,
+      "step": 7076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326104214166163e-05,
+      "loss": 0.2812,
+      "step": 7077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325290679618367e-05,
+      "loss": 0.2884,
+      "step": 7078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324477040435906e-05,
+      "loss": 0.28,
+      "step": 7079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73236632966304e-05,
+      "loss": 0.2656,
+      "step": 7080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732284944821347e-05,
+      "loss": 0.2692,
+      "step": 7081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7322035495196745e-05,
+      "loss": 0.2848,
+      "step": 7082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321221437591847e-05,
+      "loss": 0.3076,
+      "step": 7083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320407275410405e-05,
+      "loss": 0.2701,
+      "step": 7084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731959300866405e-05,
+      "loss": 0.2795,
+      "step": 7085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318778637364416e-05,
+      "loss": 0.2813,
+      "step": 7086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731796416152313e-05,
+      "loss": 0.273,
+      "step": 7087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317149581151827e-05,
+      "loss": 0.2728,
+      "step": 7088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316334896262145e-05,
+      "loss": 0.2979,
+      "step": 7089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731552010686572e-05,
+      "loss": 0.2842,
+      "step": 7090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731470521297419e-05,
+      "loss": 0.291,
+      "step": 7091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313890214599195e-05,
+      "loss": 0.2904,
+      "step": 7092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313075111752374e-05,
+      "loss": 0.2462,
+      "step": 7093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7312259904445374e-05,
+      "loss": 0.3087,
+      "step": 7094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731144459268984e-05,
+      "loss": 0.258,
+      "step": 7095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310629176497413e-05,
+      "loss": 0.2547,
+      "step": 7096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7309813655879746e-05,
+      "loss": 0.2701,
+      "step": 7097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7308998030848484e-05,
+      "loss": 0.2967,
+      "step": 7098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730818230141528e-05,
+      "loss": 0.2918,
+      "step": 7099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7307366467591782e-05,
+      "loss": 0.2949,
+      "step": 7100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730655052938965e-05,
+      "loss": 0.2767,
+      "step": 7101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730573448682053e-05,
+      "loss": 0.2911,
+      "step": 7102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304918339896087e-05,
+      "loss": 0.2654,
+      "step": 7103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304102088627976e-05,
+      "loss": 0.2935,
+      "step": 7104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7303285733027852e-05,
+      "loss": 0.2851,
+      "step": 7105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302469273107377e-05,
+      "loss": 0.292,
+      "step": 7106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730165270887822e-05,
+      "loss": 0.2556,
+      "step": 7107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300836040352044e-05,
+      "loss": 0.33,
+      "step": 7108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300019267540502e-05,
+      "loss": 0.2828,
+      "step": 7109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7299202390455278e-05,
+      "loss": 0.2959,
+      "step": 7110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298385409108026e-05,
+      "loss": 0.3022,
+      "step": 7111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297568323510422e-05,
+      "loss": 0.2918,
+      "step": 7112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729675113367414e-05,
+      "loss": 0.2635,
+      "step": 7113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295933839610847e-05,
+      "loss": 0.2838,
+      "step": 7114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295116441332223e-05,
+      "loss": 0.2712,
+      "step": 7115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729429893884994e-05,
+      "loss": 0.2734,
+      "step": 7116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293481332175674e-05,
+      "loss": 0.2819,
+      "step": 7117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292663621321106e-05,
+      "loss": 0.2835,
+      "step": 7118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291845806297916e-05,
+      "loss": 0.2292,
+      "step": 7119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291027887117787e-05,
+      "loss": 0.2885,
+      "step": 7120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72902098637924e-05,
+      "loss": 0.2943,
+      "step": 7121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728939173633344e-05,
+      "loss": 0.3016,
+      "step": 7122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288573504752593e-05,
+      "loss": 0.2806,
+      "step": 7123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728775516906155e-05,
+      "loss": 0.2759,
+      "step": 7124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286936729271993e-05,
+      "loss": 0.2747,
+      "step": 7125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728611818539562e-05,
+      "loss": 0.2637,
+      "step": 7126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728529953744412e-05,
+      "loss": 0.291,
+      "step": 7127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284480785429185e-05,
+      "loss": 0.2755,
+      "step": 7128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283661929362515e-05,
+      "loss": 0.2955,
+      "step": 7129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282842969255798e-05,
+      "loss": 0.2897,
+      "step": 7130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282023905120743e-05,
+      "loss": 0.3091,
+      "step": 7131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728120473696904e-05,
+      "loss": 0.2717,
+      "step": 7132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280385464812393e-05,
+      "loss": 0.273,
+      "step": 7133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727956608866251e-05,
+      "loss": 0.2619,
+      "step": 7134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727874660853109e-05,
+      "loss": 0.3051,
+      "step": 7135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277927024429836e-05,
+      "loss": 0.2545,
+      "step": 7136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277107336370455e-05,
+      "loss": 0.2773,
+      "step": 7137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276287544364664e-05,
+      "loss": 0.2554,
+      "step": 7138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275467648424166e-05,
+      "loss": 0.2769,
+      "step": 7139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727464764856067e-05,
+      "loss": 0.2714,
+      "step": 7140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72738275447859e-05,
+      "loss": 0.2917,
+      "step": 7141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273007337111555e-05,
+      "loss": 0.2889,
+      "step": 7142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727218702554936e-05,
+      "loss": 0.2697,
+      "step": 7143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271366610111034e-05,
+      "loss": 0.2923,
+      "step": 7144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7270546090808294e-05,
+      "loss": 0.259,
+      "step": 7145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726972546765286e-05,
+      "loss": 0.3,
+      "step": 7146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268904740656452e-05,
+      "loss": 0.2596,
+      "step": 7147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726808390983079e-05,
+      "loss": 0.241,
+      "step": 7148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267262975187605e-05,
+      "loss": 0.271,
+      "step": 7149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266441936738625e-05,
+      "loss": 0.2695,
+      "step": 7150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265620794495574e-05,
+      "loss": 0.2669,
+      "step": 7151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726479954847018e-05,
+      "loss": 0.2857,
+      "step": 7152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7263978198674172e-05,
+      "loss": 0.2479,
+      "step": 7153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726315674511929e-05,
+      "loss": 0.2732,
+      "step": 7154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726233518781726e-05,
+      "loss": 0.2887,
+      "step": 7155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261513526779822e-05,
+      "loss": 0.2796,
+      "step": 7156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7260691762018708e-05,
+      "loss": 0.292,
+      "step": 7157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259869893545663e-05,
+      "loss": 0.2875,
+      "step": 7158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259047921372418e-05,
+      "loss": 0.2813,
+      "step": 7159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7258225845510723e-05,
+      "loss": 0.2744,
+      "step": 7160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725740366597231e-05,
+      "loss": 0.2709,
+      "step": 7161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256581382768932e-05,
+      "loss": 0.2921,
+      "step": 7162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7255758995912334e-05,
+      "loss": 0.2999,
+      "step": 7163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254936505414258e-05,
+      "loss": 0.2857,
+      "step": 7164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254113911286458e-05,
+      "loss": 0.2559,
+      "step": 7165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7253291213540677e-05,
+      "loss": 0.3044,
+      "step": 7166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7252468412188675e-05,
+      "loss": 0.2674,
+      "step": 7167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251645507242197e-05,
+      "loss": 0.2851,
+      "step": 7168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250822498713004e-05,
+      "loss": 0.2548,
+      "step": 7169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249999386612844e-05,
+      "loss": 0.2815,
+      "step": 7170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724917617095348e-05,
+      "loss": 0.2678,
+      "step": 7171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7248352851746673e-05,
+      "loss": 0.3069,
+      "step": 7172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724752942900418e-05,
+      "loss": 0.2512,
+      "step": 7173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724670590273776e-05,
+      "loss": 0.3258,
+      "step": 7174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724588227295918e-05,
+      "loss": 0.2851,
+      "step": 7175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7245058539680205e-05,
+      "loss": 0.2805,
+      "step": 7176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72442347029126e-05,
+      "loss": 0.3003,
+      "step": 7177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243410762668133e-05,
+      "loss": 0.2577,
+      "step": 7178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242586718958576e-05,
+      "loss": 0.2745,
+      "step": 7179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7241762571795694e-05,
+      "loss": 0.267,
+      "step": 7180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724093832119126e-05,
+      "loss": 0.2745,
+      "step": 7181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7240113967157052e-05,
+      "loss": 0.2852,
+      "step": 7182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7239289509704843e-05,
+      "loss": 0.2733,
+      "step": 7183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723846494884641e-05,
+      "loss": 0.2828,
+      "step": 7184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7237640284593528e-05,
+      "loss": 0.2905,
+      "step": 7185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7236815516957982e-05,
+      "loss": 0.27,
+      "step": 7186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235990645951545e-05,
+      "loss": 0.2895,
+      "step": 7187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235165671586008e-05,
+      "loss": 0.2926,
+      "step": 7188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234340593873147e-05,
+      "loss": 0.282,
+      "step": 7189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233515412824758e-05,
+      "loss": 0.3227,
+      "step": 7190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232690128452615e-05,
+      "loss": 0.2804,
+      "step": 7191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723186474076852e-05,
+      "loss": 0.2721,
+      "step": 7192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723103924978425e-05,
+      "loss": 0.261,
+      "step": 7193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230213655511607e-05,
+      "loss": 0.2875,
+      "step": 7194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229387957962377e-05,
+      "loss": 0.2732,
+      "step": 7195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7228562157148355e-05,
+      "loss": 0.3034,
+      "step": 7196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722773625308134e-05,
+      "loss": 0.2724,
+      "step": 7197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226910245773125e-05,
+      "loss": 0.2767,
+      "step": 7198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226084135235512e-05,
+      "loss": 0.2772,
+      "step": 7199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72252579214803e-05,
+      "loss": 0.2992,
+      "step": 7200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224431604519292e-05,
+      "loss": 0.2881,
+      "step": 7201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722360518436429e-05,
+      "loss": 0.2783,
+      "step": 7202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7222778661027098e-05,
+      "loss": 0.2608,
+      "step": 7203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722195203451952e-05,
+      "loss": 0.2634,
+      "step": 7204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7221125304853373e-05,
+      "loss": 0.2898,
+      "step": 7205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220298472040453e-05,
+      "loss": 0.3057,
+      "step": 7206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721947153609258e-05,
+      "loss": 0.2825,
+      "step": 7207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721864449702156e-05,
+      "loss": 0.274,
+      "step": 7208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721781735483921e-05,
+      "loss": 0.2692,
+      "step": 7209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216990109557346e-05,
+      "loss": 0.2736,
+      "step": 7210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216162761187783e-05,
+      "loss": 0.2667,
+      "step": 7211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215335309742337e-05,
+      "loss": 0.2704,
+      "step": 7212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721450775523283e-05,
+      "loss": 0.269,
+      "step": 7213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721368009767108e-05,
+      "loss": 0.2647,
+      "step": 7214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212852337068912e-05,
+      "loss": 0.282,
+      "step": 7215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212024473438145e-05,
+      "loss": 0.2786,
+      "step": 7216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721119650679061e-05,
+      "loss": 0.281,
+      "step": 7217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210368437138135e-05,
+      "loss": 0.2912,
+      "step": 7218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209540264492544e-05,
+      "loss": 0.2668,
+      "step": 7219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7208711988865663e-05,
+      "loss": 0.2653,
+      "step": 7220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720788361026933e-05,
+      "loss": 0.2636,
+      "step": 7221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207055128715374e-05,
+      "loss": 0.2602,
+      "step": 7222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206226544215632e-05,
+      "loss": 0.2547,
+      "step": 7223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7205397856781935e-05,
+      "loss": 0.2649,
+      "step": 7224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7204569066426126e-05,
+      "loss": 0.2707,
+      "step": 7225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720374017316004e-05,
+      "loss": 0.2547,
+      "step": 7226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720291117699551e-05,
+      "loss": 0.2515,
+      "step": 7227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7202082077944392e-05,
+      "loss": 0.285,
+      "step": 7228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201252876018523e-05,
+      "loss": 0.2849,
+      "step": 7229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200423571229738e-05,
+      "loss": 0.2785,
+      "step": 7230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199594163589893e-05,
+      "loss": 0.2801,
+      "step": 7231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198764653110836e-05,
+      "loss": 0.3402,
+      "step": 7232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719793503980441e-05,
+      "loss": 0.2897,
+      "step": 7233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719710532368247e-05,
+      "loss": 0.2521,
+      "step": 7234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196275504756862e-05,
+      "loss": 0.3181,
+      "step": 7235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195445583039443e-05,
+      "loss": 0.2799,
+      "step": 7236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194615558542068e-05,
+      "loss": 0.2742,
+      "step": 7237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193785431276593e-05,
+      "loss": 0.2946,
+      "step": 7238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7192955201254876e-05,
+      "loss": 0.2945,
+      "step": 7239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719212486848877e-05,
+      "loss": 0.3042,
+      "step": 7240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7191294432990144e-05,
+      "loss": 0.2719,
+      "step": 7241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190463894770856e-05,
+      "loss": 0.2867,
+      "step": 7242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718963325384277e-05,
+      "loss": 0.2917,
+      "step": 7243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718880251021775e-05,
+      "loss": 0.2748,
+      "step": 7244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187971663907666e-05,
+      "loss": 0.2811,
+      "step": 7245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718714071492438e-05,
+      "loss": 0.2807,
+      "step": 7246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186309663279764e-05,
+      "loss": 0.2886,
+      "step": 7247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718547850898569e-05,
+      "loss": 0.2627,
+      "step": 7248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7184647252054027e-05,
+      "loss": 0.2559,
+      "step": 7249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7183815892496652e-05,
+      "loss": 0.2822,
+      "step": 7250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718298443032544e-05,
+      "loss": 0.2954,
+      "step": 7251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182152865552267e-05,
+      "loss": 0.2801,
+      "step": 7252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181321198189008e-05,
+      "loss": 0.2817,
+      "step": 7253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180489428247548e-05,
+      "loss": 0.2812,
+      "step": 7254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179657555739767e-05,
+      "loss": 0.2586,
+      "step": 7255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7178825580677542e-05,
+      "loss": 0.3007,
+      "step": 7256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717799350307276e-05,
+      "loss": 0.275,
+      "step": 7257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717716132293731e-05,
+      "loss": 0.2678,
+      "step": 7258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7176329040283077e-05,
+      "loss": 0.2858,
+      "step": 7259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7175496655121947e-05,
+      "loss": 0.3038,
+      "step": 7260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717466416746581e-05,
+      "loss": 0.3074,
+      "step": 7261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7173831577326564e-05,
+      "loss": 0.2881,
+      "step": 7262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717299888471609e-05,
+      "loss": 0.2947,
+      "step": 7263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717216608964629e-05,
+      "loss": 0.2976,
+      "step": 7264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717133319212906e-05,
+      "loss": 0.2619,
+      "step": 7265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7170500192176293e-05,
+      "loss": 0.2646,
+      "step": 7266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716966708979989e-05,
+      "loss": 0.2794,
+      "step": 7267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7168833885011754e-05,
+      "loss": 0.3266,
+      "step": 7268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716800057782378e-05,
+      "loss": 0.3016,
+      "step": 7269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7167167168247876e-05,
+      "loss": 0.261,
+      "step": 7270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7166333656295946e-05,
+      "loss": 0.2948,
+      "step": 7271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716550004197989e-05,
+      "loss": 0.2872,
+      "step": 7272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164666325311625e-05,
+      "loss": 0.2924,
+      "step": 7273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7163832506303056e-05,
+      "loss": 0.2665,
+      "step": 7274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716299858496609e-05,
+      "loss": 0.2944,
+      "step": 7275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716216456131264e-05,
+      "loss": 0.3142,
+      "step": 7276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161330435354623e-05,
+      "loss": 0.2799,
+      "step": 7277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716049620710395e-05,
+      "loss": 0.2799,
+      "step": 7278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159661876572535e-05,
+      "loss": 0.3,
+      "step": 7279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71588274437723e-05,
+      "loss": 0.2818,
+      "step": 7280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157992908715163e-05,
+      "loss": 0.2792,
+      "step": 7281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157158271413047e-05,
+      "loss": 0.2755,
+      "step": 7282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715632353187787e-05,
+      "loss": 0.2607,
+      "step": 7283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155488690121557e-05,
+      "loss": 0.266,
+      "step": 7284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715465374615603e-05,
+      "loss": 0.2534,
+      "step": 7285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715381869999322e-05,
+      "loss": 0.2752,
+      "step": 7286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152983551645054e-05,
+      "loss": 0.2657,
+      "step": 7287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152148301123458e-05,
+      "loss": 0.2698,
+      "step": 7288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151312948440364e-05,
+      "loss": 0.2591,
+      "step": 7289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7150477493607706e-05,
+      "loss": 0.2596,
+      "step": 7290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714964193663742e-05,
+      "loss": 0.2717,
+      "step": 7291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7148806277541435e-05,
+      "loss": 0.2699,
+      "step": 7292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147970516331688e-05,
+      "loss": 0.2442,
+      "step": 7293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147134653020122e-05,
+      "loss": 0.2932,
+      "step": 7294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146298687618677e-05,
+      "loss": 0.2531,
+      "step": 7295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714546262013929e-05,
+      "loss": 0.294,
+      "step": 7296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144626450593897e-05,
+      "loss": 0.3092,
+      "step": 7297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7143790178994456e-05,
+      "loss": 0.2957,
+      "step": 7298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142953805352905e-05,
+      "loss": 0.2598,
+      "step": 7299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714211732968119e-05,
+      "loss": 0.2699,
+      "step": 7300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714128075199126e-05,
+      "loss": 0.2681,
+      "step": 7301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7140444072295065e-05,
+      "loss": 0.2521,
+      "step": 7302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713960729060456e-05,
+      "loss": 0.2771,
+      "step": 7303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713877040693169e-05,
+      "loss": 0.2615,
+      "step": 7304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137933421288412e-05,
+      "loss": 0.2505,
+      "step": 7305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713709633368669e-05,
+      "loss": 0.2486,
+      "step": 7306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136259144138467e-05,
+      "loss": 0.2811,
+      "step": 7307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713542185265571e-05,
+      "loss": 0.286,
+      "step": 7308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134584459250378e-05,
+      "loss": 0.2836,
+      "step": 7309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133746963934425e-05,
+      "loss": 0.2805,
+      "step": 7310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7132909366719826e-05,
+      "loss": 0.2711,
+      "step": 7311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713207166761854e-05,
+      "loss": 0.2928,
+      "step": 7312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713123386664253e-05,
+      "loss": 0.3137,
+      "step": 7313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7130395963803765e-05,
+      "loss": 0.2603,
+      "step": 7314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129557959114212e-05,
+      "loss": 0.2708,
+      "step": 7315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7128719852585847e-05,
+      "loss": 0.267,
+      "step": 7316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127881644230634e-05,
+      "loss": 0.2876,
+      "step": 7317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127043334060553e-05,
+      "loss": 0.2444,
+      "step": 7318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126204922087574e-05,
+      "loss": 0.2758,
+      "step": 7319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125366408323675e-05,
+      "loss": 0.252,
+      "step": 7320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124527792780828e-05,
+      "loss": 0.2911,
+      "step": 7321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712368907547102e-05,
+      "loss": 0.2962,
+      "step": 7322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122850256406226e-05,
+      "loss": 0.2744,
+      "step": 7323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122011335598428e-05,
+      "loss": 0.2426,
+      "step": 7324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121172313059608e-05,
+      "loss": 0.2664,
+      "step": 7325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120333188801756e-05,
+      "loss": 0.2648,
+      "step": 7326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7119493962836858e-05,
+      "loss": 0.2562,
+      "step": 7327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118654635176892e-05,
+      "loss": 0.2518,
+      "step": 7328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117815205833856e-05,
+      "loss": 0.2792,
+      "step": 7329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711697567481974e-05,
+      "loss": 0.257,
+      "step": 7330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711613604214653e-05,
+      "loss": 0.2579,
+      "step": 7331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115296307826224e-05,
+      "loss": 0.2818,
+      "step": 7332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114456471870816e-05,
+      "loss": 0.27,
+      "step": 7333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113616534292302e-05,
+      "loss": 0.2426,
+      "step": 7334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711277649510268e-05,
+      "loss": 0.2335,
+      "step": 7335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711193635431395e-05,
+      "loss": 0.2455,
+      "step": 7336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111096111938113e-05,
+      "loss": 0.273,
+      "step": 7337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7110255767987167e-05,
+      "loss": 0.2944,
+      "step": 7338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109415322473117e-05,
+      "loss": 0.2859,
+      "step": 7339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7108574775407966e-05,
+      "loss": 0.2894,
+      "step": 7340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710773412680373e-05,
+      "loss": 0.2903,
+      "step": 7341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710689337667241e-05,
+      "loss": 0.275,
+      "step": 7342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7106052525026007e-05,
+      "loss": 0.2644,
+      "step": 7343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710521157187655e-05,
+      "loss": 0.2615,
+      "step": 7344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104370517236037e-05,
+      "loss": 0.2843,
+      "step": 7345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103529361116487e-05,
+      "loss": 0.2692,
+      "step": 7346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102688103529915e-05,
+      "loss": 0.2704,
+      "step": 7347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710184674448834e-05,
+      "loss": 0.2732,
+      "step": 7348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7101005284003765e-05,
+      "loss": 0.2859,
+      "step": 7349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100163722088233e-05,
+      "loss": 0.317,
+      "step": 7350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709932205875375e-05,
+      "loss": 0.2635,
+      "step": 7351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7098480294012337e-05,
+      "loss": 0.2627,
+      "step": 7352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097638427876028e-05,
+      "loss": 0.2674,
+      "step": 7353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7096796460356838e-05,
+      "loss": 0.2627,
+      "step": 7354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.70959543914668e-05,
+      "loss": 0.2745,
+      "step": 7355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095112221217943e-05,
+      "loss": 0.2901,
+      "step": 7356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709426994962229e-05,
+      "loss": 0.2805,
+      "step": 7357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093427576691877e-05,
+      "loss": 0.2964,
+      "step": 7358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7092585102438737e-05,
+      "loss": 0.2287,
+      "step": 7359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7091742526874903e-05,
+      "loss": 0.2802,
+      "step": 7360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7090899850012406e-05,
+      "loss": 0.2762,
+      "step": 7361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7090057071863288e-05,
+      "loss": 0.269,
+      "step": 7362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7089214192439584e-05,
+      "loss": 0.2689,
+      "step": 7363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088371211753342e-05,
+      "loss": 0.2906,
+      "step": 7364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708752812981659e-05,
+      "loss": 0.2543,
+      "step": 7365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086684946641382e-05,
+      "loss": 0.2595,
+      "step": 7366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085841662239754e-05,
+      "loss": 0.2743,
+      "step": 7367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7084998276623753e-05,
+      "loss": 0.2715,
+      "step": 7368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7084154789805432e-05,
+      "loss": 0.2694,
+      "step": 7369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083311201796834e-05,
+      "loss": 0.244,
+      "step": 7370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708246751261001e-05,
+      "loss": 0.2847,
+      "step": 7371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081623722257014e-05,
+      "loss": 0.2492,
+      "step": 7372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708077983074989e-05,
+      "loss": 0.2697,
+      "step": 7373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70799358381007e-05,
+      "loss": 0.2803,
+      "step": 7374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70790917443215e-05,
+      "loss": 0.2795,
+      "step": 7375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078247549424344e-05,
+      "loss": 0.3162,
+      "step": 7376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7077403253421292e-05,
+      "loss": 0.263,
+      "step": 7377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70765588563244e-05,
+      "loss": 0.2856,
+      "step": 7378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075714358145735e-05,
+      "loss": 0.2754,
+      "step": 7379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707486975889736e-05,
+      "loss": 0.2799,
+      "step": 7380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074025058591334e-05,
+      "loss": 0.2852,
+      "step": 7381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073180257239725e-05,
+      "loss": 0.2819,
+      "step": 7382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072335354854602e-05,
+      "loss": 0.2682,
+      "step": 7383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071490351448034e-05,
+      "loss": 0.2967,
+      "step": 7384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070645247032086e-05,
+      "loss": 0.2836,
+      "step": 7385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7069800041618836e-05,
+      "loss": 0.2908,
+      "step": 7386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706895473522035e-05,
+      "loss": 0.264,
+      "step": 7387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068109327848708e-05,
+      "loss": 0.2621,
+      "step": 7388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067263819515988e-05,
+      "loss": 0.2906,
+      "step": 7389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066418210234258e-05,
+      "loss": 0.2679,
+      "step": 7390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065572500015602e-05,
+      "loss": 0.2833,
+      "step": 7391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7064726688872103e-05,
+      "loss": 0.2901,
+      "step": 7392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063880776815837e-05,
+      "loss": 0.2891,
+      "step": 7393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706303476385889e-05,
+      "loss": 0.2951,
+      "step": 7394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706218865001335e-05,
+      "loss": 0.2492,
+      "step": 7395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061342435291296e-05,
+      "loss": 0.2792,
+      "step": 7396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706049611970482e-05,
+      "loss": 0.2811,
+      "step": 7397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059649703266007e-05,
+      "loss": 0.2949,
+      "step": 7398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058803185986953e-05,
+      "loss": 0.2841,
+      "step": 7399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057956567879742e-05,
+      "loss": 0.3108,
+      "step": 7400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057109848956475e-05,
+      "loss": 0.2613,
+      "step": 7401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705626302922924e-05,
+      "loss": 0.2591,
+      "step": 7402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705541610871014e-05,
+      "loss": 0.2698,
+      "step": 7403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054569087411262e-05,
+      "loss": 0.2668,
+      "step": 7404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7053721965344717e-05,
+      "loss": 0.2384,
+      "step": 7405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052874742522596e-05,
+      "loss": 0.2388,
+      "step": 7406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052027418957003e-05,
+      "loss": 0.2699,
+      "step": 7407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705117999466005e-05,
+      "loss": 0.2489,
+      "step": 7408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050332469643827e-05,
+      "loss": 0.2722,
+      "step": 7409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7049484843920453e-05,
+      "loss": 0.2663,
+      "step": 7410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048637117502026e-05,
+      "loss": 0.287,
+      "step": 7411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047789290400655e-05,
+      "loss": 0.2674,
+      "step": 7412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704694136262846e-05,
+      "loss": 0.2772,
+      "step": 7413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704609333419754e-05,
+      "loss": 0.2399,
+      "step": 7414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045245205120023e-05,
+      "loss": 0.2851,
+      "step": 7415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7044396975408012e-05,
+      "loss": 0.2281,
+      "step": 7416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704354864507363e-05,
+      "loss": 0.2518,
+      "step": 7417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704270021412899e-05,
+      "loss": 0.2731,
+      "step": 7418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704185168258621e-05,
+      "loss": 0.2649,
+      "step": 7419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704100305045741e-05,
+      "loss": 0.265,
+      "step": 7420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040154317754724e-05,
+      "loss": 0.2737,
+      "step": 7421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703930548449026e-05,
+      "loss": 0.2812,
+      "step": 7422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038456550676152e-05,
+      "loss": 0.2822,
+      "step": 7423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037607516324522e-05,
+      "loss": 0.3094,
+      "step": 7424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70367583814475e-05,
+      "loss": 0.2853,
+      "step": 7425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703590914605721e-05,
+      "loss": 0.2926,
+      "step": 7426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703505981016579e-05,
+      "loss": 0.2774,
+      "step": 7427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7034210373785368e-05,
+      "loss": 0.2838,
+      "step": 7428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033360836928077e-05,
+      "loss": 0.2978,
+      "step": 7429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703251119960605e-05,
+      "loss": 0.284,
+      "step": 7430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703166146183143e-05,
+      "loss": 0.2783,
+      "step": 7431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030811623616345e-05,
+      "loss": 0.317,
+      "step": 7432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029961684972943e-05,
+      "loss": 0.2755,
+      "step": 7433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029111645913362e-05,
+      "loss": 0.2699,
+      "step": 7434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028261506449745e-05,
+      "loss": 0.275,
+      "step": 7435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702741126659423e-05,
+      "loss": 0.2778,
+      "step": 7436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026560926358963e-05,
+      "loss": 0.3277,
+      "step": 7437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025710485756098e-05,
+      "loss": 0.2908,
+      "step": 7438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024859944797773e-05,
+      "loss": 0.2812,
+      "step": 7439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024009303496143e-05,
+      "loss": 0.288,
+      "step": 7440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702315856186336e-05,
+      "loss": 0.263,
+      "step": 7441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702230771991157e-05,
+      "loss": 0.2486,
+      "step": 7442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702145677765293e-05,
+      "loss": 0.2633,
+      "step": 7443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702060573509959e-05,
+      "loss": 0.2803,
+      "step": 7444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701975459226372e-05,
+      "loss": 0.2402,
+      "step": 7445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701890334915746e-05,
+      "loss": 0.2687,
+      "step": 7446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701805200579298e-05,
+      "loss": 0.2964,
+      "step": 7447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017200562182442e-05,
+      "loss": 0.2876,
+      "step": 7448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016349018338e-05,
+      "loss": 0.2719,
+      "step": 7449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015497374271826e-05,
+      "loss": 0.321,
+      "step": 7450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7014645629996077e-05,
+      "loss": 0.2886,
+      "step": 7451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013793785522926e-05,
+      "loss": 0.3042,
+      "step": 7452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7012941840864535e-05,
+      "loss": 0.2891,
+      "step": 7453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701208979603308e-05,
+      "loss": 0.2856,
+      "step": 7454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011237651040723e-05,
+      "loss": 0.2699,
+      "step": 7455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701038540589964e-05,
+      "loss": 0.2873,
+      "step": 7456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700953306062201e-05,
+      "loss": 0.2534,
+      "step": 7457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700868061522e-05,
+      "loss": 0.2725,
+      "step": 7458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700782806970579e-05,
+      "loss": 0.3025,
+      "step": 7459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006975424091558e-05,
+      "loss": 0.2739,
+      "step": 7460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006122678389484e-05,
+      "loss": 0.3335,
+      "step": 7461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7005269832611744e-05,
+      "loss": 0.3106,
+      "step": 7462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004416886770525e-05,
+      "loss": 0.2993,
+      "step": 7463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700356384087801e-05,
+      "loss": 0.2542,
+      "step": 7464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700271069494638e-05,
+      "loss": 0.2695,
+      "step": 7465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001857448987825e-05,
+      "loss": 0.2662,
+      "step": 7466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001004103014534e-05,
+      "loss": 0.298,
+      "step": 7467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000150657038687e-05,
+      "loss": 0.254,
+      "step": 7468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999297111072488e-05,
+      "loss": 0.2758,
+      "step": 7469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699844346512812e-05,
+      "loss": 0.2695,
+      "step": 7470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699758971921778e-05,
+      "loss": 0.2543,
+      "step": 7471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996735873353664e-05,
+      "loss": 0.2413,
+      "step": 7472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995881927547965e-05,
+      "loss": 0.2572,
+      "step": 7473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995027881812882e-05,
+      "loss": 0.2999,
+      "step": 7474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994173736160617e-05,
+      "loss": 0.2594,
+      "step": 7475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6993319490603368e-05,
+      "loss": 0.2712,
+      "step": 7476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699246514515333e-05,
+      "loss": 0.2821,
+      "step": 7477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991610699822723e-05,
+      "loss": 0.3067,
+      "step": 7478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6990756154623736e-05,
+      "loss": 0.2845,
+      "step": 7479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989901509568588e-05,
+      "loss": 0.2528,
+      "step": 7480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989046764669478e-05,
+      "loss": 0.2744,
+      "step": 7481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6988191919938618e-05,
+      "loss": 0.2682,
+      "step": 7482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6987336975388218e-05,
+      "loss": 0.2693,
+      "step": 7483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698648193103049e-05,
+      "loss": 0.2922,
+      "step": 7484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985626786877646e-05,
+      "loss": 0.2633,
+      "step": 7485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698477154294191e-05,
+      "loss": 0.2855,
+      "step": 7486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983916199235487e-05,
+      "loss": 0.2673,
+      "step": 7487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69830607557706e-05,
+      "loss": 0.3183,
+      "step": 7488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982205212559467e-05,
+      "loss": 0.2708,
+      "step": 7489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698134956961431e-05,
+      "loss": 0.2509,
+      "step": 7490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698049382694735e-05,
+      "loss": 0.2525,
+      "step": 7491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979637984570807e-05,
+      "loss": 0.2793,
+      "step": 7492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6978782042496912e-05,
+      "loss": 0.3199,
+      "step": 7493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977926000737888e-05,
+      "loss": 0.2691,
+      "step": 7494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977069859305966e-05,
+      "loss": 0.2574,
+      "step": 7495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976213618213367e-05,
+      "loss": 0.3129,
+      "step": 7496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697535727747233e-05,
+      "loss": 0.262,
+      "step": 7497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974500837095084e-05,
+      "loss": 0.2599,
+      "step": 7498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973644297093862e-05,
+      "loss": 0.2703,
+      "step": 7499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69727876574809e-05,
+      "loss": 0.3074,
+      "step": 7500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6971930918268436e-05,
+      "loss": 0.254,
+      "step": 7501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6971074079468702e-05,
+      "loss": 0.2829,
+      "step": 7502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697021714109394e-05,
+      "loss": 0.2956,
+      "step": 7503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6969360103156393e-05,
+      "loss": 0.3083,
+      "step": 7504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968502965668298e-05,
+      "loss": 0.2735,
+      "step": 7505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6967645728641905e-05,
+      "loss": 0.2861,
+      "step": 7506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966788392089456e-05,
+      "loss": 0.2649,
+      "step": 7507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965930956023192e-05,
+      "loss": 0.2577,
+      "step": 7508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965073420455365e-05,
+      "loss": 0.2446,
+      "step": 7509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964215785398222e-05,
+      "loss": 0.293,
+      "step": 7510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696335805086402e-05,
+      "loss": 0.2696,
+      "step": 7511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962500216865004e-05,
+      "loss": 0.2541,
+      "step": 7512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961642283413435e-05,
+      "loss": 0.2802,
+      "step": 7513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696078425052156e-05,
+      "loss": 0.2562,
+      "step": 7514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695992611820163e-05,
+      "loss": 0.2567,
+      "step": 7515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6959067886465917e-05,
+      "loss": 0.2553,
+      "step": 7516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958209555326673e-05,
+      "loss": 0.2603,
+      "step": 7517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6957351124796157e-05,
+      "loss": 0.2827,
+      "step": 7518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956492594886632e-05,
+      "loss": 0.2879,
+      "step": 7519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6955633965610365e-05,
+      "loss": 0.2531,
+      "step": 7520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6954775236979616e-05,
+      "loss": 0.2555,
+      "step": 7521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695391640900665e-05,
+      "loss": 0.278,
+      "step": 7522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695305748170374e-05,
+      "loss": 0.2899,
+      "step": 7523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695219845508315e-05,
+      "loss": 0.2613,
+      "step": 7524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951339329157154e-05,
+      "loss": 0.2854,
+      "step": 7525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695048010393802e-05,
+      "loss": 0.2732,
+      "step": 7526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949620779438027e-05,
+      "loss": 0.2614,
+      "step": 7527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6948761355669442e-05,
+      "loss": 0.2589,
+      "step": 7528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947901832644544e-05,
+      "loss": 0.2575,
+      "step": 7529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947042210375615e-05,
+      "loss": 0.2784,
+      "step": 7530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946182488874926e-05,
+      "loss": 0.2436,
+      "step": 7531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6945322668154766e-05,
+      "loss": 0.3097,
+      "step": 7532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944462748227414e-05,
+      "loss": 0.266,
+      "step": 7533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943602729105148e-05,
+      "loss": 0.2853,
+      "step": 7534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942742610800254e-05,
+      "loss": 0.2922,
+      "step": 7535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6941882393325023e-05,
+      "loss": 0.2759,
+      "step": 7536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6941022076691738e-05,
+      "loss": 0.2412,
+      "step": 7537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6940161660912688e-05,
+      "loss": 0.2544,
+      "step": 7538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939301146000167e-05,
+      "loss": 0.2743,
+      "step": 7539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938440531966462e-05,
+      "loss": 0.2422,
+      "step": 7540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6937579818823868e-05,
+      "loss": 0.2491,
+      "step": 7541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693671900658468e-05,
+      "loss": 0.2848,
+      "step": 7542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6935858095261195e-05,
+      "loss": 0.2795,
+      "step": 7543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934997084865706e-05,
+      "loss": 0.2606,
+      "step": 7544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934135975410514e-05,
+      "loss": 0.2783,
+      "step": 7545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693327476690792e-05,
+      "loss": 0.2658,
+      "step": 7546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932413459370226e-05,
+      "loss": 0.2909,
+      "step": 7547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931552052809734e-05,
+      "loss": 0.2794,
+      "step": 7548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930690547238747e-05,
+      "loss": 0.2887,
+      "step": 7549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6929828942669572e-05,
+      "loss": 0.251,
+      "step": 7550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6928967239114514e-05,
+      "loss": 0.3226,
+      "step": 7551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6928105436585885e-05,
+      "loss": 0.2793,
+      "step": 7552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 0.274,
+      "step": 7553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926381534657153e-05,
+      "loss": 0.2478,
+      "step": 7554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6925519435281678e-05,
+      "loss": 0.2861,
+      "step": 7555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692465723698187e-05,
+      "loss": 0.3041,
+      "step": 7556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692379493977006e-05,
+      "loss": 0.2897,
+      "step": 7557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6922932543658557e-05,
+      "loss": 0.2774,
+      "step": 7558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6922070048659684e-05,
+      "loss": 0.2498,
+      "step": 7559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921207454785754e-05,
+      "loss": 0.2661,
+      "step": 7560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69203447620491e-05,
+      "loss": 0.3175,
+      "step": 7561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691948197046203e-05,
+      "loss": 0.2916,
+      "step": 7562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6918619080036878e-05,
+      "loss": 0.2666,
+      "step": 7563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691775609078597e-05,
+      "loss": 0.2841,
+      "step": 7564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6916893002721627e-05,
+      "loss": 0.2847,
+      "step": 7565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691602981585618e-05,
+      "loss": 0.284,
+      "step": 7566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6915166530201958e-05,
+      "loss": 0.2567,
+      "step": 7567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6914303145771297e-05,
+      "loss": 0.2668,
+      "step": 7568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6913439662576523e-05,
+      "loss": 0.2817,
+      "step": 7569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912576080629977e-05,
+      "loss": 0.2704,
+      "step": 7570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691171239994399e-05,
+      "loss": 0.2931,
+      "step": 7571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6910848620530895e-05,
+      "loss": 0.2681,
+      "step": 7572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690998474240304e-05,
+      "loss": 0.2879,
+      "step": 7573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909120765572754e-05,
+      "loss": 0.2601,
+      "step": 7574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908256690052387e-05,
+      "loss": 0.2774,
+      "step": 7575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6907392515854274e-05,
+      "loss": 0.2877,
+      "step": 7576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690652824299077e-05,
+      "loss": 0.2742,
+      "step": 7577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905663871474203e-05,
+      "loss": 0.301,
+      "step": 7578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904799401316938e-05,
+      "loss": 0.2873,
+      "step": 7579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903934832531312e-05,
+      "loss": 0.2862,
+      "step": 7580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903070165129675e-05,
+      "loss": 0.2728,
+      "step": 7581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6902205399124384e-05,
+      "loss": 0.2141,
+      "step": 7582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901340534527785e-05,
+      "loss": 0.2783,
+      "step": 7583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900475571352234e-05,
+      "loss": 0.2441,
+      "step": 7584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689961050961009e-05,
+      "loss": 0.2591,
+      "step": 7585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6898745349313705e-05,
+      "loss": 0.2838,
+      "step": 7586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897880090475436e-05,
+      "loss": 0.2295,
+      "step": 7587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897014733107645e-05,
+      "loss": 0.2635,
+      "step": 7588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896149277222692e-05,
+      "loss": 0.2661,
+      "step": 7589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6895283722832944e-05,
+      "loss": 0.2906,
+      "step": 7590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894418069950754e-05,
+      "loss": 0.2973,
+      "step": 7591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893552318588494e-05,
+      "loss": 0.2776,
+      "step": 7592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892686468758533e-05,
+      "loss": 0.2613,
+      "step": 7593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6891820520473233e-05,
+      "loss": 0.2707,
+      "step": 7594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890954473744964e-05,
+      "loss": 0.2938,
+      "step": 7595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68900883285861e-05,
+      "loss": 0.3178,
+      "step": 7596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688922208500901e-05,
+      "loss": 0.2704,
+      "step": 7597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688835574302607e-05,
+      "loss": 0.2637,
+      "step": 7598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887489302649657e-05,
+      "loss": 0.2671,
+      "step": 7599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688662276389214e-05,
+      "loss": 0.2708,
+      "step": 7600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6885756126765897e-05,
+      "loss": 0.2825,
+      "step": 7601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884889391283317e-05,
+      "loss": 0.2557,
+      "step": 7602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688402255745677e-05,
+      "loss": 0.2581,
+      "step": 7603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883155625298642e-05,
+      "loss": 0.2594,
+      "step": 7604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882288594821318e-05,
+      "loss": 0.2623,
+      "step": 7605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688142146603718e-05,
+      "loss": 0.2741,
+      "step": 7606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880554238958616e-05,
+      "loss": 0.2928,
+      "step": 7607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687968691359801e-05,
+      "loss": 0.284,
+      "step": 7608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6878819489967754e-05,
+      "loss": 0.2636,
+      "step": 7609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6877951968080236e-05,
+      "loss": 0.2661,
+      "step": 7610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6877084347947854e-05,
+      "loss": 0.2609,
+      "step": 7611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876216629582995e-05,
+      "loss": 0.235,
+      "step": 7612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875348812998054e-05,
+      "loss": 0.2496,
+      "step": 7613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6874480898205427e-05,
+      "loss": 0.2835,
+      "step": 7614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687361288521751e-05,
+      "loss": 0.2697,
+      "step": 7615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687274477404671e-05,
+      "loss": 0.2782,
+      "step": 7616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687187656470542e-05,
+      "loss": 0.2399,
+      "step": 7617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871008257206037e-05,
+      "loss": 0.2639,
+      "step": 7618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6870139851560975e-05,
+      "loss": 0.2645,
+      "step": 7619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6869271347782633e-05,
+      "loss": 0.2449,
+      "step": 7620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686840274588341e-05,
+      "loss": 0.286,
+      "step": 7621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867534045875725e-05,
+      "loss": 0.2891,
+      "step": 7622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686666524777198e-05,
+      "loss": 0.2837,
+      "step": 7623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686579635158459e-05,
+      "loss": 0.3147,
+      "step": 7624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686492735732596e-05,
+      "loss": 0.2945,
+      "step": 7625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864058265008503e-05,
+      "loss": 0.2721,
+      "step": 7626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6863189074644638e-05,
+      "loss": 0.2754,
+      "step": 7627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6862319786246778e-05,
+      "loss": 0.2726,
+      "step": 7628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861450399827338e-05,
+      "loss": 0.2581,
+      "step": 7629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6860580915398736e-05,
+      "loss": 0.2707,
+      "step": 7630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68597113329734e-05,
+      "loss": 0.2627,
+      "step": 7631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.685884165256374e-05,
+      "loss": 0.2542,
+      "step": 7632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6857971874182187e-05,
+      "loss": 0.2954,
+      "step": 7633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685710199784116e-05,
+      "loss": 0.2742,
+      "step": 7634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6856232023553087e-05,
+      "loss": 0.2615,
+      "step": 7635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855361951330392e-05,
+      "loss": 0.257,
+      "step": 7636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685449178118551e-05,
+      "loss": 0.3059,
+      "step": 7637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6853621513130857e-05,
+      "loss": 0.2991,
+      "step": 7638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852751147178878e-05,
+      "loss": 0.3134,
+      "step": 7639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851880683342e-05,
+      "loss": 0.2632,
+      "step": 7640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851010121632653e-05,
+      "loss": 0.2719,
+      "step": 7641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850139462063276e-05,
+      "loss": 0.2896,
+      "step": 7642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6849268704646306e-05,
+      "loss": 0.2284,
+      "step": 7643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6848397849394178e-05,
+      "loss": 0.2527,
+      "step": 7644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847526896319337e-05,
+      "loss": 0.2754,
+      "step": 7645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684665584543422e-05,
+      "loss": 0.275,
+      "step": 7646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845784696751268e-05,
+      "loss": 0.2675,
+      "step": 7647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6844913450282922e-05,
+      "loss": 0.3022,
+      "step": 7648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6844042106041635e-05,
+      "loss": 0.3029,
+      "step": 7649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843170664039844e-05,
+      "loss": 0.2767,
+      "step": 7650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6842299124290003e-05,
+      "loss": 0.25,
+      "step": 7651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841427486804565e-05,
+      "loss": 0.304,
+      "step": 7652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684055575159597e-05,
+      "loss": 0.2654,
+      "step": 7653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6839683918676675e-05,
+      "loss": 0.2925,
+      "step": 7654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838811988059138e-05,
+      "loss": 0.3284,
+      "step": 7655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837939959755804e-05,
+      "loss": 0.289,
+      "step": 7656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837067833779136e-05,
+      "loss": 0.2641,
+      "step": 7657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683619561014159e-05,
+      "loss": 0.2923,
+      "step": 7658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683532328885562e-05,
+      "loss": 0.3006,
+      "step": 7659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6834450869933695e-05,
+      "loss": 0.2607,
+      "step": 7660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6833578353388272e-05,
+      "loss": 0.2579,
+      "step": 7661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832705739231816e-05,
+      "loss": 0.298,
+      "step": 7662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831833027476787e-05,
+      "loss": 0.297,
+      "step": 7663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6830960218135652e-05,
+      "loss": 0.2735,
+      "step": 7664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6830087311220885e-05,
+      "loss": 0.2949,
+      "step": 7665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6829214306744945e-05,
+      "loss": 0.2685,
+      "step": 7666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682834120472031e-05,
+      "loss": 0.277,
+      "step": 7667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827468005159447e-05,
+      "loss": 0.2962,
+      "step": 7668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6826594708074827e-05,
+      "loss": 0.2661,
+      "step": 7669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682572131347893e-05,
+      "loss": 0.2981,
+      "step": 7670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6824847821384228e-05,
+      "loss": 0.2598,
+      "step": 7671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823974231803196e-05,
+      "loss": 0.263,
+      "step": 7672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823100544748317e-05,
+      "loss": 0.2993,
+      "step": 7673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822226760232068e-05,
+      "loss": 0.2611,
+      "step": 7674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682135287826693e-05,
+      "loss": 0.3059,
+      "step": 7675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682047889886539e-05,
+      "loss": 0.2929,
+      "step": 7676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819604822039924e-05,
+      "loss": 0.306,
+      "step": 7677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818730647803027e-05,
+      "loss": 0.2708,
+      "step": 7678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681785637616718e-05,
+      "loss": 0.2858,
+      "step": 7679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6816982007144868e-05,
+      "loss": 0.2996,
+      "step": 7680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6816107540748584e-05,
+      "loss": 0.2844,
+      "step": 7681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681523297699082e-05,
+      "loss": 0.2608,
+      "step": 7682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814358315884066e-05,
+      "loss": 0.2739,
+      "step": 7683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6813483557440817e-05,
+      "loss": 0.268,
+      "step": 7684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6812608701673573e-05,
+      "loss": 0.2998,
+      "step": 7685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681173374859482e-05,
+      "loss": 0.2797,
+      "step": 7686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681085869821706e-05,
+      "loss": 0.2553,
+      "step": 7687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6809983550552798e-05,
+      "loss": 0.2828,
+      "step": 7688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680910830561453e-05,
+      "loss": 0.2624,
+      "step": 7689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808232963414756e-05,
+      "loss": 0.2674,
+      "step": 7690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807357523965986e-05,
+      "loss": 0.2381,
+      "step": 7691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6806481987280716e-05,
+      "loss": 0.296,
+      "step": 7692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680560635337146e-05,
+      "loss": 0.2598,
+      "step": 7693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680473062225072e-05,
+      "loss": 0.2506,
+      "step": 7694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6803854793931006e-05,
+      "loss": 0.2441,
+      "step": 7695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680297886842483e-05,
+      "loss": 0.2719,
+      "step": 7696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6802102845744707e-05,
+      "loss": 0.2436,
+      "step": 7697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801226725903142e-05,
+      "loss": 0.2369,
+      "step": 7698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800350508912657e-05,
+      "loss": 0.2642,
+      "step": 7699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799474194785763e-05,
+      "loss": 0.2934,
+      "step": 7700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798597783534977e-05,
+      "loss": 0.2805,
+      "step": 7701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6797721275172823e-05,
+      "loss": 0.2956,
+      "step": 7702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679684466971182e-05,
+      "loss": 0.2816,
+      "step": 7703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795967967164482e-05,
+      "loss": 0.3134,
+      "step": 7704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795091167543344e-05,
+      "loss": 0.2507,
+      "step": 7705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6794214270860917e-05,
+      "loss": 0.2652,
+      "step": 7706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6793337277129738e-05,
+      "loss": 0.2648,
+      "step": 7707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6792460186362325e-05,
+      "loss": 0.2495,
+      "step": 7708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6791582998571212e-05,
+      "loss": 0.2824,
+      "step": 7709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790705713768927e-05,
+      "loss": 0.2931,
+      "step": 7710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789828331968004e-05,
+      "loss": 0.2709,
+      "step": 7711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678895085318097e-05,
+      "loss": 0.2906,
+      "step": 7712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788073277420362e-05,
+      "loss": 0.2648,
+      "step": 7713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787195604698715e-05,
+      "loss": 0.2461,
+      "step": 7714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786317835028568e-05,
+      "loss": 0.2739,
+      "step": 7715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785439968422456e-05,
+      "loss": 0.284,
+      "step": 7716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6784562004892918e-05,
+      "loss": 0.2477,
+      "step": 7717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67836839444525e-05,
+      "loss": 0.274,
+      "step": 7718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782805787113737e-05,
+      "loss": 0.2718,
+      "step": 7719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781927532889176e-05,
+      "loss": 0.2775,
+      "step": 7720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781049181791368e-05,
+      "loss": 0.2919,
+      "step": 7721
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6780170733832845e-05,
+      "loss": 0.288,
+      "step": 7722
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6779292189026168e-05,
+      "loss": 0.2528,
+      "step": 7723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778413547383886e-05,
+      "loss": 0.2978,
+      "step": 7724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777534808918538e-05,
+      "loss": 0.2726,
+      "step": 7725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776655973642687e-05,
+      "loss": 0.2329,
+      "step": 7726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677577704156888e-05,
+      "loss": 0.2983,
+      "step": 7727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774898012709674e-05,
+      "loss": 0.2581,
+      "step": 7728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774018887077624e-05,
+      "loss": 0.286,
+      "step": 7729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677313966468529e-05,
+      "loss": 0.2753,
+      "step": 7730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677226034554523e-05,
+      "loss": 0.2761,
+      "step": 7731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6771380929670005e-05,
+      "loss": 0.2687,
+      "step": 7732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677050141707217e-05,
+      "loss": 0.2841,
+      "step": 7733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67696218077643e-05,
+      "loss": 0.2772,
+      "step": 7734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676874210175895e-05,
+      "loss": 0.2591,
+      "step": 7735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6767862299068684e-05,
+      "loss": 0.295,
+      "step": 7736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676698239970608e-05,
+      "loss": 0.2953,
+      "step": 7737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6766102403683696e-05,
+      "loss": 0.2482,
+      "step": 7738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765222311014103e-05,
+      "loss": 0.2707,
+      "step": 7739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764342121709883e-05,
+      "loss": 0.273,
+      "step": 7740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6763461835783594e-05,
+      "loss": 0.293,
+      "step": 7741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676258145324782e-05,
+      "loss": 0.2793,
+      "step": 7742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6761700974115132e-05,
+      "loss": 0.2895,
+      "step": 7743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676082039839811e-05,
+      "loss": 0.2584,
+      "step": 7744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675993972610933e-05,
+      "loss": 0.2322,
+      "step": 7745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675905895726137e-05,
+      "loss": 0.2654,
+      "step": 7746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675817809186681e-05,
+      "loss": 0.2714,
+      "step": 7747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757297129938237e-05,
+      "loss": 0.2794,
+      "step": 7748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756416071488234e-05,
+      "loss": 0.3007,
+      "step": 7749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6755534916529387e-05,
+      "loss": 0.2654,
+      "step": 7750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754653665074277e-05,
+      "loss": 0.276,
+      "step": 7751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753772317135494e-05,
+      "loss": 0.3072,
+      "step": 7752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752890872725628e-05,
+      "loss": 0.2661,
+      "step": 7753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752009331857273e-05,
+      "loss": 0.2643,
+      "step": 7754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6751127694543012e-05,
+      "loss": 0.2964,
+      "step": 7755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675024596079545e-05,
+      "loss": 0.2806,
+      "step": 7756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749364130627172e-05,
+      "loss": 0.2771,
+      "step": 7757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674848220405078e-05,
+      "loss": 0.2591,
+      "step": 7758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6747600181078868e-05,
+      "loss": 0.2619,
+      "step": 7759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746718061724036e-05,
+      "loss": 0.2704,
+      "step": 7760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6745835845998887e-05,
+      "loss": 0.2877,
+      "step": 7761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674495353391602e-05,
+      "loss": 0.2794,
+      "step": 7762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674407112548803e-05,
+      "loss": 0.275,
+      "step": 7763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743188620727536e-05,
+      "loss": 0.2888,
+      "step": 7764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674230601964714e-05,
+      "loss": 0.2765,
+      "step": 7765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674142332225944e-05,
+      "loss": 0.2793,
+      "step": 7766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740540528577053e-05,
+      "loss": 0.2566,
+      "step": 7767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739657638612587e-05,
+      "loss": 0.2808,
+      "step": 7768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673877465237865e-05,
+      "loss": 0.2898,
+      "step": 7769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673789156988786e-05,
+      "loss": 0.2758,
+      "step": 7770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6737008391152826e-05,
+      "loss": 0.2972,
+      "step": 7771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6736125116186167e-05,
+      "loss": 0.2578,
+      "step": 7772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735241745000498e-05,
+      "loss": 0.2747,
+      "step": 7773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734358277608437e-05,
+      "loss": 0.2608,
+      "step": 7774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733474714022602e-05,
+      "loss": 0.2862,
+      "step": 7775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673259105425562e-05,
+      "loss": 0.2725,
+      "step": 7776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673170729832011e-05,
+      "loss": 0.27,
+      "step": 7777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730823446228685e-05,
+      "loss": 0.2573,
+      "step": 7778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729939497993987e-05,
+      "loss": 0.2558,
+      "step": 7779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729055453628636e-05,
+      "loss": 0.2473,
+      "step": 7780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6728171313145252e-05,
+      "loss": 0.2864,
+      "step": 7781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6727287076556477e-05,
+      "loss": 0.2904,
+      "step": 7782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672640274387493e-05,
+      "loss": 0.267,
+      "step": 7783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6725518315113255e-05,
+      "loss": 0.2777,
+      "step": 7784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724633790284074e-05,
+      "loss": 0.3115,
+      "step": 7785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6723749169400024e-05,
+      "loss": 0.2734,
+      "step": 7786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6722864452473745e-05,
+      "loss": 0.2928,
+      "step": 7787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672197963951787e-05,
+      "loss": 0.2883,
+      "step": 7788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672109473054504e-05,
+      "loss": 0.2612,
+      "step": 7789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67202097255679e-05,
+      "loss": 0.2453,
+      "step": 7790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671932462459908e-05,
+      "loss": 0.2938,
+      "step": 7791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671843942765123e-05,
+      "loss": 0.2907,
+      "step": 7792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6717554134736994e-05,
+      "loss": 0.2789,
+      "step": 7793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671666874586902e-05,
+      "loss": 0.2754,
+      "step": 7794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715783261059946e-05,
+      "loss": 0.2834,
+      "step": 7795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714897680322426e-05,
+      "loss": 0.2835,
+      "step": 7796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714012003669115e-05,
+      "loss": 0.2624,
+      "step": 7797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713126231112654e-05,
+      "loss": 0.274,
+      "step": 7798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6712240362665703e-05,
+      "loss": 0.2529,
+      "step": 7799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671135439834091e-05,
+      "loss": 0.292,
+      "step": 7800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6710468338150936e-05,
+      "loss": 0.2905,
+      "step": 7801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670958218210843e-05,
+      "loss": 0.2707,
+      "step": 7802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670869593022606e-05,
+      "loss": 0.26,
+      "step": 7803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6707809582516478e-05,
+      "loss": 0.2494,
+      "step": 7804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706923138992344e-05,
+      "loss": 0.2907,
+      "step": 7805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706036599666324e-05,
+      "loss": 0.2652,
+      "step": 7806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670514996455108e-05,
+      "loss": 0.2981,
+      "step": 7807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670426323365927e-05,
+      "loss": 0.2879,
+      "step": 7808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6703376407003574e-05,
+      "loss": 0.2661,
+      "step": 7809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6702489484596648e-05,
+      "loss": 0.2941,
+      "step": 7810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6701602466451166e-05,
+      "loss": 0.2595,
+      "step": 7811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700715352579797e-05,
+      "loss": 0.2836,
+      "step": 7812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.669982814299521e-05,
+      "loss": 0.3018,
+      "step": 7813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698940837710082e-05,
+      "loss": 0.2812,
+      "step": 7814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698053436737086e-05,
+      "loss": 0.2661,
+      "step": 7815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66971659400889e-05,
+      "loss": 0.253,
+      "step": 7816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6696278347778194e-05,
+      "loss": 0.289,
+      "step": 7817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695390659817655e-05,
+      "loss": 0.2654,
+      "step": 7818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669450287621995e-05,
+      "loss": 0.2659,
+      "step": 7819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693614996997776e-05,
+      "loss": 0.2788,
+      "step": 7820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6692727022163807e-05,
+      "loss": 0.282,
+      "step": 7821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691838951730728e-05,
+      "loss": 0.2883,
+      "step": 7822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690950785711224e-05,
+      "loss": 0.2476,
+      "step": 7823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669006252411798e-05,
+      "loss": 0.285,
+      "step": 7824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6689174166963687e-05,
+      "loss": 0.2614,
+      "step": 7825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668828571426103e-05,
+      "loss": 0.2401,
+      "step": 7826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687397166022707e-05,
+      "loss": 0.2918,
+      "step": 7827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6686508522261405e-05,
+      "loss": 0.2655,
+      "step": 7828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668561978298982e-05,
+      "loss": 0.2926,
+      "step": 7829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668473094822064e-05,
+      "loss": 0.2743,
+      "step": 7830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668384201796657e-05,
+      "loss": 0.2903,
+      "step": 7831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66829529922403e-05,
+      "loss": 0.2451,
+      "step": 7832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6682063871054534e-05,
+      "loss": 0.265,
+      "step": 7833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6681174654421973e-05,
+      "loss": 0.2699,
+      "step": 7834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680285342355314e-05,
+      "loss": 0.296,
+      "step": 7835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679395934867263e-05,
+      "loss": 0.2647,
+      "step": 7836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6678506431970523e-05,
+      "loss": 0.2596,
+      "step": 7837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66776168336778e-05,
+      "loss": 0.2594,
+      "step": 7838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6676727140001803e-05,
+      "loss": 0.2615,
+      "step": 7839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667583735095524e-05,
+      "loss": 0.2609,
+      "step": 7840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674947466550816e-05,
+      "loss": 0.2572,
+      "step": 7841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674057486801245e-05,
+      "loss": 0.2645,
+      "step": 7842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667316741171924e-05,
+      "loss": 0.2792,
+      "step": 7843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667227724131752e-05,
+      "loss": 0.2873,
+      "step": 7844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667138697560879e-05,
+      "loss": 0.2706,
+      "step": 7845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670496614605776e-05,
+      "loss": 0.2519,
+      "step": 7846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669606158321188e-05,
+      "loss": 0.2793,
+      "step": 7847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666871560676775e-05,
+      "loss": 0.2785,
+      "step": 7848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6667824959958183e-05,
+      "loss": 0.2702,
+      "step": 7849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666693421790521e-05,
+      "loss": 0.2489,
+      "step": 7850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666604338062155e-05,
+      "loss": 0.2541,
+      "step": 7851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6665152448119924e-05,
+      "loss": 0.2338,
+      "step": 7852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666426142041307e-05,
+      "loss": 0.2662,
+      "step": 7853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6663370297513707e-05,
+      "loss": 0.2685,
+      "step": 7854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662479079434568e-05,
+      "loss": 0.2535,
+      "step": 7855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6661587766188387e-05,
+      "loss": 0.2654,
+      "step": 7856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666069635778788e-05,
+      "loss": 0.2969,
+      "step": 7857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6659804854245792e-05,
+      "loss": 0.2475,
+      "step": 7858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658913255574858e-05,
+      "loss": 0.2693,
+      "step": 7859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665802156178781e-05,
+      "loss": 0.2678,
+      "step": 7860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6657129772897388e-05,
+      "loss": 0.2856,
+      "step": 7861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6656237888916327e-05,
+      "loss": 0.2575,
+      "step": 7862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6655345909857367e-05,
+      "loss": 0.2764,
+      "step": 7863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665445383573325e-05,
+      "loss": 0.2679,
+      "step": 7864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653561666556714e-05,
+      "loss": 0.2672,
+      "step": 7865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665266940234051e-05,
+      "loss": 0.2677,
+      "step": 7866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665177704309738e-05,
+      "loss": 0.3125,
+      "step": 7867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650884588840073e-05,
+      "loss": 0.2819,
+      "step": 7868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6649992039581334e-05,
+      "loss": 0.2808,
+      "step": 7869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664909939533391e-05,
+      "loss": 0.2827,
+      "step": 7870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648206656110556e-05,
+      "loss": 0.3358,
+      "step": 7871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6647313821924022e-05,
+      "loss": 0.2879,
+      "step": 7872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664642089278706e-05,
+      "loss": 0.2729,
+      "step": 7873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664552786871243e-05,
+      "loss": 0.2801,
+      "step": 7874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664463474971288e-05,
+      "loss": 0.2303,
+      "step": 7875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6643741535801174e-05,
+      "loss": 0.2552,
+      "step": 7876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664284822699007e-05,
+      "loss": 0.2654,
+      "step": 7877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664195482329232e-05,
+      "loss": 0.2705,
+      "step": 7878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6641061324720694e-05,
+      "loss": 0.2676,
+      "step": 7879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640167731287952e-05,
+      "loss": 0.2677,
+      "step": 7880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6639274043006862e-05,
+      "loss": 0.273,
+      "step": 7881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638380259890184e-05,
+      "loss": 0.2866,
+      "step": 7882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6637486381950686e-05,
+      "loss": 0.2851,
+      "step": 7883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6636592409201136e-05,
+      "loss": 0.2861,
+      "step": 7884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635698341654305e-05,
+      "loss": 0.254,
+      "step": 7885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634804179322963e-05,
+      "loss": 0.2605,
+      "step": 7886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633909922219882e-05,
+      "loss": 0.2688,
+      "step": 7887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633015570357837e-05,
+      "loss": 0.2788,
+      "step": 7888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6632121123749603e-05,
+      "loss": 0.2723,
+      "step": 7889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 0.2787,
+      "step": 7890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630331946345667e-05,
+      "loss": 0.2822,
+      "step": 7891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6629437215575528e-05,
+      "loss": 0.2645,
+      "step": 7892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662854239011031e-05,
+      "loss": 0.2449,
+      "step": 7893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66276474699628e-05,
+      "loss": 0.2796,
+      "step": 7894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626752455145775e-05,
+      "loss": 0.2632,
+      "step": 7895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662585734567202e-05,
+      "loss": 0.2451,
+      "step": 7896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624962141554327e-05,
+      "loss": 0.2875,
+      "step": 7897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624066842805476e-05,
+      "loss": 0.2885,
+      "step": 7898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662317144943826e-05,
+      "loss": 0.2755,
+      "step": 7899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6622275961465472e-05,
+      "loss": 0.2761,
+      "step": 7900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621380378899896e-05,
+      "loss": 0.2843,
+      "step": 7901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662048470175433e-05,
+      "loss": 0.2659,
+      "step": 7902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6619588930041563e-05,
+      "loss": 0.2796,
+      "step": 7903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6618693063774393e-05,
+      "loss": 0.2778,
+      "step": 7904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617797102965614e-05,
+      "loss": 0.223,
+      "step": 7905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661690104762803e-05,
+      "loss": 0.2642,
+      "step": 7906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661600489777443e-05,
+      "loss": 0.2898,
+      "step": 7907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615108653417632e-05,
+      "loss": 0.2601,
+      "step": 7908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661421231457042e-05,
+      "loss": 0.2798,
+      "step": 7909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613315881245604e-05,
+      "loss": 0.3008,
+      "step": 7910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661241935345599e-05,
+      "loss": 0.2803,
+      "step": 7911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661152273121438e-05,
+      "loss": 0.2678,
+      "step": 7912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661062601453359e-05,
+      "loss": 0.2898,
+      "step": 7913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6609729203426418e-05,
+      "loss": 0.2742,
+      "step": 7914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6608832297905682e-05,
+      "loss": 0.2886,
+      "step": 7915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660793529798419e-05,
+      "loss": 0.2734,
+      "step": 7916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6607038203674757e-05,
+      "loss": 0.2923,
+      "step": 7917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606141014990195e-05,
+      "loss": 0.2681,
+      "step": 7918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660524373194332e-05,
+      "loss": 0.2596,
+      "step": 7919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604346354546947e-05,
+      "loss": 0.2586,
+      "step": 7920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6603448882813898e-05,
+      "loss": 0.2906,
+      "step": 7921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660255131675699e-05,
+      "loss": 0.2705,
+      "step": 7922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660165365638904e-05,
+      "loss": 0.2381,
+      "step": 7923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660075590172288e-05,
+      "loss": 0.2838,
+      "step": 7924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659985805277133e-05,
+      "loss": 0.2837,
+      "step": 7925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6598960109547207e-05,
+      "loss": 0.2569,
+      "step": 7926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6598062072063346e-05,
+      "loss": 0.2703,
+      "step": 7927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6597163940332572e-05,
+      "loss": 0.2484,
+      "step": 7928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6596265714367716e-05,
+      "loss": 0.2619,
+      "step": 7929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6595367394181602e-05,
+      "loss": 0.2761,
+      "step": 7930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594468979787067e-05,
+      "loss": 0.2526,
+      "step": 7931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6593570471196945e-05,
+      "loss": 0.2739,
+      "step": 7932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592671868424065e-05,
+      "loss": 0.2754,
+      "step": 7933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591773171481267e-05,
+      "loss": 0.3065,
+      "step": 7934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590874380381388e-05,
+      "loss": 0.2588,
+      "step": 7935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6589975495137264e-05,
+      "loss": 0.2502,
+      "step": 7936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6589076515761734e-05,
+      "loss": 0.2598,
+      "step": 7937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588177442267644e-05,
+      "loss": 0.2766,
+      "step": 7938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658727827466783e-05,
+      "loss": 0.2854,
+      "step": 7939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658637901297514e-05,
+      "loss": 0.268,
+      "step": 7940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658547965720242e-05,
+      "loss": 0.2905,
+      "step": 7941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6584580207362514e-05,
+      "loss": 0.2584,
+      "step": 7942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583680663468267e-05,
+      "loss": 0.2725,
+      "step": 7943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6582781025532536e-05,
+      "loss": 0.3059,
+      "step": 7944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658188129356816e-05,
+      "loss": 0.2847,
+      "step": 7945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580981467588002e-05,
+      "loss": 0.2965,
+      "step": 7946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580081547604915e-05,
+      "loss": 0.2361,
+      "step": 7947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6579181533631743e-05,
+      "loss": 0.2811,
+      "step": 7948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657828142568135e-05,
+      "loss": 0.2608,
+      "step": 7949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577381223766592e-05,
+      "loss": 0.3074,
+      "step": 7950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6576480927900326e-05,
+      "loss": 0.2545,
+      "step": 7951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575580538095413e-05,
+      "loss": 0.2917,
+      "step": 7952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657468005436471e-05,
+      "loss": 0.2638,
+      "step": 7953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657377947672109e-05,
+      "loss": 0.2706,
+      "step": 7954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6572878805177402e-05,
+      "loss": 0.2755,
+      "step": 7955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657197803974653e-05,
+      "loss": 0.2945,
+      "step": 7956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6571077180441323e-05,
+      "loss": 0.2926,
+      "step": 7957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6570176227274654e-05,
+      "loss": 0.3032,
+      "step": 7958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569275180259398e-05,
+      "loss": 0.2501,
+      "step": 7959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6568374039408422e-05,
+      "loss": 0.303,
+      "step": 7960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567472804734595e-05,
+      "loss": 0.2778,
+      "step": 7961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6566571476250796e-05,
+      "loss": 0.2485,
+      "step": 7962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656567005396989e-05,
+      "loss": 0.2472,
+      "step": 7963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6564768537904767e-05,
+      "loss": 0.2724,
+      "step": 7964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6563866928068294e-05,
+      "loss": 0.2513,
+      "step": 7965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656296522447335e-05,
+      "loss": 0.2629,
+      "step": 7966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656206342713282e-05,
+      "loss": 0.2696,
+      "step": 7967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561161536059583e-05,
+      "loss": 0.2414,
+      "step": 7968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656025955126652e-05,
+      "loss": 0.2841,
+      "step": 7969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655935747276652e-05,
+      "loss": 0.2934,
+      "step": 7970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558455300572455e-05,
+      "loss": 0.2632,
+      "step": 7971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6557553034697228e-05,
+      "loss": 0.3007,
+      "step": 7972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655665067515372e-05,
+      "loss": 0.2628,
+      "step": 7973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655574822195482e-05,
+      "loss": 0.2707,
+      "step": 7974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655484567511342e-05,
+      "loss": 0.2766,
+      "step": 7975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655394303464241e-05,
+      "loss": 0.2945,
+      "step": 7976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655304030055468e-05,
+      "loss": 0.2624,
+      "step": 7977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552137472863137e-05,
+      "loss": 0.2783,
+      "step": 7978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6551234551580664e-05,
+      "loss": 0.2494,
+      "step": 7979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6550331536720165e-05,
+      "loss": 0.2855,
+      "step": 7980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654942842829454e-05,
+      "loss": 0.2854,
+      "step": 7981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6548525226316684e-05,
+      "loss": 0.3007,
+      "step": 7982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65476219307995e-05,
+      "loss": 0.263,
+      "step": 7983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654671854175589e-05,
+      "loss": 0.254,
+      "step": 7984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654581505919876e-05,
+      "loss": 0.2205,
+      "step": 7985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544911483141018e-05,
+      "loss": 0.2942,
+      "step": 7986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544007813595565e-05,
+      "loss": 0.2632,
+      "step": 7987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6543104050575313e-05,
+      "loss": 0.2442,
+      "step": 7988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654220019409317e-05,
+      "loss": 0.2725,
+      "step": 7989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541296244162044e-05,
+      "loss": 0.2902,
+      "step": 7990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654039220079485e-05,
+      "loss": 0.2578,
+      "step": 7991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6539488064004503e-05,
+      "loss": 0.2672,
+      "step": 7992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538583833803914e-05,
+      "loss": 0.2366,
+      "step": 7993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6537679510206e-05,
+      "loss": 0.268,
+      "step": 7994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6536775093223686e-05,
+      "loss": 0.2718,
+      "step": 7995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6535870582869876e-05,
+      "loss": 0.2382,
+      "step": 7996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65349659791575e-05,
+      "loss": 0.2864,
+      "step": 7997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534061282099476e-05,
+      "loss": 0.2781,
+      "step": 7998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6533156491708734e-05,
+      "loss": 0.3124,
+      "step": 7999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532251607998183e-05,
+      "loss": 0.2622,
+      "step": 8000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531346630980766e-05,
+      "loss": 0.2711,
+      "step": 8001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530441560669396e-05,
+      "loss": 0.3127,
+      "step": 8002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529536397077008e-05,
+      "loss": 0.291,
+      "step": 8003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652863114021653e-05,
+      "loss": 0.2725,
+      "step": 8004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652772579010089e-05,
+      "loss": 0.2966,
+      "step": 8005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6526820346743027e-05,
+      "loss": 0.2884,
+      "step": 8006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652591481015587e-05,
+      "loss": 0.2603,
+      "step": 8007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652500918035235e-05,
+      "loss": 0.2837,
+      "step": 8008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524103457345406e-05,
+      "loss": 0.2732,
+      "step": 8009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6523197641147976e-05,
+      "loss": 0.2932,
+      "step": 8010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6522291731772998e-05,
+      "loss": 0.2943,
+      "step": 8011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521385729233412e-05,
+      "loss": 0.2738,
+      "step": 8012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520479633542167e-05,
+      "loss": 0.2459,
+      "step": 8013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6519573444712192e-05,
+      "loss": 0.2687,
+      "step": 8014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651866716275644e-05,
+      "loss": 0.2962,
+      "step": 8015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517760787687854e-05,
+      "loss": 0.2638,
+      "step": 8016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651685431951938e-05,
+      "loss": 0.2793,
+      "step": 8017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515947758263966e-05,
+      "loss": 0.2586,
+      "step": 8018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515041103934564e-05,
+      "loss": 0.2755,
+      "step": 8019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514134356544124e-05,
+      "loss": 0.3055,
+      "step": 8020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513227516105595e-05,
+      "loss": 0.2835,
+      "step": 8021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6512320582631934e-05,
+      "loss": 0.249,
+      "step": 8022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511413556136088e-05,
+      "loss": 0.2808,
+      "step": 8023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510506436631024e-05,
+      "loss": 0.2828,
+      "step": 8024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6509599224129694e-05,
+      "loss": 0.2792,
+      "step": 8025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6508691918645055e-05,
+      "loss": 0.2577,
+      "step": 8026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650778452019007e-05,
+      "loss": 0.283,
+      "step": 8027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65068770287777e-05,
+      "loss": 0.2457,
+      "step": 8028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505969444420904e-05,
+      "loss": 0.2777,
+      "step": 8029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650506176713265e-05,
+      "loss": 0.2985,
+      "step": 8030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504153996925902e-05,
+      "loss": 0.2977,
+      "step": 8031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6503246133813625e-05,
+      "loss": 0.2444,
+      "step": 8032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650233817780879e-05,
+      "loss": 0.2548,
+      "step": 8033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650143012892437e-05,
+      "loss": 0.2816,
+      "step": 8034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6500521987173325e-05,
+      "loss": 0.2595,
+      "step": 8035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499613752568634e-05,
+      "loss": 0.2728,
+      "step": 8036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498705425123263e-05,
+      "loss": 0.2504,
+      "step": 8037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6497797004850197e-05,
+      "loss": 0.3025,
+      "step": 8038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496888491762408e-05,
+      "loss": 0.2672,
+      "step": 8039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495979885872872e-05,
+      "loss": 0.2614,
+      "step": 8040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495071187194564e-05,
+      "loss": 0.2759,
+      "step": 8041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6494162395740472e-05,
+      "loss": 0.3048,
+      "step": 8042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649325351152357e-05,
+      "loss": 0.2747,
+      "step": 8043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6492344534556847e-05,
+      "loss": 0.2765,
+      "step": 8044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649143546485328e-05,
+      "loss": 0.2797,
+      "step": 8045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649052630242586e-05,
+      "loss": 0.2643,
+      "step": 8046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6489617047287568e-05,
+      "loss": 0.2733,
+      "step": 8047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6488707699451398e-05,
+      "loss": 0.2902,
+      "step": 8048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648779825893033e-05,
+      "loss": 0.2682,
+      "step": 8049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486888725737365e-05,
+      "loss": 0.2596,
+      "step": 8050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6485979099885493e-05,
+      "loss": 0.2702,
+      "step": 8051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64850693813877e-05,
+      "loss": 0.2603,
+      "step": 8052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484159570256988e-05,
+      "loss": 0.2576,
+      "step": 8053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483249666506346e-05,
+      "loss": 0.2717,
+      "step": 8054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6482339670148774e-05,
+      "loss": 0.2789,
+      "step": 8055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481429581197275e-05,
+      "loss": 0.2736,
+      "step": 8056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6480519399664844e-05,
+      "loss": 0.2978,
+      "step": 8057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647960912556448e-05,
+      "loss": 0.2707,
+      "step": 8058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647869875890919e-05,
+      "loss": 0.2985,
+      "step": 8059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647778829971198e-05,
+      "loss": 0.2265,
+      "step": 8060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476877747985846e-05,
+      "loss": 0.2756,
+      "step": 8061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64759671037438e-05,
+      "loss": 0.2495,
+      "step": 8062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475056366998848e-05,
+      "loss": 0.2689,
+      "step": 8063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474145537764e-05,
+      "loss": 0.2525,
+      "step": 8064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647323461605227e-05,
+      "loss": 0.2851,
+      "step": 8065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472323601876663e-05,
+      "loss": 0.2905,
+      "step": 8066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6471412495250195e-05,
+      "loss": 0.2744,
+      "step": 8067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647050129618588e-05,
+      "loss": 0.2656,
+      "step": 8068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469590004696735e-05,
+      "loss": 0.2669,
+      "step": 8069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6468678620795773e-05,
+      "loss": 0.2923,
+      "step": 8070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646776714449602e-05,
+      "loss": 0.2702,
+      "step": 8071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6466855575810487e-05,
+      "loss": 0.2669,
+      "step": 8072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64659439147522e-05,
+      "loss": 0.2514,
+      "step": 8073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646503216133418e-05,
+      "loss": 0.2536,
+      "step": 8074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6464120315569448e-05,
+      "loss": 0.2605,
+      "step": 8075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463208377471028e-05,
+      "loss": 0.2521,
+      "step": 8076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6462296347051956e-05,
+      "loss": 0.2648,
+      "step": 8077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646138422432525e-05,
+      "loss": 0.3129,
+      "step": 8078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460472009303935e-05,
+      "loss": 0.26,
+      "step": 8079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459559702001054e-05,
+      "loss": 0.2742,
+      "step": 8080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645864730242963e-05,
+      "loss": 0.264,
+      "step": 8081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64577348106027e-05,
+      "loss": 0.2518,
+      "step": 8082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645682222653329e-05,
+      "loss": 0.2704,
+      "step": 8083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6455909550234446e-05,
+      "loss": 0.2903,
+      "step": 8084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64549967817192e-05,
+      "loss": 0.2682,
+      "step": 8085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6454083921000586e-05,
+      "loss": 0.2997,
+      "step": 8086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645317096809165e-05,
+      "loss": 0.2442,
+      "step": 8087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645225792300543e-05,
+      "loss": 0.2632,
+      "step": 8088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6451344785754965e-05,
+      "loss": 0.3044,
+      "step": 8089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6450431556353302e-05,
+      "loss": 0.3002,
+      "step": 8090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449518234813482e-05,
+      "loss": 0.2812,
+      "step": 8091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6448604821148557e-05,
+      "loss": 0.2911,
+      "step": 8092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644769131537157e-05,
+      "loss": 0.2372,
+      "step": 8093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644677771749557e-05,
+      "loss": 0.2658,
+      "step": 8094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644586402753361e-05,
+      "loss": 0.2494,
+      "step": 8095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444950245498733e-05,
+      "loss": 0.2766,
+      "step": 8096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444036371403997e-05,
+      "loss": 0.2503,
+      "step": 8097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443122405262458e-05,
+      "loss": 0.261,
+      "step": 8098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442208347087168e-05,
+      "loss": 0.3035,
+      "step": 8099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441294196891187e-05,
+      "loss": 0.2704,
+      "step": 8100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6440379954687564e-05,
+      "loss": 0.281,
+      "step": 8101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643946562048937e-05,
+      "loss": 0.2619,
+      "step": 8102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6438551194309654e-05,
+      "loss": 0.254,
+      "step": 8103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437636676161484e-05,
+      "loss": 0.2897,
+      "step": 8104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6436722066057923e-05,
+      "loss": 0.2754,
+      "step": 8105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6435807364012035e-05,
+      "loss": 0.284,
+      "step": 8106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434892570036886e-05,
+      "loss": 0.265,
+      "step": 8107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433977684145536e-05,
+      "loss": 0.2872,
+      "step": 8108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433062706351064e-05,
+      "loss": 0.2698,
+      "step": 8109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432147636666535e-05,
+      "loss": 0.2445,
+      "step": 8110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6431232475105016e-05,
+      "loss": 0.2521,
+      "step": 8111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430317221679585e-05,
+      "loss": 0.262,
+      "step": 8112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642940187640332e-05,
+      "loss": 0.2886,
+      "step": 8113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428486439289276e-05,
+      "loss": 0.2715,
+      "step": 8114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642757091035055e-05,
+      "loss": 0.246,
+      "step": 8115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642665528960021e-05,
+      "loss": 0.2677,
+      "step": 8116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425739577051337e-05,
+      "loss": 0.263,
+      "step": 8117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642482377271701e-05,
+      "loss": 0.2703,
+      "step": 8118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642390787661031e-05,
+      "loss": 0.3121,
+      "step": 8119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422991888744323e-05,
+      "loss": 0.2797,
+      "step": 8120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422075809132127e-05,
+      "loss": 0.2643,
+      "step": 8121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421159637786813e-05,
+      "loss": 0.2903,
+      "step": 8122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420243374721464e-05,
+      "loss": 0.2603,
+      "step": 8123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419327019949172e-05,
+      "loss": 0.2841,
+      "step": 8124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6418410573483023e-05,
+      "loss": 0.2604,
+      "step": 8125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6417494035336107e-05,
+      "loss": 0.2446,
+      "step": 8126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416577405521518e-05,
+      "loss": 0.2807,
+      "step": 8127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641566068405235e-05,
+      "loss": 0.2551,
+      "step": 8128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641474387094169e-05,
+      "loss": 0.2801,
+      "step": 8129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413826966202643e-05,
+      "loss": 0.2833,
+      "step": 8130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412909969848304e-05,
+      "loss": 0.2545,
+      "step": 8131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641199288189177e-05,
+      "loss": 0.2707,
+      "step": 8132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6411075702346138e-05,
+      "loss": 0.2724,
+      "step": 8133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6410158431224513e-05,
+      "loss": 0.2812,
+      "step": 8134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409241068539996e-05,
+      "loss": 0.2638,
+      "step": 8135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6408323614305688e-05,
+      "loss": 0.2942,
+      "step": 8136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64074060685347e-05,
+      "loss": 0.2555,
+      "step": 8137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406488431240133e-05,
+      "loss": 0.281,
+      "step": 8138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6405570702435098e-05,
+      "loss": 0.2697,
+      "step": 8139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404652882132703e-05,
+      "loss": 0.2828,
+      "step": 8140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403734970346053e-05,
+      "loss": 0.2555,
+      "step": 8141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6402816967088265e-05,
+      "loss": 0.298,
+      "step": 8142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6401898872372454e-05,
+      "loss": 0.2778,
+      "step": 8143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640098068621173e-05,
+      "loss": 0.2941,
+      "step": 8144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640006240861921e-05,
+      "loss": 0.294,
+      "step": 8145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639914403960801e-05,
+      "loss": 0.2912,
+      "step": 8146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6398225579191246e-05,
+      "loss": 0.2617,
+      "step": 8147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639730702738204e-05,
+      "loss": 0.2511,
+      "step": 8148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6396388384193513e-05,
+      "loss": 0.2769,
+      "step": 8149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639546964963878e-05,
+      "loss": 0.279,
+      "step": 8150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394550823730973e-05,
+      "loss": 0.2682,
+      "step": 8151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6393631906483215e-05,
+      "loss": 0.2606,
+      "step": 8152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392712897908632e-05,
+      "loss": 0.2883,
+      "step": 8153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6391793798020347e-05,
+      "loss": 0.2923,
+      "step": 8154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639087460683149e-05,
+      "loss": 0.268,
+      "step": 8155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389955324355195e-05,
+      "loss": 0.2508,
+      "step": 8156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638903595060459e-05,
+      "loss": 0.2464,
+      "step": 8157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63881164855928e-05,
+      "loss": 0.2872,
+      "step": 8158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6387196929332976e-05,
+      "loss": 0.2737,
+      "step": 8159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6386277281838236e-05,
+      "loss": 0.2619,
+      "step": 8160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385357543121723e-05,
+      "loss": 0.2978,
+      "step": 8161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638443771319658e-05,
+      "loss": 0.2906,
+      "step": 8162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383517792075938e-05,
+      "loss": 0.2628,
+      "step": 8163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382597779772937e-05,
+      "loss": 0.2615,
+      "step": 8164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6381677676300724e-05,
+      "loss": 0.2654,
+      "step": 8165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380757481672438e-05,
+      "loss": 0.3037,
+      "step": 8166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6379837195901224e-05,
+      "loss": 0.259,
+      "step": 8167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378916819000226e-05,
+      "loss": 0.2694,
+      "step": 8168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377996350982593e-05,
+      "loss": 0.2831,
+      "step": 8169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377075791861474e-05,
+      "loss": 0.2413,
+      "step": 8170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6376155141650015e-05,
+      "loss": 0.2503,
+      "step": 8171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637523440036137e-05,
+      "loss": 0.2947,
+      "step": 8172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6374313568008686e-05,
+      "loss": 0.2747,
+      "step": 8173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637339264460512e-05,
+      "loss": 0.2603,
+      "step": 8174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372471630163826e-05,
+      "loss": 0.2806,
+      "step": 8175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6371550524697958e-05,
+      "loss": 0.2635,
+      "step": 8176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6370629328220674e-05,
+      "loss": 0.2758,
+      "step": 8177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369708040745136e-05,
+      "loss": 0.2667,
+      "step": 8178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63687866622845e-05,
+      "loss": 0.2855,
+      "step": 8179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367865192851927e-05,
+      "loss": 0.3291,
+      "step": 8180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6366943632460577e-05,
+      "loss": 0.2716,
+      "step": 8181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6366021981123622e-05,
+      "loss": 0.2727,
+      "step": 8182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6365100238854218e-05,
+      "loss": 0.2677,
+      "step": 8183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364178405665534e-05,
+      "loss": 0.2921,
+      "step": 8184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636325648157074e-05,
+      "loss": 0.2629,
+      "step": 8185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362334466583e-05,
+      "loss": 0.2793,
+      "step": 8186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636141236071549e-05,
+      "loss": 0.2857,
+      "step": 8187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6360490163981377e-05,
+      "loss": 0.3182,
+      "step": 8188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359567876393835e-05,
+      "loss": 0.2822,
+      "step": 8189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6358645497966037e-05,
+      "loss": 0.268,
+      "step": 8190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635772302871116e-05,
+      "loss": 0.2701,
+      "step": 8191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356800468642383e-05,
+      "loss": 0.2889,
+      "step": 8192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355877817772876e-05,
+      "loss": 0.2602,
+      "step": 8193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6354955076115826e-05,
+      "loss": 0.2747,
+      "step": 8194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6354032243684407e-05,
+      "loss": 0.259,
+      "step": 8195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353109320491808e-05,
+      "loss": 0.2801,
+      "step": 8196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352186306551206e-05,
+      "loss": 0.2829,
+      "step": 8197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6351263201875787e-05,
+      "loss": 0.3049,
+      "step": 8198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635034000647874e-05,
+      "loss": 0.2871,
+      "step": 8199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6349416720373248e-05,
+      "loss": 0.2634,
+      "step": 8200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63484933435725e-05,
+      "loss": 0.2899,
+      "step": 8201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347569876089685e-05,
+      "loss": 0.2854,
+      "step": 8202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6346646317937994e-05,
+      "loss": 0.2882,
+      "step": 8203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345722669130623e-05,
+      "loss": 0.3039,
+      "step": 8204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634479892968076e-05,
+      "loss": 0.2878,
+      "step": 8205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343875099601604e-05,
+      "loss": 0.2859,
+      "step": 8206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6342951178906347e-05,
+      "loss": 0.2665,
+      "step": 8207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634202716760819e-05,
+      "loss": 0.2563,
+      "step": 8208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634110306572033e-05,
+      "loss": 0.2414,
+      "step": 8209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340178873255965e-05,
+      "loss": 0.2736,
+      "step": 8210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63392545902283e-05,
+      "loss": 0.2549,
+      "step": 8211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6338330216650534e-05,
+      "loss": 0.2891,
+      "step": 8212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337405752535876e-05,
+      "loss": 0.2698,
+      "step": 8213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633648119789752e-05,
+      "loss": 0.266,
+      "step": 8214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6335556552748686e-05,
+      "loss": 0.2806,
+      "step": 8215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6334631817102577e-05,
+      "loss": 0.2741,
+      "step": 8216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333706990972396e-05,
+      "loss": 0.2787,
+      "step": 8217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332782074371358e-05,
+      "loss": 0.2671,
+      "step": 8218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331857067312676e-05,
+      "loss": 0.2743,
+      "step": 8219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633093196980956e-05,
+      "loss": 0.272,
+      "step": 8220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6330006781875227e-05,
+      "loss": 0.2691,
+      "step": 8221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6329081503522887e-05,
+      "loss": 0.2453,
+      "step": 8222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632815613476576e-05,
+      "loss": 0.2632,
+      "step": 8223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6327230675617067e-05,
+      "loss": 0.2779,
+      "step": 8224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6326305126090026e-05,
+      "loss": 0.2966,
+      "step": 8225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6325379486197854e-05,
+      "loss": 0.2771,
+      "step": 8226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 0.2836,
+      "step": 8227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632352793537101e-05,
+      "loss": 0.2925,
+      "step": 8228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6322602024462787e-05,
+      "loss": 0.2902,
+      "step": 8229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632167602324233e-05,
+      "loss": 0.2723,
+      "step": 8230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632074993172287e-05,
+      "loss": 0.2452,
+      "step": 8231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631982374991763e-05,
+      "loss": 0.2722,
+      "step": 8232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631889747783984e-05,
+      "loss": 0.2756,
+      "step": 8233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631797111550273e-05,
+      "loss": 0.2685,
+      "step": 8234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631704466291954e-05,
+      "loss": 0.2647,
+      "step": 8235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6316118120103492e-05,
+      "loss": 0.2255,
+      "step": 8236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631519148706783e-05,
+      "loss": 0.3054,
+      "step": 8237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6314264763825786e-05,
+      "loss": 0.3148,
+      "step": 8238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313337950390597e-05,
+      "loss": 0.3127,
+      "step": 8239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312411046775504e-05,
+      "loss": 0.2655,
+      "step": 8240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6311484052993743e-05,
+      "loss": 0.2818,
+      "step": 8241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310556969058562e-05,
+      "loss": 0.2768,
+      "step": 8242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6309629794983196e-05,
+      "loss": 0.2717,
+      "step": 8243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6308702530780895e-05,
+      "loss": 0.2561,
+      "step": 8244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307775176464896e-05,
+      "loss": 0.2741,
+      "step": 8245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6306847732048455e-05,
+      "loss": 0.2912,
+      "step": 8246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305920197544814e-05,
+      "loss": 0.244,
+      "step": 8247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6304992572967223e-05,
+      "loss": 0.2729,
+      "step": 8248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6304064858328928e-05,
+      "loss": 0.2286,
+      "step": 8249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630313705364319e-05,
+      "loss": 0.2882,
+      "step": 8250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302209158923258e-05,
+      "loss": 0.2921,
+      "step": 8251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630128117418238e-05,
+      "loss": 0.2551,
+      "step": 8252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300353099433816e-05,
+      "loss": 0.2276,
+      "step": 8253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6299424934690822e-05,
+      "loss": 0.2449,
+      "step": 8254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629849667996666e-05,
+      "loss": 0.2607,
+      "step": 8255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297568335274584e-05,
+      "loss": 0.2662,
+      "step": 8256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629663990062786e-05,
+      "loss": 0.2797,
+      "step": 8257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629571137603974e-05,
+      "loss": 0.263,
+      "step": 8258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6294782761523497e-05,
+      "loss": 0.2647,
+      "step": 8259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629385405709239e-05,
+      "loss": 0.2425,
+      "step": 8260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6292925262759686e-05,
+      "loss": 0.2745,
+      "step": 8261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291996378538653e-05,
+      "loss": 0.2607,
+      "step": 8262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629106740444256e-05,
+      "loss": 0.262,
+      "step": 8263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6290138340484672e-05,
+      "loss": 0.267,
+      "step": 8264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6289209186678263e-05,
+      "loss": 0.2424,
+      "step": 8265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6288279943036604e-05,
+      "loss": 0.2574,
+      "step": 8266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628735060957297e-05,
+      "loss": 0.2611,
+      "step": 8267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286421186300632e-05,
+      "loss": 0.2742,
+      "step": 8268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285491673232873e-05,
+      "loss": 0.2763,
+      "step": 8269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6284562070382962e-05,
+      "loss": 0.2649,
+      "step": 8270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283632377764184e-05,
+      "loss": 0.2606,
+      "step": 8271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628270259538981e-05,
+      "loss": 0.2788,
+      "step": 8272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6281772723273133e-05,
+      "loss": 0.2739,
+      "step": 8273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628084276142743e-05,
+      "loss": 0.2921,
+      "step": 8274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627991270986598e-05,
+      "loss": 0.2738,
+      "step": 8275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278982568602072e-05,
+      "loss": 0.2752,
+      "step": 8276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278052337648994e-05,
+      "loss": 0.2626,
+      "step": 8277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627712201702003e-05,
+      "loss": 0.2591,
+      "step": 8278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276191606728474e-05,
+      "loss": 0.2962,
+      "step": 8279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275261106787605e-05,
+      "loss": 0.2467,
+      "step": 8280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274330517210728e-05,
+      "loss": 0.2637,
+      "step": 8281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6273399838011124e-05,
+      "loss": 0.2653,
+      "step": 8282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272469069202093e-05,
+      "loss": 0.2824,
+      "step": 8283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627153821079693e-05,
+      "loss": 0.273,
+      "step": 8284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270607262808932e-05,
+      "loss": 0.2967,
+      "step": 8285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6269676225251392e-05,
+      "loss": 0.2638,
+      "step": 8286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6268745098137612e-05,
+      "loss": 0.2581,
+      "step": 8287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267813881480892e-05,
+      "loss": 0.2218,
+      "step": 8288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266882575294535e-05,
+      "loss": 0.323,
+      "step": 8289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6265951179591845e-05,
+      "loss": 0.2675,
+      "step": 8290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626501969438612e-05,
+      "loss": 0.2618,
+      "step": 8291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626408811969067e-05,
+      "loss": 0.2631,
+      "step": 8292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62631564555188e-05,
+      "loss": 0.2638,
+      "step": 8293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626222470188382e-05,
+      "loss": 0.2572,
+      "step": 8294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626129285879904e-05,
+      "loss": 0.2824,
+      "step": 8295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6260360926277763e-05,
+      "loss": 0.284,
+      "step": 8296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625942890433331e-05,
+      "loss": 0.2819,
+      "step": 8297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258496792978987e-05,
+      "loss": 0.2589,
+      "step": 8298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6257564592228113e-05,
+      "loss": 0.2769,
+      "step": 8299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256632302094e-05,
+      "loss": 0.2671,
+      "step": 8300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255699922589968e-05,
+      "loss": 0.2685,
+      "step": 8301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254767453729338e-05,
+      "loss": 0.2832,
+      "step": 8302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625383489552542e-05,
+      "loss": 0.2436,
+      "step": 8303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6252902247991542e-05,
+      "loss": 0.2625,
+      "step": 8304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251969511141024e-05,
+      "loss": 0.2603,
+      "step": 8305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625103668498719e-05,
+      "loss": 0.253,
+      "step": 8306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250103769543358e-05,
+      "loss": 0.2415,
+      "step": 8307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249170764822865e-05,
+      "loss": 0.2872,
+      "step": 8308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248237670839035e-05,
+      "loss": 0.2617,
+      "step": 8309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6247304487605192e-05,
+      "loss": 0.2538,
+      "step": 8310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246371215134664e-05,
+      "loss": 0.289,
+      "step": 8311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624543785344079e-05,
+      "loss": 0.2826,
+      "step": 8312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624450440253689e-05,
+      "loss": 0.3233,
+      "step": 8313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243570862436312e-05,
+      "loss": 0.2812,
+      "step": 8314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242637233152382e-05,
+      "loss": 0.2833,
+      "step": 8315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241703514698438e-05,
+      "loss": 0.2616,
+      "step": 8316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240769707087814e-05,
+      "loss": 0.2396,
+      "step": 8317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239835810333858e-05,
+      "loss": 0.2478,
+      "step": 8318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6238901824449898e-05,
+      "loss": 0.283,
+      "step": 8319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237967749449282e-05,
+      "loss": 0.2642,
+      "step": 8320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623703358534535e-05,
+      "loss": 0.2542,
+      "step": 8321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623609933215145e-05,
+      "loss": 0.267,
+      "step": 8322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6235164989880922e-05,
+      "loss": 0.2861,
+      "step": 8323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234230558547112e-05,
+      "loss": 0.2651,
+      "step": 8324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233296038163366e-05,
+      "loss": 0.2958,
+      "step": 8325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623236142874304e-05,
+      "loss": 0.2807,
+      "step": 8326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6231426730299474e-05,
+      "loss": 0.2814,
+      "step": 8327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6230491942846033e-05,
+      "loss": 0.2534,
+      "step": 8328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229557066396057e-05,
+      "loss": 0.2596,
+      "step": 8329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228622100962905e-05,
+      "loss": 0.2805,
+      "step": 8330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6227687046559932e-05,
+      "loss": 0.2727,
+      "step": 8331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6226751903200493e-05,
+      "loss": 0.2479,
+      "step": 8332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225816670897945e-05,
+      "loss": 0.2777,
+      "step": 8333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622488134966565e-05,
+      "loss": 0.2702,
+      "step": 8334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223945939516963e-05,
+      "loss": 0.3007,
+      "step": 8335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223010440465255e-05,
+      "loss": 0.2592,
+      "step": 8336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222074852523876e-05,
+      "loss": 0.2733,
+      "step": 8337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221139175706203e-05,
+      "loss": 0.2735,
+      "step": 8338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622020341002559e-05,
+      "loss": 0.2492,
+      "step": 8339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6219267555495407e-05,
+      "loss": 0.2893,
+      "step": 8340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218331612129022e-05,
+      "loss": 0.2699,
+      "step": 8341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621739557993981e-05,
+      "loss": 0.2445,
+      "step": 8342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216459458941137e-05,
+      "loss": 0.2666,
+      "step": 8343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621552324914637e-05,
+      "loss": 0.2741,
+      "step": 8344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621458695056889e-05,
+      "loss": 0.2503,
+      "step": 8345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621365056322206e-05,
+      "loss": 0.2738,
+      "step": 8346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6212714087119263e-05,
+      "loss": 0.2561,
+      "step": 8347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6211777522273878e-05,
+      "loss": 0.2614,
+      "step": 8348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621084086869928e-05,
+      "loss": 0.2616,
+      "step": 8349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6209904126408845e-05,
+      "loss": 0.2684,
+      "step": 8350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208967295415964e-05,
+      "loss": 0.263,
+      "step": 8351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208030375734003e-05,
+      "loss": 0.2991,
+      "step": 8352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6207093367376353e-05,
+      "loss": 0.2853,
+      "step": 8353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206156270356403e-05,
+      "loss": 0.2694,
+      "step": 8354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6205219084687534e-05,
+      "loss": 0.2697,
+      "step": 8355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204281810383127e-05,
+      "loss": 0.2677,
+      "step": 8356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6203344447456584e-05,
+      "loss": 0.2791,
+      "step": 8357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202406995921277e-05,
+      "loss": 0.3077,
+      "step": 8358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6201469455790613e-05,
+      "loss": 0.2522,
+      "step": 8359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6200531827077976e-05,
+      "loss": 0.2566,
+      "step": 8360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199594109796754e-05,
+      "loss": 0.2704,
+      "step": 8361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619865630396035e-05,
+      "loss": 0.2577,
+      "step": 8362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6197718409582157e-05,
+      "loss": 0.2928,
+      "step": 8363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619678042667557e-05,
+      "loss": 0.2429,
+      "step": 8364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195842355253993e-05,
+      "loss": 0.2556,
+      "step": 8365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6194904195330816e-05,
+      "loss": 0.276,
+      "step": 8366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619396594691945e-05,
+      "loss": 0.2678,
+      "step": 8367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6193027610033287e-05,
+      "loss": 0.2788,
+      "step": 8368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192089184685736e-05,
+      "loss": 0.2719,
+      "step": 8369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6191150670890205e-05,
+      "loss": 0.2707,
+      "step": 8370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619021206866009e-05,
+      "loss": 0.2698,
+      "step": 8371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189273378008806e-05,
+      "loss": 0.2655,
+      "step": 8372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618833459894976e-05,
+      "loss": 0.2468,
+      "step": 8373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187395731496355e-05,
+      "loss": 0.2599,
+      "step": 8374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618645677566201e-05,
+      "loss": 0.274,
+      "step": 8375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185517731460136e-05,
+      "loss": 0.2827,
+      "step": 8376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6184578598904143e-05,
+      "loss": 0.2963,
+      "step": 8377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618363937800745e-05,
+      "loss": 0.2547,
+      "step": 8378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182700068783463e-05,
+      "loss": 0.2582,
+      "step": 8379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181760671245606e-05,
+      "loss": 0.272,
+      "step": 8380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6180821185407306e-05,
+      "loss": 0.2662,
+      "step": 8381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179881611281966e-05,
+      "loss": 0.2494,
+      "step": 8382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617894194888302e-05,
+      "loss": 0.2341,
+      "step": 8383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6178002198223882e-05,
+      "loss": 0.287,
+      "step": 8384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617706235931798e-05,
+      "loss": 0.282,
+      "step": 8385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6176122432178733e-05,
+      "loss": 0.2568,
+      "step": 8386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617518241681958e-05,
+      "loss": 0.264,
+      "step": 8387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617424231325393e-05,
+      "loss": 0.2687,
+      "step": 8388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173302121495223e-05,
+      "loss": 0.2614,
+      "step": 8389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617236184155689e-05,
+      "loss": 0.2647,
+      "step": 8390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171421473452356e-05,
+      "loss": 0.2477,
+      "step": 8391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617048101719506e-05,
+      "loss": 0.253,
+      "step": 8392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169540472798427e-05,
+      "loss": 0.2708,
+      "step": 8393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168599840275895e-05,
+      "loss": 0.2921,
+      "step": 8394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6167659119640902e-05,
+      "loss": 0.2819,
+      "step": 8395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166718310906887e-05,
+      "loss": 0.2542,
+      "step": 8396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6165777414087283e-05,
+      "loss": 0.2798,
+      "step": 8397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164836429195535e-05,
+      "loss": 0.2595,
+      "step": 8398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616389535624508e-05,
+      "loss": 0.2614,
+      "step": 8399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616295419524936e-05,
+      "loss": 0.274,
+      "step": 8400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6162012946221825e-05,
+      "loss": 0.273,
+      "step": 8401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161071609175912e-05,
+      "loss": 0.2515,
+      "step": 8402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160130184125074e-05,
+      "loss": 0.2689,
+      "step": 8403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6159188671082754e-05,
+      "loss": 0.2679,
+      "step": 8404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6158247070062397e-05,
+      "loss": 0.2729,
+      "step": 8405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6157305381077462e-05,
+      "loss": 0.283,
+      "step": 8406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156363604141393e-05,
+      "loss": 0.2897,
+      "step": 8407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155421739267648e-05,
+      "loss": 0.2692,
+      "step": 8408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154479786469673e-05,
+      "loss": 0.279,
+      "step": 8409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615353774576093e-05,
+      "loss": 0.2697,
+      "step": 8410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615259561715487e-05,
+      "loss": 0.2778,
+      "step": 8411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151653400664956e-05,
+      "loss": 0.2552,
+      "step": 8412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6150711096304642e-05,
+      "loss": 0.2486,
+      "step": 8413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614976870408739e-05,
+      "loss": 0.2623,
+      "step": 8414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614882622402666e-05,
+      "loss": 0.2736,
+      "step": 8415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147883656135917e-05,
+      "loss": 0.2809,
+      "step": 8416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614694100042862e-05,
+      "loss": 0.2821,
+      "step": 8417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145998256918238e-05,
+      "loss": 0.2601,
+      "step": 8418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145055425618236e-05,
+      "loss": 0.2602,
+      "step": 8419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614411250654208e-05,
+      "loss": 0.3001,
+      "step": 8420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614316949970324e-05,
+      "loss": 0.279,
+      "step": 8421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142226405115188e-05,
+      "loss": 0.2721,
+      "step": 8422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614128322279139e-05,
+      "loss": 0.2669,
+      "step": 8423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140339952745324e-05,
+      "loss": 0.2561,
+      "step": 8424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613939659499046e-05,
+      "loss": 0.2601,
+      "step": 8425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6138453149540268e-05,
+      "loss": 0.2689,
+      "step": 8426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613750961640824e-05,
+      "loss": 0.2838,
+      "step": 8427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136565995607837e-05,
+      "loss": 0.2765,
+      "step": 8428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135622287152547e-05,
+      "loss": 0.3001,
+      "step": 8429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613467849105585e-05,
+      "loss": 0.2785,
+      "step": 8430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133734607331223e-05,
+      "loss": 0.2758,
+      "step": 8431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613279063599215e-05,
+      "loss": 0.2639,
+      "step": 8432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613184657705211e-05,
+      "loss": 0.2976,
+      "step": 8433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6130902430524598e-05,
+      "loss": 0.2446,
+      "step": 8434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129958196423093e-05,
+      "loss": 0.2416,
+      "step": 8435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129013874761086e-05,
+      "loss": 0.2695,
+      "step": 8436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128069465552063e-05,
+      "loss": 0.2941,
+      "step": 8437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6127124968809515e-05,
+      "loss": 0.2777,
+      "step": 8438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612618038454693e-05,
+      "loss": 0.2886,
+      "step": 8439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612523571277781e-05,
+      "loss": 0.2813,
+      "step": 8440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612429095351564e-05,
+      "loss": 0.2794,
+      "step": 8441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6123346106773922e-05,
+      "loss": 0.2664,
+      "step": 8442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612240117256614e-05,
+      "loss": 0.2724,
+      "step": 8443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6121456150905802e-05,
+      "loss": 0.2558,
+      "step": 8444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6120511041806405e-05,
+      "loss": 0.289,
+      "step": 8445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6119565845281447e-05,
+      "loss": 0.2654,
+      "step": 8446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118620561344428e-05,
+      "loss": 0.2561,
+      "step": 8447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117675190008858e-05,
+      "loss": 0.267,
+      "step": 8448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611672973128823e-05,
+      "loss": 0.2663,
+      "step": 8449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6115784185196058e-05,
+      "loss": 0.2848,
+      "step": 8450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611483855174584e-05,
+      "loss": 0.2472,
+      "step": 8451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611389283095109e-05,
+      "loss": 0.2597,
+      "step": 8452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112947022825317e-05,
+      "loss": 0.2929,
+      "step": 8453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112001127382026e-05,
+      "loss": 0.2335,
+      "step": 8454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611105514463473e-05,
+      "loss": 0.3001,
+      "step": 8455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6110109074596946e-05,
+      "loss": 0.2651,
+      "step": 8456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610916291728218e-05,
+      "loss": 0.2786,
+      "step": 8457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108216672703954e-05,
+      "loss": 0.274,
+      "step": 8458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107270340875777e-05,
+      "loss": 0.2519,
+      "step": 8459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106323921811172e-05,
+      "loss": 0.28,
+      "step": 8460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6105377415523658e-05,
+      "loss": 0.2591,
+      "step": 8461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6104430822026752e-05,
+      "loss": 0.2767,
+      "step": 8462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6103484141333975e-05,
+      "loss": 0.2818,
+      "step": 8463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6102537373458854e-05,
+      "loss": 0.2743,
+      "step": 8464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101590518414905e-05,
+      "loss": 0.269,
+      "step": 8465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610064357621566e-05,
+      "loss": 0.262,
+      "step": 8466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609969654687464e-05,
+      "loss": 0.2645,
+      "step": 8467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609874943040538e-05,
+      "loss": 0.265,
+      "step": 8468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60978022268214e-05,
+      "loss": 0.273,
+      "step": 8469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609685493613623e-05,
+      "loss": 0.2748,
+      "step": 8470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6095907558363406e-05,
+      "loss": 0.2432,
+      "step": 8471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609496009351646e-05,
+      "loss": 0.2681,
+      "step": 8472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094012541608924e-05,
+      "loss": 0.2757,
+      "step": 8473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6093064902654333e-05,
+      "loss": 0.28,
+      "step": 8474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092117176666225e-05,
+      "loss": 0.2669,
+      "step": 8475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6091169363658134e-05,
+      "loss": 0.2696,
+      "step": 8476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60902214636436e-05,
+      "loss": 0.2644,
+      "step": 8477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089273476636164e-05,
+      "loss": 0.2436,
+      "step": 8478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6088325402649364e-05,
+      "loss": 0.2698,
+      "step": 8479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608737724169675e-05,
+      "loss": 0.2682,
+      "step": 8480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6086428993791857e-05,
+      "loss": 0.2348,
+      "step": 8481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608548065894823e-05,
+      "loss": 0.2731,
+      "step": 8482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608453223717942e-05,
+      "loss": 0.2645,
+      "step": 8483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6083583728498973e-05,
+      "loss": 0.2735,
+      "step": 8484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082635132920435e-05,
+      "loss": 0.2706,
+      "step": 8485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608168645045736e-05,
+      "loss": 0.2808,
+      "step": 8486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080737681123294e-05,
+      "loss": 0.2751,
+      "step": 8487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6079788824931793e-05,
+      "loss": 0.2614,
+      "step": 8488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607883988189641e-05,
+      "loss": 0.288,
+      "step": 8489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6077890852030696e-05,
+      "loss": 0.2549,
+      "step": 8490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607694173534821e-05,
+      "loss": 0.25,
+      "step": 8491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075992531862512e-05,
+      "loss": 0.266,
+      "step": 8492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075043241587155e-05,
+      "loss": 0.2558,
+      "step": 8493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074093864535704e-05,
+      "loss": 0.2565,
+      "step": 8494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607314440072172e-05,
+      "loss": 0.275,
+      "step": 8495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6072194850158755e-05,
+      "loss": 0.2489,
+      "step": 8496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6071245212860382e-05,
+      "loss": 0.2661,
+      "step": 8497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070295488840167e-05,
+      "loss": 0.2785,
+      "step": 8498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606934567811167e-05,
+      "loss": 0.2451,
+      "step": 8499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068395780688458e-05,
+      "loss": 0.2678,
+      "step": 8500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067445796584107e-05,
+      "loss": 0.277,
+      "step": 8501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066495725812177e-05,
+      "loss": 0.2698,
+      "step": 8502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065545568386244e-05,
+      "loss": 0.2817,
+      "step": 8503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606459532431988e-05,
+      "loss": 0.2591,
+      "step": 8504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6063644993626656e-05,
+      "loss": 0.259,
+      "step": 8505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606269457632015e-05,
+      "loss": 0.2275,
+      "step": 8506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061744072413937e-05,
+      "loss": 0.2831,
+      "step": 8507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6060793481921592e-05,
+      "loss": 0.2601,
+      "step": 8508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059842804856693e-05,
+      "loss": 0.265,
+      "step": 8509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6058892041232823e-05,
+      "loss": 0.3014,
+      "step": 8510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057941191063558e-05,
+      "loss": 0.2747,
+      "step": 8511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056990254362485e-05,
+      "loss": 0.275,
+      "step": 8512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056039231143182e-05,
+      "loss": 0.2622,
+      "step": 8513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055088121419238e-05,
+      "loss": 0.2424,
+      "step": 8514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6054136925204237e-05,
+      "loss": 0.2597,
+      "step": 8515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6053185642511763e-05,
+      "loss": 0.277,
+      "step": 8516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605223427335541e-05,
+      "loss": 0.3102,
+      "step": 8517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051282817748762e-05,
+      "loss": 0.2634,
+      "step": 8518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050331275705412e-05,
+      "loss": 0.295,
+      "step": 8519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604937964723895e-05,
+      "loss": 0.2711,
+      "step": 8520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048427932362976e-05,
+      "loss": 0.28,
+      "step": 8521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047476131091077e-05,
+      "loss": 0.2871,
+      "step": 8522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604652424343685e-05,
+      "loss": 0.2575,
+      "step": 8523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604557226941389e-05,
+      "loss": 0.2527,
+      "step": 8524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044620209035802e-05,
+      "loss": 0.284,
+      "step": 8525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604366806231618e-05,
+      "loss": 0.2591,
+      "step": 8526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6042715829268622e-05,
+      "loss": 0.2809,
+      "step": 8527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6041763509906738e-05,
+      "loss": 0.237,
+      "step": 8528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040811104244122e-05,
+      "loss": 0.2534,
+      "step": 8529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039858612294382e-05,
+      "loss": 0.3062,
+      "step": 8530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038906034071124e-05,
+      "loss": 0.2596,
+      "step": 8531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6037953369587955e-05,
+      "loss": 0.2495,
+      "step": 8532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6037000618858483e-05,
+      "loss": 0.2799,
+      "step": 8533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036047781896316e-05,
+      "loss": 0.2531,
+      "step": 8534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035094858715065e-05,
+      "loss": 0.2835,
+      "step": 8535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603414184932834e-05,
+      "loss": 0.2796,
+      "step": 8536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6033188753749752e-05,
+      "loss": 0.2963,
+      "step": 8537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603223557199292e-05,
+      "loss": 0.259,
+      "step": 8538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603128230407146e-05,
+      "loss": 0.2992,
+      "step": 8539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6030328949998982e-05,
+      "loss": 0.2657,
+      "step": 8540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602937550978911e-05,
+      "loss": 0.2692,
+      "step": 8541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6028421983455462e-05,
+      "loss": 0.2597,
+      "step": 8542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027468371011653e-05,
+      "loss": 0.2821,
+      "step": 8543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602651467247131e-05,
+      "loss": 0.2755,
+      "step": 8544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602556088784805e-05,
+      "loss": 0.2731,
+      "step": 8545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024607017155505e-05,
+      "loss": 0.2909,
+      "step": 8546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.60236530604073e-05,
+      "loss": 0.2652,
+      "step": 8547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602269901761705e-05,
+      "loss": 0.276,
+      "step": 8548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6021744888798393e-05,
+      "loss": 0.2774,
+      "step": 8549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602079067396495e-05,
+      "loss": 0.2531,
+      "step": 8550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601983637313036e-05,
+      "loss": 0.2508,
+      "step": 8551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601888198630825e-05,
+      "loss": 0.2604,
+      "step": 8552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601792751351225e-05,
+      "loss": 0.2582,
+      "step": 8553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016972954755998e-05,
+      "loss": 0.2432,
+      "step": 8554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016018310053124e-05,
+      "loss": 0.2901,
+      "step": 8555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601506357941727e-05,
+      "loss": 0.2634,
+      "step": 8556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601410876286207e-05,
+      "loss": 0.2695,
+      "step": 8557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601315386040116e-05,
+      "loss": 0.2541,
+      "step": 8558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012198872048185e-05,
+      "loss": 0.2402,
+      "step": 8559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6011243797816787e-05,
+      "loss": 0.2304,
+      "step": 8560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010288637720605e-05,
+      "loss": 0.255,
+      "step": 8561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600933339177328e-05,
+      "loss": 0.2587,
+      "step": 8562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600837805998846e-05,
+      "loss": 0.2712,
+      "step": 8563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600742264237979e-05,
+      "loss": 0.2937,
+      "step": 8564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006467138960918e-05,
+      "loss": 0.2846,
+      "step": 8565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6005511549745495e-05,
+      "loss": 0.2717,
+      "step": 8566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004555874747167e-05,
+      "loss": 0.2775,
+      "step": 8567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6003600113979586e-05,
+      "loss": 0.2559,
+      "step": 8568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002644267456405e-05,
+      "loss": 0.2253,
+      "step": 8569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6001688335191273e-05,
+      "loss": 0.2688,
+      "step": 8570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600073231719785e-05,
+      "loss": 0.2561,
+      "step": 8571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599977621348979e-05,
+      "loss": 0.302,
+      "step": 8572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599882002408075e-05,
+      "loss": 0.2451,
+      "step": 8573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997863748984384e-05,
+      "loss": 0.254,
+      "step": 8574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599690738821436e-05,
+      "loss": 0.31,
+      "step": 8575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5995950941784334e-05,
+      "loss": 0.2435,
+      "step": 8576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599499440970797e-05,
+      "loss": 0.2867,
+      "step": 8577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599403779199892e-05,
+      "loss": 0.2613,
+      "step": 8578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5993081088670864e-05,
+      "loss": 0.2917,
+      "step": 8579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599212429973746e-05,
+      "loss": 0.2657,
+      "step": 8580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599116742521238e-05,
+      "loss": 0.2777,
+      "step": 8581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599021046510928e-05,
+      "loss": 0.2579,
+      "step": 8582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989253419441843e-05,
+      "loss": 0.2558,
+      "step": 8583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988296288223733e-05,
+      "loss": 0.256,
+      "step": 8584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598733907146862e-05,
+      "loss": 0.259,
+      "step": 8585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598638176919018e-05,
+      "loss": 0.2717,
+      "step": 8586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985424381402087e-05,
+      "loss": 0.2579,
+      "step": 8587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598446690811802e-05,
+      "loss": 0.267,
+      "step": 8588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5983509349351643e-05,
+      "loss": 0.2677,
+      "step": 8589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982551705116644e-05,
+      "loss": 0.2794,
+      "step": 8590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981593975426703e-05,
+      "loss": 0.2627,
+      "step": 8591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5980636160295496e-05,
+      "loss": 0.3025,
+      "step": 8592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5979678259736706e-05,
+      "loss": 0.2639,
+      "step": 8593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5978720273764018e-05,
+      "loss": 0.2637,
+      "step": 8594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5977762202391107e-05,
+      "loss": 0.2715,
+      "step": 8595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597680404563167e-05,
+      "loss": 0.2853,
+      "step": 8596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975845803499384e-05,
+      "loss": 0.2775,
+      "step": 8597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974887476007945e-05,
+      "loss": 0.2751,
+      "step": 8598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5973929063171035e-05,
+      "loss": 0.3095,
+      "step": 8599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5972970565002345e-05,
+      "loss": 0.2838,
+      "step": 8600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597201198151557e-05,
+      "loss": 0.2541,
+      "step": 8601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59710533127244e-05,
+      "loss": 0.2711,
+      "step": 8602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970094558642523e-05,
+      "loss": 0.2426,
+      "step": 8603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5969135719283643e-05,
+      "loss": 0.2778,
+      "step": 8604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5968176794661454e-05,
+      "loss": 0.2558,
+      "step": 8605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596721778478965e-05,
+      "loss": 0.2485,
+      "step": 8606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5966258689681932e-05,
+      "loss": 0.255,
+      "step": 8607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965299509351996e-05,
+      "loss": 0.2582,
+      "step": 8608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596434024381355e-05,
+      "loss": 0.2424,
+      "step": 8609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596338089308029e-05,
+      "loss": 0.2627,
+      "step": 8610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596242145716592e-05,
+      "loss": 0.2493,
+      "step": 8611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596146193608415e-05,
+      "loss": 0.2635,
+      "step": 8612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5960502329848683e-05,
+      "loss": 0.2831,
+      "step": 8613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595954263847322e-05,
+      "loss": 0.2848,
+      "step": 8614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5958582861971475e-05,
+      "loss": 0.2651,
+      "step": 8615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595762300035716e-05,
+      "loss": 0.2734,
+      "step": 8616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595666305364398e-05,
+      "loss": 0.2543,
+      "step": 8617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595570302184565e-05,
+      "loss": 0.2885,
+      "step": 8618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954742904975882e-05,
+      "loss": 0.2427,
+      "step": 8619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5953782703048392e-05,
+      "loss": 0.2957,
+      "step": 8620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5952822416076893e-05,
+      "loss": 0.2684,
+      "step": 8621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5951862044075107e-05,
+      "loss": 0.2549,
+      "step": 8622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5950901587056743e-05,
+      "loss": 0.2798,
+      "step": 8623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949941045035525e-05,
+      "loss": 0.2722,
+      "step": 8624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5948980418025177e-05,
+      "loss": 0.258,
+      "step": 8625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5948019706039416e-05,
+      "loss": 0.2548,
+      "step": 8626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5947058909091972e-05,
+      "loss": 0.2845,
+      "step": 8627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5946098027196555e-05,
+      "loss": 0.2764,
+      "step": 8628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5945137060366902e-05,
+      "loss": 0.2837,
+      "step": 8629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5944176008616734e-05,
+      "loss": 0.2796,
+      "step": 8630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594321487195978e-05,
+      "loss": 0.2812,
+      "step": 8631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942253650409773e-05,
+      "loss": 0.263,
+      "step": 8632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594129234398044e-05,
+      "loss": 0.2712,
+      "step": 8633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594033095268551e-05,
+      "loss": 0.2876,
+      "step": 8634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939369476538717e-05,
+      "loss": 0.2837,
+      "step": 8635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938407915553795e-05,
+      "loss": 0.2589,
+      "step": 8636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593744626974448e-05,
+      "loss": 0.2451,
+      "step": 8637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593648453912451e-05,
+      "loss": 0.3014,
+      "step": 8638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5935522723707617e-05,
+      "loss": 0.273,
+      "step": 8639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5934560823507544e-05,
+      "loss": 0.2862,
+      "step": 8640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933598838538027e-05,
+      "loss": 0.2668,
+      "step": 8641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593263676881281e-05,
+      "loss": 0.2623,
+      "step": 8642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593167461434564e-05,
+      "loss": 0.2737,
+      "step": 8643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5930712375150248e-05,
+      "loss": 0.2816,
+      "step": 8644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592975005124039e-05,
+      "loss": 0.2744,
+      "step": 8645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5928787642629803e-05,
+      "loss": 0.2503,
+      "step": 8646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592782514933224e-05,
+      "loss": 0.2763,
+      "step": 8647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592686257136145e-05,
+      "loss": 0.2832,
+      "step": 8648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925899908731176e-05,
+      "loss": 0.2659,
+      "step": 8649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5924937161455178e-05,
+      "loss": 0.234,
+      "step": 8650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923974329547202e-05,
+      "loss": 0.2704,
+      "step": 8651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923011413021e-05,
+      "loss": 0.2917,
+      "step": 8652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592204841189033e-05,
+      "loss": 0.2665,
+      "step": 8653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5921085326168944e-05,
+      "loss": 0.2826,
+      "step": 8654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920122155870604e-05,
+      "loss": 0.2811,
+      "step": 8655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591915890100906e-05,
+      "loss": 0.2666,
+      "step": 8656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591819556159808e-05,
+      "loss": 0.25,
+      "step": 8657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917232137651417e-05,
+      "loss": 0.2728,
+      "step": 8658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5916268629182837e-05,
+      "loss": 0.2731,
+      "step": 8659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5915305036206105e-05,
+      "loss": 0.2938,
+      "step": 8660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914341358734977e-05,
+      "loss": 0.2588,
+      "step": 8661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5913377596783222e-05,
+      "loss": 0.2947,
+      "step": 8662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591241375036461e-05,
+      "loss": 0.2682,
+      "step": 8663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5911449819492903e-05,
+      "loss": 0.2886,
+      "step": 8664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910485804181878e-05,
+      "loss": 0.2957,
+      "step": 8665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5909521704445293e-05,
+      "loss": 0.2794,
+      "step": 8666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590855752029693e-05,
+      "loss": 0.2703,
+      "step": 8667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907593251750556e-05,
+      "loss": 0.2747,
+      "step": 8668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906628898819943e-05,
+      "loss": 0.2679,
+      "step": 8669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905664461518875e-05,
+      "loss": 0.2847,
+      "step": 8670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590469993986112e-05,
+      "loss": 0.2785,
+      "step": 8671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5903735333860453e-05,
+      "loss": 0.2404,
+      "step": 8672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5902770643530657e-05,
+      "loss": 0.2739,
+      "step": 8673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901805868885514e-05,
+      "loss": 0.2962,
+      "step": 8674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59008410099388e-05,
+      "loss": 0.2622,
+      "step": 8675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5899876066704303e-05,
+      "loss": 0.2727,
+      "step": 8676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898911039195797e-05,
+      "loss": 0.2618,
+      "step": 8677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5897945927427076e-05,
+      "loss": 0.2823,
+      "step": 8678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589698073141192e-05,
+      "loss": 0.2497,
+      "step": 8679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896015451164116e-05,
+      "loss": 0.2796,
+      "step": 8680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5895050086697454e-05,
+      "loss": 0.2252,
+      "step": 8681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589408463802572e-05,
+      "loss": 0.2956,
+      "step": 8682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5893119105162712e-05,
+      "loss": 0.2691,
+      "step": 8683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892153488122216e-05,
+      "loss": 0.2516,
+      "step": 8684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589118778691803e-05,
+      "loss": 0.2522,
+      "step": 8685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5890222001563937e-05,
+      "loss": 0.2441,
+      "step": 8686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588925613207374e-05,
+      "loss": 0.283,
+      "step": 8687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5888290178461237e-05,
+      "loss": 0.2657,
+      "step": 8688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588732414074022e-05,
+      "loss": 0.3045,
+      "step": 8689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886358018924498e-05,
+      "loss": 0.2514,
+      "step": 8690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588539181302786e-05,
+      "loss": 0.2665,
+      "step": 8691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5884425523064114e-05,
+      "loss": 0.2909,
+      "step": 8692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5883459149047057e-05,
+      "loss": 0.2927,
+      "step": 8693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5882492690990502e-05,
+      "loss": 0.2573,
+      "step": 8694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588152614890824e-05,
+      "loss": 0.2839,
+      "step": 8695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588055952281409e-05,
+      "loss": 0.2912,
+      "step": 8696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5879592812721856e-05,
+      "loss": 0.2618,
+      "step": 8697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878626018645342e-05,
+      "loss": 0.2659,
+      "step": 8698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877659140598362e-05,
+      "loss": 0.2656,
+      "step": 8699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5876692178594727e-05,
+      "loss": 0.2711,
+      "step": 8700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875725132648242e-05,
+      "loss": 0.2854,
+      "step": 8701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5874758002772733e-05,
+      "loss": 0.2746,
+      "step": 8702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873790788982005e-05,
+      "loss": 0.2746,
+      "step": 8703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587282349128987e-05,
+      "loss": 0.2797,
+      "step": 8704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5871856109710163e-05,
+      "loss": 0.2565,
+      "step": 8705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587088864425668e-05,
+      "loss": 0.2485,
+      "step": 8706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586992109494326e-05,
+      "loss": 0.2724,
+      "step": 8707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868953461783706e-05,
+      "loss": 0.2964,
+      "step": 8708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586798574479185e-05,
+      "loss": 0.2568,
+      "step": 8709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586701794398151e-05,
+      "loss": 0.2747,
+      "step": 8710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5866050059366517e-05,
+      "loss": 0.2556,
+      "step": 8711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5865082090960692e-05,
+      "loss": 0.2807,
+      "step": 8712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586411403877786e-05,
+      "loss": 0.2302,
+      "step": 8713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586314590283185e-05,
+      "loss": 0.249,
+      "step": 8714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5862177683136493e-05,
+      "loss": 0.28,
+      "step": 8715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5861209379705617e-05,
+      "loss": 0.285,
+      "step": 8716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5860240992553052e-05,
+      "loss": 0.2963,
+      "step": 8717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5859272521692633e-05,
+      "loss": 0.2537,
+      "step": 8718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585830396713819e-05,
+      "loss": 0.2497,
+      "step": 8719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585733532890356e-05,
+      "loss": 0.2508,
+      "step": 8720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5856366607002583e-05,
+      "loss": 0.2671,
+      "step": 8721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585539780144909e-05,
+      "loss": 0.2754,
+      "step": 8722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5854428912256925e-05,
+      "loss": 0.2602,
+      "step": 8723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853459939439923e-05,
+      "loss": 0.2707,
+      "step": 8724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585249088301193e-05,
+      "loss": 0.2705,
+      "step": 8725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585152174298678e-05,
+      "loss": 0.2536,
+      "step": 8726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850552519378322e-05,
+      "loss": 0.2582,
+      "step": 8727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58495832122004e-05,
+      "loss": 0.2394,
+      "step": 8728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584861382146686e-05,
+      "loss": 0.254,
+      "step": 8729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847644347191545e-05,
+      "loss": 0.2575,
+      "step": 8730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5846674789388307e-05,
+      "loss": 0.2253,
+      "step": 8731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5845705148070995e-05,
+      "loss": 0.2769,
+      "step": 8732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584473542325346e-05,
+      "loss": 0.267,
+      "step": 8733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584376561494955e-05,
+      "loss": 0.2512,
+      "step": 8734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5842795723173118e-05,
+      "loss": 0.2727,
+      "step": 8735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841825747938024e-05,
+      "loss": 0.2414,
+      "step": 8736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5840855689258115e-05,
+      "loss": 0.2732,
+      "step": 8737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583988554714725e-05,
+      "loss": 0.2811,
+      "step": 8738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838915321619294e-05,
+      "loss": 0.2792,
+      "step": 8739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5837945012688096e-05,
+      "loss": 0.2695,
+      "step": 8740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836974620367517e-05,
+      "loss": 0.2641,
+      "step": 8741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583600414467142e-05,
+      "loss": 0.2781,
+      "step": 8742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583503358561367e-05,
+      "loss": 0.2947,
+      "step": 8743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583406294320813e-05,
+      "loss": 0.2832,
+      "step": 8744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583309221746866e-05,
+      "loss": 0.2716,
+      "step": 8745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583212140840913e-05,
+      "loss": 0.275,
+      "step": 8746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831150516043407e-05,
+      "loss": 0.2799,
+      "step": 8747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583017954038536e-05,
+      "loss": 0.2516,
+      "step": 8748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829208481448848e-05,
+      "loss": 0.2582,
+      "step": 8749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5828237339247758e-05,
+      "loss": 0.2824,
+      "step": 8750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827266113795953e-05,
+      "loss": 0.2905,
+      "step": 8751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826294805107305e-05,
+      "loss": 0.2671,
+      "step": 8752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582532341319569e-05,
+      "loss": 0.2912,
+      "step": 8753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5824351938074986e-05,
+      "loss": 0.2748,
+      "step": 8754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5823380379759067e-05,
+      "loss": 0.269,
+      "step": 8755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582240873826181e-05,
+      "loss": 0.2558,
+      "step": 8756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821437013597093e-05,
+      "loss": 0.2819,
+      "step": 8757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58204652057788e-05,
+      "loss": 0.2651,
+      "step": 8758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5819493314820812e-05,
+      "loss": 0.2897,
+      "step": 8759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818521340737006e-05,
+      "loss": 0.2849,
+      "step": 8760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5817549283541273e-05,
+      "loss": 0.2532,
+      "step": 8761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816577143247496e-05,
+      "loss": 0.2956,
+      "step": 8762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5815604919869553e-05,
+      "loss": 0.2578,
+      "step": 8763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814632613421345e-05,
+      "loss": 0.2764,
+      "step": 8764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5813660223916752e-05,
+      "loss": 0.2838,
+      "step": 8765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5812687751369662e-05,
+      "loss": 0.2934,
+      "step": 8766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581171519579397e-05,
+      "loss": 0.2598,
+      "step": 8767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810742557203568e-05,
+      "loss": 0.246,
+      "step": 8768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580976983561235e-05,
+      "loss": 0.2726,
+      "step": 8769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5808797031034205e-05,
+      "loss": 0.2412,
+      "step": 8770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580782414348303e-05,
+      "loss": 0.2556,
+      "step": 8771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580685117297273e-05,
+      "loss": 0.2735,
+      "step": 8772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580587811951719e-05,
+      "loss": 0.2546,
+      "step": 8773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580490498313032e-05,
+      "loss": 0.2719,
+      "step": 8774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5803931763826015e-05,
+      "loss": 0.282,
+      "step": 8775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802958461618177e-05,
+      "loss": 0.2466,
+      "step": 8776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580198507652071e-05,
+      "loss": 0.264,
+      "step": 8777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801011608547518e-05,
+      "loss": 0.2835,
+      "step": 8778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5800038057712504e-05,
+      "loss": 0.27,
+      "step": 8779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799064424029576e-05,
+      "loss": 0.298,
+      "step": 8780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579809070751264e-05,
+      "loss": 0.2477,
+      "step": 8781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797116908175606e-05,
+      "loss": 0.2957,
+      "step": 8782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579614302603238e-05,
+      "loss": 0.2616,
+      "step": 8783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5795169061096882e-05,
+      "loss": 0.2406,
+      "step": 8784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794195013383017e-05,
+      "loss": 0.2843,
+      "step": 8785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5793220882904696e-05,
+      "loss": 0.2734,
+      "step": 8786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579224666967584e-05,
+      "loss": 0.2512,
+      "step": 8787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579127237371036e-05,
+      "loss": 0.276,
+      "step": 8788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790297995022175e-05,
+      "loss": 0.2788,
+      "step": 8789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5789323533625205e-05,
+      "loss": 0.2866,
+      "step": 8790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5788348989533366e-05,
+      "loss": 0.2884,
+      "step": 8791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578737436276058e-05,
+      "loss": 0.2613,
+      "step": 8792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5786399653320767e-05,
+      "loss": 0.2866,
+      "step": 8793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578542486122785e-05,
+      "loss": 0.2735,
+      "step": 8794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5784449986495755e-05,
+      "loss": 0.2826,
+      "step": 8795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783475029138405e-05,
+      "loss": 0.2874,
+      "step": 8796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578249998916973e-05,
+      "loss": 0.2747,
+      "step": 8797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5781524866603654e-05,
+      "loss": 0.264,
+      "step": 8798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57805496614541e-05,
+      "loss": 0.2428,
+      "step": 8799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577957437373501e-05,
+      "loss": 0.2601,
+      "step": 8800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778599003460308e-05,
+      "loss": 0.2622,
+      "step": 8801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5777623550643924e-05,
+      "loss": 0.2654,
+      "step": 8802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5776648015299802e-05,
+      "loss": 0.2538,
+      "step": 8803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5775672397441868e-05,
+      "loss": 0.2812,
+      "step": 8804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5774696697084055e-05,
+      "loss": 0.2689,
+      "step": 8805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5773720914240308e-05,
+      "loss": 0.2181,
+      "step": 8806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577274504892456e-05,
+      "loss": 0.268,
+      "step": 8807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5771769101150752e-05,
+      "loss": 0.2547,
+      "step": 8808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577079307093282e-05,
+      "loss": 0.2625,
+      "step": 8809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769816958284714e-05,
+      "loss": 0.2569,
+      "step": 8810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5768840763220375e-05,
+      "loss": 0.2484,
+      "step": 8811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767864485753737e-05,
+      "loss": 0.2563,
+      "step": 8812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576688812589876e-05,
+      "loss": 0.2622,
+      "step": 8813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576591168366938e-05,
+      "loss": 0.2707,
+      "step": 8814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5764935159079545e-05,
+      "loss": 0.2539,
+      "step": 8815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576395855214321e-05,
+      "loss": 0.2574,
+      "step": 8816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576298186287432e-05,
+      "loss": 0.2702,
+      "step": 8817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5762005091286824e-05,
+      "loss": 0.2731,
+      "step": 8818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761028237394682e-05,
+      "loss": 0.2554,
+      "step": 8819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576005130121184e-05,
+      "loss": 0.2687,
+      "step": 8820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5759074282752254e-05,
+      "loss": 0.2725,
+      "step": 8821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5758097182029887e-05,
+      "loss": 0.2946,
+      "step": 8822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5757119999058686e-05,
+      "loss": 0.3025,
+      "step": 8823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756142733852616e-05,
+      "loss": 0.2747,
+      "step": 8824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575516538642563e-05,
+      "loss": 0.2761,
+      "step": 8825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754187956791694e-05,
+      "loss": 0.2598,
+      "step": 8826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753210444964765e-05,
+      "loss": 0.2707,
+      "step": 8827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5752232850958812e-05,
+      "loss": 0.2767,
+      "step": 8828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5751255174787793e-05,
+      "loss": 0.2757,
+      "step": 8829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750277416465678e-05,
+      "loss": 0.2671,
+      "step": 8830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574929957600643e-05,
+      "loss": 0.2592,
+      "step": 8831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5748321653424018e-05,
+      "loss": 0.2492,
+      "step": 8832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747343648732406e-05,
+      "loss": 0.243,
+      "step": 8833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574636556194557e-05,
+      "loss": 0.2501,
+      "step": 8834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5745387393077482e-05,
+      "loss": 0.2724,
+      "step": 8835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574440914214211e-05,
+      "loss": 0.2787,
+      "step": 8836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574343080915343e-05,
+      "loss": 0.2743,
+      "step": 8837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574245239412541e-05,
+      "loss": 0.2569,
+      "step": 8838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741473897072033e-05,
+      "loss": 0.2651,
+      "step": 8839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574049531800727e-05,
+      "loss": 0.2564,
+      "step": 8840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5739516656945108e-05,
+      "loss": 0.2288,
+      "step": 8841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573853791389952e-05,
+      "loss": 0.2429,
+      "step": 8842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737559088884484e-05,
+      "loss": 0.2684,
+      "step": 8843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573658018191399e-05,
+      "loss": 0.2785,
+      "step": 8844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5735601193002015e-05,
+      "loss": 0.2649,
+      "step": 8845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573462212216254e-05,
+      "loss": 0.2713,
+      "step": 8846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733642969409553e-05,
+      "loss": 0.2923,
+      "step": 8847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732663734757044e-05,
+      "loss": 0.2621,
+      "step": 8848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731684418218994e-05,
+      "loss": 0.3041,
+      "step": 8849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57307050198094e-05,
+      "loss": 0.289,
+      "step": 8850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5729725539542245e-05,
+      "loss": 0.2753,
+      "step": 8851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5728745977431515e-05,
+      "loss": 0.2667,
+      "step": 8852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727766333491216e-05,
+      "loss": 0.2593,
+      "step": 8853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726786607735333e-05,
+      "loss": 0.3001,
+      "step": 8854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5725806800177857e-05,
+      "loss": 0.293,
+      "step": 8855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572482691083279e-05,
+      "loss": 0.2744,
+      "step": 8856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572384693971413e-05,
+      "loss": 0.2642,
+      "step": 8857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722866886835868e-05,
+      "loss": 0.2862,
+      "step": 8858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5721886752212004e-05,
+      "loss": 0.2471,
+      "step": 8859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572090653585655e-05,
+      "loss": 0.26,
+      "step": 8860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571992623778349e-05,
+      "loss": 0.2569,
+      "step": 8861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5718945858006838e-05,
+      "loss": 0.2601,
+      "step": 8862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5717965396540594e-05,
+      "loss": 0.2828,
+      "step": 8863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716984853398764e-05,
+      "loss": 0.2462,
+      "step": 8864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716004228595355e-05,
+      "loss": 0.2918,
+      "step": 8865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5715023522144374e-05,
+      "loss": 0.2822,
+      "step": 8866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5714042734059827e-05,
+      "loss": 0.2951,
+      "step": 8867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713061864355723e-05,
+      "loss": 0.2543,
+      "step": 8868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712080913046077e-05,
+      "loss": 0.2936,
+      "step": 8869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57110998801449e-05,
+      "loss": 0.2581,
+      "step": 8870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5710118765666203e-05,
+      "loss": 0.257,
+      "step": 8871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5709137569624e-05,
+      "loss": 0.283,
+      "step": 8872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5708156292032308e-05,
+      "loss": 0.2702,
+      "step": 8873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5707174932905144e-05,
+      "loss": 0.2765,
+      "step": 8874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706193492256524e-05,
+      "loss": 0.2575,
+      "step": 8875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570521197010047e-05,
+      "loss": 0.2705,
+      "step": 8876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704230366451e-05,
+      "loss": 0.2552,
+      "step": 8877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5703248681322136e-05,
+      "loss": 0.2884,
+      "step": 8878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702266914727896e-05,
+      "loss": 0.2717,
+      "step": 8879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701285066682307e-05,
+      "loss": 0.2432,
+      "step": 8880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57003031371994e-05,
+      "loss": 0.233,
+      "step": 8881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569932112629319e-05,
+      "loss": 0.2378,
+      "step": 8882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569833903397771e-05,
+      "loss": 0.2565,
+      "step": 8883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569735686026699e-05,
+      "loss": 0.2432,
+      "step": 8884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696374605175052e-05,
+      "loss": 0.3,
+      "step": 8885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695392268715934e-05,
+      "loss": 0.2856,
+      "step": 8886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694409850903666e-05,
+      "loss": 0.2449,
+      "step": 8887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5693427351752276e-05,
+      "loss": 0.2773,
+      "step": 8888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5692444771275804e-05,
+      "loss": 0.2533,
+      "step": 8889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691462109488285e-05,
+      "loss": 0.2493,
+      "step": 8890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569047936640375e-05,
+      "loss": 0.2718,
+      "step": 8891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5689496542036243e-05,
+      "loss": 0.2623,
+      "step": 8892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688513636399797e-05,
+      "loss": 0.2902,
+      "step": 8893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568753064950846e-05,
+      "loss": 0.2783,
+      "step": 8894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568654758137626e-05,
+      "loss": 0.3118,
+      "step": 8895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568556443201725e-05,
+      "loss": 0.2971,
+      "step": 8896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568458120144547e-05,
+      "loss": 0.276,
+      "step": 8897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5683597889674966e-05,
+      "loss": 0.2649,
+      "step": 8898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568261449671978e-05,
+      "loss": 0.2574,
+      "step": 8899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5681631022593964e-05,
+      "loss": 0.261,
+      "step": 8900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.2308,
+      "step": 8901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567966383088662e-05,
+      "loss": 0.2866,
+      "step": 8902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5678680113333196e-05,
+      "loss": 0.2469,
+      "step": 8903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677696314665337e-05,
+      "loss": 0.2692,
+      "step": 8904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676712434897098e-05,
+      "loss": 0.2677,
+      "step": 8905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5675728474042527e-05,
+      "loss": 0.2944,
+      "step": 8906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567474443211569e-05,
+      "loss": 0.2681,
+      "step": 8907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673760309130633e-05,
+      "loss": 0.2713,
+      "step": 8908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5672776105101415e-05,
+      "loss": 0.2795,
+      "step": 8909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671791820042097e-05,
+      "loss": 0.2628,
+      "step": 8910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5670807453966737e-05,
+      "loss": 0.2424,
+      "step": 8911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669823006889397e-05,
+      "loss": 0.2801,
+      "step": 8912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566883847882414e-05,
+      "loss": 0.2487,
+      "step": 8913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5667853869785022e-05,
+      "loss": 0.2507,
+      "step": 8914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666869179786113e-05,
+      "loss": 0.2579,
+      "step": 8915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5665884408841484e-05,
+      "loss": 0.2444,
+      "step": 8916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566489955696519e-05,
+      "loss": 0.2504,
+      "step": 8917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663914624171304e-05,
+      "loss": 0.2754,
+      "step": 8918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662929610473895e-05,
+      "loss": 0.2727,
+      "step": 8919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5661944515887033e-05,
+      "loss": 0.2661,
+      "step": 8920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5660959340424787e-05,
+      "loss": 0.2491,
+      "step": 8921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659974084101232e-05,
+      "loss": 0.2645,
+      "step": 8922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5658988746930442e-05,
+      "loss": 0.2959,
+      "step": 8923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5658003328926488e-05,
+      "loss": 0.2599,
+      "step": 8924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5657017830103448e-05,
+      "loss": 0.2523,
+      "step": 8925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5656032250475396e-05,
+      "loss": 0.2842,
+      "step": 8926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5655046590056416e-05,
+      "loss": 0.2639,
+      "step": 8927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654060848860584e-05,
+      "loss": 0.2606,
+      "step": 8928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565307502690198e-05,
+      "loss": 0.253,
+      "step": 8929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652089124194683e-05,
+      "loss": 0.2738,
+      "step": 8930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565110314075278e-05,
+      "loss": 0.253,
+      "step": 8931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5650117076590353e-05,
+      "loss": 0.2659,
+      "step": 8932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564913093172149e-05,
+      "loss": 0.2687,
+      "step": 8933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564814470616027e-05,
+      "loss": 0.2657,
+      "step": 8934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564715839992079e-05,
+      "loss": 0.2637,
+      "step": 8935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646172013017127e-05,
+      "loss": 0.334,
+      "step": 8936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564518554546338e-05,
+      "loss": 0.2891,
+      "step": 8937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5644198997273634e-05,
+      "loss": 0.2729,
+      "step": 8938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5643212368461984e-05,
+      "loss": 0.2596,
+      "step": 8939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5642225659042522e-05,
+      "loss": 0.3021,
+      "step": 8940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641238869029342e-05,
+      "loss": 0.2597,
+      "step": 8941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640251998436542e-05,
+      "loss": 0.2334,
+      "step": 8942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5639265047278213e-05,
+      "loss": 0.2856,
+      "step": 8943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5638278015568458e-05,
+      "loss": 0.2881,
+      "step": 8944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5637290903321373e-05,
+      "loss": 0.2981,
+      "step": 8945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636303710551057e-05,
+      "loss": 0.2406,
+      "step": 8946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635316437271614e-05,
+      "loss": 0.2609,
+      "step": 8947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634329083497147e-05,
+      "loss": 0.2391,
+      "step": 8948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633341649241756e-05,
+      "loss": 0.2824,
+      "step": 8949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5632354134519546e-05,
+      "loss": 0.251,
+      "step": 8950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631366539344622e-05,
+      "loss": 0.3062,
+      "step": 8951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630378863731096e-05,
+      "loss": 0.2637,
+      "step": 8952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562939110769307e-05,
+      "loss": 0.2849,
+      "step": 8953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5628403271244655e-05,
+      "loss": 0.2787,
+      "step": 8954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5627415354399964e-05,
+      "loss": 0.2747,
+      "step": 8955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626427357173106e-05,
+      "loss": 0.2901,
+      "step": 8956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562543927957819e-05,
+      "loss": 0.2542,
+      "step": 8957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562445112162934e-05,
+      "loss": 0.2531,
+      "step": 8958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562346288334066e-05,
+      "loss": 0.2686,
+      "step": 8959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562247456472627e-05,
+      "loss": 0.2631,
+      "step": 8960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5621486165800287e-05,
+      "loss": 0.2693,
+      "step": 8961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5620497686576835e-05,
+      "loss": 0.2803,
+      "step": 8962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619509127070022e-05,
+      "loss": 0.2587,
+      "step": 8963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561852048729398e-05,
+      "loss": 0.2804,
+      "step": 8964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5617531767262822e-05,
+      "loss": 0.2679,
+      "step": 8965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616542966990675e-05,
+      "loss": 0.2812,
+      "step": 8966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5615554086491663e-05,
+      "loss": 0.273,
+      "step": 8967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5614565125779916e-05,
+      "loss": 0.3337,
+      "step": 8968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5613576084869547e-05,
+      "loss": 0.2701,
+      "step": 8969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56125869637747e-05,
+      "loss": 0.2739,
+      "step": 8970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561159776250949e-05,
+      "loss": 0.2835,
+      "step": 8971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5610608481088054e-05,
+      "loss": 0.2611,
+      "step": 8972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5609619119524517e-05,
+      "loss": 0.2872,
+      "step": 8973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560862967783302e-05,
+      "loss": 0.2652,
+      "step": 8974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5607640156027693e-05,
+      "loss": 0.2707,
+      "step": 8975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5606650554122663e-05,
+      "loss": 0.2728,
+      "step": 8976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605660872132074e-05,
+      "loss": 0.2695,
+      "step": 8977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560467111007006e-05,
+      "loss": 0.2499,
+      "step": 8978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5603681267950757e-05,
+      "loss": 0.2479,
+      "step": 8979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602691345788306e-05,
+      "loss": 0.2952,
+      "step": 8980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5601701343596848e-05,
+      "loss": 0.2504,
+      "step": 8981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5600711261390523e-05,
+      "loss": 0.2821,
+      "step": 8982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599721099183475e-05,
+      "loss": 0.2624,
+      "step": 8983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5598730856989842e-05,
+      "loss": 0.2443,
+      "step": 8984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597740534823773e-05,
+      "loss": 0.2737,
+      "step": 8985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5596750132699417e-05,
+      "loss": 0.3122,
+      "step": 8986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559575965063091e-05,
+      "loss": 0.2779,
+      "step": 8987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5594769088632417e-05,
+      "loss": 0.3068,
+      "step": 8988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559377844671807e-05,
+      "loss": 0.2247,
+      "step": 8989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559278772490203e-05,
+      "loss": 0.2445,
+      "step": 8990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591796923198442e-05,
+      "loss": 0.2725,
+      "step": 8991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5590806041621464e-05,
+      "loss": 0.2689,
+      "step": 8992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5589815080185247e-05,
+      "loss": 0.2237,
+      "step": 8993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588824038903947e-05,
+      "loss": 0.2399,
+      "step": 8994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5587832917791722e-05,
+      "loss": 0.2754,
+      "step": 8995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586841716862723e-05,
+      "loss": 0.2647,
+      "step": 8996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585850436131114e-05,
+      "loss": 0.2922,
+      "step": 8997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558485907561105e-05,
+      "loss": 0.2878,
+      "step": 8998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55838676353167e-05,
+      "loss": 0.2531,
+      "step": 8999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5582876115262218e-05,
+      "loss": 0.2794,
+      "step": 9000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5581884515461768e-05,
+      "loss": 0.2891,
+      "step": 9001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580892835929514e-05,
+      "loss": 0.2889,
+      "step": 9002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579901076679625e-05,
+      "loss": 0.2509,
+      "step": 9003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557890923772626e-05,
+      "loss": 0.2703,
+      "step": 9004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5577917319083598e-05,
+      "loss": 0.2783,
+      "step": 9005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576925320765795e-05,
+      "loss": 0.2561,
+      "step": 9006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557593324278703e-05,
+      "loss": 0.2581,
+      "step": 9007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5574941085161466e-05,
+      "loss": 0.2817,
+      "step": 9008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573948847903282e-05,
+      "loss": 0.2623,
+      "step": 9009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5572956531026647e-05,
+      "loss": 0.2553,
+      "step": 9010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5571964134545736e-05,
+      "loss": 0.2856,
+      "step": 9011
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570971658474727e-05,
+      "loss": 0.2777,
+      "step": 9012
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569979102827795e-05,
+      "loss": 0.2863,
+      "step": 9013
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5568986467619116e-05,
+      "loss": 0.2925,
+      "step": 9014
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556799375286287e-05,
+      "loss": 0.258,
+      "step": 9015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5567000958573236e-05,
+      "loss": 0.2733,
+      "step": 9016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5566008084764398e-05,
+      "loss": 0.248,
+      "step": 9017
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565015131450537e-05,
+      "loss": 0.2673,
+      "step": 9018
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5564022098645835e-05,
+      "loss": 0.2373,
+      "step": 9019
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563028986364476e-05,
+      "loss": 0.299,
+      "step": 9020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562035794620653e-05,
+      "loss": 0.2396,
+      "step": 9021
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556104252342854e-05,
+      "loss": 0.2697,
+      "step": 9022
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5560049172802335e-05,
+      "loss": 0.2673,
+      "step": 9023
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555905574275622e-05,
+      "loss": 0.2985,
+      "step": 9024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5558062233304392e-05,
+      "loss": 0.3038,
+      "step": 9025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557068644461036e-05,
+      "loss": 0.274,
+      "step": 9026
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556074976240353e-05,
+      "loss": 0.2592,
+      "step": 9027
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555508122865653e-05,
+      "loss": 0.3101,
+      "step": 9028
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5554087401723758e-05,
+      "loss": 0.2405,
+      "step": 9029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553093495456245e-05,
+      "loss": 0.2539,
+      "step": 9030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552099509868176e-05,
+      "loss": 0.2668,
+      "step": 9031
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5551105444973754e-05,
+      "loss": 0.2803,
+      "step": 9032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550111300787182e-05,
+      "loss": 0.3009,
+      "step": 9033
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554911707732265e-05,
+      "loss": 0.288,
+      "step": 9034
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5548122774594372e-05,
+      "loss": 0.2438,
+      "step": 9035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547128392616545e-05,
+      "loss": 0.2243,
+      "step": 9036
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5546133931403365e-05,
+      "loss": 0.2527,
+      "step": 9037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5545139390969054e-05,
+      "loss": 0.2257,
+      "step": 9038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55441447713278e-05,
+      "loss": 0.2525,
+      "step": 9039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5543150072493823e-05,
+      "loss": 0.2306,
+      "step": 9040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554215529448133e-05,
+      "loss": 0.2506,
+      "step": 9041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5541160437304524e-05,
+      "loss": 0.2674,
+      "step": 9042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5540165500977618e-05,
+      "loss": 0.2403,
+      "step": 9043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5539170485514826e-05,
+      "loss": 0.2536,
+      "step": 9044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553817539093036e-05,
+      "loss": 0.2447,
+      "step": 9045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553718021723843e-05,
+      "loss": 0.2445,
+      "step": 9046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553618496445326e-05,
+      "loss": 0.244,
+      "step": 9047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5535189632589057e-05,
+      "loss": 0.2907,
+      "step": 9048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534194221660043e-05,
+      "loss": 0.2579,
+      "step": 9049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533198731680436e-05,
+      "loss": 0.2413,
+      "step": 9050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5532203162664456e-05,
+      "loss": 0.2512,
+      "step": 9051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553120751462632e-05,
+      "loss": 0.2402,
+      "step": 9052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530211787580256e-05,
+      "loss": 0.2487,
+      "step": 9053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5529215981540483e-05,
+      "loss": 0.2502,
+      "step": 9054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5528220096521227e-05,
+      "loss": 0.2436,
+      "step": 9055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552722413253671e-05,
+      "loss": 0.2421,
+      "step": 9056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552622808960116e-05,
+      "loss": 0.2313,
+      "step": 9057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525231967728806e-05,
+      "loss": 0.2237,
+      "step": 9058
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524235766933878e-05,
+      "loss": 0.2367,
+      "step": 9059
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55232394872306e-05,
+      "loss": 0.2291,
+      "step": 9060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552224312863321e-05,
+      "loss": 0.2267,
+      "step": 9061
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521246691155933e-05,
+      "loss": 0.2286,
+      "step": 9062
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552025017481301e-05,
+      "loss": 0.2667,
+      "step": 9063
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5519253579618666e-05,
+      "loss": 0.2694,
+      "step": 9064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5518256905587142e-05,
+      "loss": 0.2561,
+      "step": 9065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5517260152732677e-05,
+      "loss": 0.2447,
+      "step": 9066
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5516263321069504e-05,
+      "loss": 0.2333,
+      "step": 9067
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551526641061186e-05,
+      "loss": 0.2888,
+      "step": 9068
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5514269421373992e-05,
+      "loss": 0.2339,
+      "step": 9069
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551327235337013e-05,
+      "loss": 0.2252,
+      "step": 9070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551227520661453e-05,
+      "loss": 0.2251,
+      "step": 9071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5511277981121426e-05,
+      "loss": 0.2407,
+      "step": 9072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510280676905067e-05,
+      "loss": 0.2582,
+      "step": 9073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5509283293979692e-05,
+      "loss": 0.2581,
+      "step": 9074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550828583235956e-05,
+      "loss": 0.2464,
+      "step": 9075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507288292058902e-05,
+      "loss": 0.2504,
+      "step": 9076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550629067309198e-05,
+      "loss": 0.2509,
+      "step": 9077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550529297547304e-05,
+      "loss": 0.2131,
+      "step": 9078
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550429519921633e-05,
+      "loss": 0.2566,
+      "step": 9079
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503297344336108e-05,
+      "loss": 0.2501,
+      "step": 9080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5502299410846626e-05,
+      "loss": 0.2657,
+      "step": 9081
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5501301398762132e-05,
+      "loss": 0.2567,
+      "step": 9082
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550030330809689e-05,
+      "loss": 0.2581,
+      "step": 9083
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5499305138865152e-05,
+      "loss": 0.2202,
+      "step": 9084
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498306891081178e-05,
+      "loss": 0.2788,
+      "step": 9085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5497308564759224e-05,
+      "loss": 0.2491,
+      "step": 9086
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549631015991356e-05,
+      "loss": 0.2525,
+      "step": 9087
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495311676558436e-05,
+      "loss": 0.243,
+      "step": 9088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5494313114708113e-05,
+      "loss": 0.2461,
+      "step": 9089
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5493314474376863e-05,
+      "loss": 0.2507,
+      "step": 9090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492315755578947e-05,
+      "loss": 0.2161,
+      "step": 9091
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549131695832863e-05,
+      "loss": 0.2547,
+      "step": 9092
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5490318082640182e-05,
+      "loss": 0.267,
+      "step": 9093
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489319128527868e-05,
+      "loss": 0.2335,
+      "step": 9094
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5488320096005957e-05,
+      "loss": 0.2514,
+      "step": 9095
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548732098508872e-05,
+      "loss": 0.2128,
+      "step": 9096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486321795790427e-05,
+      "loss": 0.2516,
+      "step": 9097
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485322528125353e-05,
+      "loss": 0.246,
+      "step": 9098
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548432318210777e-05,
+      "loss": 0.2092,
+      "step": 9099
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5483323757751955e-05,
+      "loss": 0.2367,
+      "step": 9100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548232425507218e-05,
+      "loss": 0.2553,
+      "step": 9101
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481324674082724e-05,
+      "loss": 0.2566,
+      "step": 9102
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480325014797862e-05,
+      "loss": 0.2479,
+      "step": 9103
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479325277231882e-05,
+      "loss": 0.2397,
+      "step": 9104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5478325461399055e-05,
+      "loss": 0.2438,
+      "step": 9105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477325567313662e-05,
+      "loss": 0.2474,
+      "step": 9106
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5476325594989998e-05,
+      "loss": 0.2595,
+      "step": 9107
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475325544442332e-05,
+      "loss": 0.2771,
+      "step": 9108
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5474325415684953e-05,
+      "loss": 0.2285,
+      "step": 9109
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5473325208732152e-05,
+      "loss": 0.2559,
+      "step": 9110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472324923598207e-05,
+      "loss": 0.2235,
+      "step": 9111
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547132456029742e-05,
+      "loss": 0.2629,
+      "step": 9112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5470324118844064e-05,
+      "loss": 0.2213,
+      "step": 9113
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469323599252443e-05,
+      "loss": 0.282,
+      "step": 9114
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468323001536835e-05,
+      "loss": 0.2263,
+      "step": 9115
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5467322325711548e-05,
+      "loss": 0.2544,
+      "step": 9116
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5466321571790864e-05,
+      "loss": 0.2494,
+      "step": 9117
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546532073978908e-05,
+      "loss": 0.2321,
+      "step": 9118
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5464319829720494e-05,
+      "loss": 0.2459,
+      "step": 9119
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463318841599408e-05,
+      "loss": 0.2378,
+      "step": 9120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546231777544011e-05,
+      "loss": 0.2544,
+      "step": 9121
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5461316631256905e-05,
+      "loss": 0.265,
+      "step": 9122
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546031540906409e-05,
+      "loss": 0.2317,
+      "step": 9123
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545931410887597e-05,
+      "loss": 0.2263,
+      "step": 9124
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458312730706847e-05,
+      "loss": 0.238,
+      "step": 9125
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457311274571024e-05,
+      "loss": 0.2204,
+      "step": 9126
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5456309740482804e-05,
+      "loss": 0.229,
+      "step": 9127
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455308128456496e-05,
+      "loss": 0.2409,
+      "step": 9128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454306438506404e-05,
+      "loss": 0.2395,
+      "step": 9129
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545330467064684e-05,
+      "loss": 0.2554,
+      "step": 9130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452302824892112e-05,
+      "loss": 0.2542,
+      "step": 9131
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451300901256533e-05,
+      "loss": 0.2499,
+      "step": 9132
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450298899754407e-05,
+      "loss": 0.2754,
+      "step": 9133
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5449296820400053e-05,
+      "loss": 0.2594,
+      "step": 9134
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448294663207783e-05,
+      "loss": 0.2455,
+      "step": 9135
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447292428191904e-05,
+      "loss": 0.2311,
+      "step": 9136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5446290115366745e-05,
+      "loss": 0.2432,
+      "step": 9137
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445287724746618e-05,
+      "loss": 0.2469,
+      "step": 9138
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544428525634584e-05,
+      "loss": 0.2304,
+      "step": 9139
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5443282710178734e-05,
+      "loss": 0.2334,
+      "step": 9140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5442280086259614e-05,
+      "loss": 0.2655,
+      "step": 9141
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5441277384602802e-05,
+      "loss": 0.2488,
+      "step": 9142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544027460522263e-05,
+      "loss": 0.2488,
+      "step": 9143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5439271748133408e-05,
+      "loss": 0.2365,
+      "step": 9144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543826881334947e-05,
+      "loss": 0.2312,
+      "step": 9145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437265800885145e-05,
+      "loss": 0.2435,
+      "step": 9146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543626271075475e-05,
+      "loss": 0.2417,
+      "step": 9147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543525954297262e-05,
+      "loss": 0.2515,
+      "step": 9148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5434256297553085e-05,
+      "loss": 0.2709,
+      "step": 9149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543325297451047e-05,
+      "loss": 0.2485,
+      "step": 9150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5432249573859108e-05,
+      "loss": 0.2314,
+      "step": 9151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543124609561333e-05,
+      "loss": 0.2447,
+      "step": 9152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430242539787482e-05,
+      "loss": 0.228,
+      "step": 9153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5429238906395884e-05,
+      "loss": 0.2348,
+      "step": 9154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5428235195452878e-05,
+      "loss": 0.2478,
+      "step": 9155
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54272314069728e-05,
+      "loss": 0.2326,
+      "step": 9156
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5426227540969988e-05,
+      "loss": 0.2721,
+      "step": 9157
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5425223597458784e-05,
+      "loss": 0.2631,
+      "step": 9158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5424219576453526e-05,
+      "loss": 0.2146,
+      "step": 9159
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5423215477968555e-05,
+      "loss": 0.2377,
+      "step": 9160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422211302018216e-05,
+      "loss": 0.2329,
+      "step": 9161
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542120704861685e-05,
+      "loss": 0.2455,
+      "step": 9162
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54202027177788e-05,
+      "loss": 0.2392,
+      "step": 9163
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419198309518417e-05,
+      "loss": 0.2296,
+      "step": 9164
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5418193823850043e-05,
+      "loss": 0.2514,
+      "step": 9165
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5417189260788033e-05,
+      "loss": 0.2631,
+      "step": 9166
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5416184620346727e-05,
+      "loss": 0.2426,
+      "step": 9167
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5415179902540484e-05,
+      "loss": 0.224,
+      "step": 9168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541417510738365e-05,
+      "loss": 0.2545,
+      "step": 9169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541317023489058e-05,
+      "loss": 0.2283,
+      "step": 9170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5412165285075625e-05,
+      "loss": 0.2458,
+      "step": 9171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5411160257953143e-05,
+      "loss": 0.2539,
+      "step": 9172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541015515353749e-05,
+      "loss": 0.246,
+      "step": 9173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540914997184302e-05,
+      "loss": 0.2508,
+      "step": 9174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408144712884094e-05,
+      "loss": 0.2275,
+      "step": 9175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5407139376675067e-05,
+      "loss": 0.2461,
+      "step": 9176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.54061339632303e-05,
+      "loss": 0.2419,
+      "step": 9177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540512847256416e-05,
+      "loss": 0.2457,
+      "step": 9178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5404122904691002e-05,
+      "loss": 0.2484,
+      "step": 9179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5403117259625195e-05,
+      "loss": 0.2511,
+      "step": 9180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5402111537381107e-05,
+      "loss": 0.2353,
+      "step": 9181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540110573797309e-05,
+      "loss": 0.2514,
+      "step": 9182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400099861415522e-05,
+      "loss": 0.2588,
+      "step": 9183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5399093907722767e-05,
+      "loss": 0.2422,
+      "step": 9184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53980878769092e-05,
+      "loss": 0.2217,
+      "step": 9185
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5397081768989185e-05,
+      "loss": 0.2304,
+      "step": 9186
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5396075583977096e-05,
+      "loss": 0.2425,
+      "step": 9187
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5395069321887302e-05,
+      "loss": 0.2494,
+      "step": 9188
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5394062982734176e-05,
+      "loss": 0.2212,
+      "step": 9189
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53930565665321e-05,
+      "loss": 0.224,
+      "step": 9190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392050073295446e-05,
+      "loss": 0.2302,
+      "step": 9191
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539104350303859e-05,
+      "loss": 0.2295,
+      "step": 9192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390036855775905e-05,
+      "loss": 0.2687,
+      "step": 9193
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5389030131521777e-05,
+      "loss": 0.2548,
+      "step": 9194
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5388023330290583e-05,
+      "loss": 0.2207,
+      "step": 9195
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387016452096704e-05,
+      "loss": 0.2441,
+      "step": 9196
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386009496954528e-05,
+      "loss": 0.2605,
+      "step": 9197
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538500246487843e-05,
+      "loss": 0.2258,
+      "step": 9198
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5383995355882797e-05,
+      "loss": 0.2199,
+      "step": 9199
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538298816998202e-05,
+      "loss": 0.2429,
+      "step": 9200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5381980907190475e-05,
+      "loss": 0.2709,
+      "step": 9201
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380973567522558e-05,
+      "loss": 0.2496,
+      "step": 9202
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537996615099266e-05,
+      "loss": 0.2296,
+      "step": 9203
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378958657615165e-05,
+      "loss": 0.2424,
+      "step": 9204
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5377951087404466e-05,
+      "loss": 0.2427,
+      "step": 9205
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5376943440374953e-05,
+      "loss": 0.2481,
+      "step": 9206
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375935716541022e-05,
+      "loss": 0.2498,
+      "step": 9207
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537492791591707e-05,
+      "loss": 0.246,
+      "step": 9208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5373920038517485e-05,
+      "loss": 0.2598,
+      "step": 9209
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537291208435667e-05,
+      "loss": 0.2447,
+      "step": 9210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537190405344902e-05,
+      "loss": 0.2402,
+      "step": 9211
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5370895945808934e-05,
+      "loss": 0.2624,
+      "step": 9212
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5369887761450813e-05,
+      "loss": 0.2444,
+      "step": 9213
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5368879500389054e-05,
+      "loss": 0.2364,
+      "step": 9214
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5367871162638067e-05,
+      "loss": 0.2426,
+      "step": 9215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5366862748212245e-05,
+      "loss": 0.2467,
+      "step": 9216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365854257126e-05,
+      "loss": 0.2449,
+      "step": 9217
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536484568939373e-05,
+      "loss": 0.2437,
+      "step": 9218
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363837045029855e-05,
+      "loss": 0.2523,
+      "step": 9219
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536282832404877e-05,
+      "loss": 0.2327,
+      "step": 9220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5361819526464886e-05,
+      "loss": 0.2135,
+      "step": 9221
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360810652292613e-05,
+      "loss": 0.2446,
+      "step": 9222
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5359801701546363e-05,
+      "loss": 0.2536,
+      "step": 9223
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358792674240548e-05,
+      "loss": 0.2393,
+      "step": 9224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357783570389582e-05,
+      "loss": 0.2383,
+      "step": 9225
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356774390007875e-05,
+      "loss": 0.2457,
+      "step": 9226
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355765133109842e-05,
+      "loss": 0.2404,
+      "step": 9227
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535475579970991e-05,
+      "loss": 0.236,
+      "step": 9228
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353746389822483e-05,
+      "loss": 0.2593,
+      "step": 9229
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352736903461985e-05,
+      "loss": 0.2645,
+      "step": 9230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5351727340642836e-05,
+      "loss": 0.2339,
+      "step": 9231
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5350717701379454e-05,
+      "loss": 0.2484,
+      "step": 9232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349707985686263e-05,
+      "loss": 0.24,
+      "step": 9233
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534869819357769e-05,
+      "loss": 0.2465,
+      "step": 9234
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5347688325068148e-05,
+      "loss": 0.2372,
+      "step": 9235
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346678380172073e-05,
+      "loss": 0.242,
+      "step": 9236
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345668358903886e-05,
+      "loss": 0.2276,
+      "step": 9237
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344658261278013e-05,
+      "loss": 0.25,
+      "step": 9238
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343648087308885e-05,
+      "loss": 0.2278,
+      "step": 9239
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5342637837010927e-05,
+      "loss": 0.2313,
+      "step": 9240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5341627510398577e-05,
+      "loss": 0.2258,
+      "step": 9241
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534061710748626e-05,
+      "loss": 0.2533,
+      "step": 9242
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533960662828841e-05,
+      "loss": 0.2193,
+      "step": 9243
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533859607281946e-05,
+      "loss": 0.2184,
+      "step": 9244
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337585441093853e-05,
+      "loss": 0.2418,
+      "step": 9245
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336574733126013e-05,
+      "loss": 0.2208,
+      "step": 9246
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5335563948930385e-05,
+      "loss": 0.2154,
+      "step": 9247
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5334553088521403e-05,
+      "loss": 0.2641,
+      "step": 9248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533354215191351e-05,
+      "loss": 0.2302,
+      "step": 9249
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332531139121137e-05,
+      "loss": 0.2407,
+      "step": 9250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5331520050158735e-05,
+      "loss": 0.253,
+      "step": 9251
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5330508885040746e-05,
+      "loss": 0.2344,
+      "step": 9252
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532949764378161e-05,
+      "loss": 0.2348,
+      "step": 9253
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5328486326395773e-05,
+      "loss": 0.2504,
+      "step": 9254
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5327474932897677e-05,
+      "loss": 0.2452,
+      "step": 9255
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326463463301772e-05,
+      "loss": 0.2328,
+      "step": 9256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5325451917622508e-05,
+      "loss": 0.2707,
+      "step": 9257
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5324440295874336e-05,
+      "loss": 0.2281,
+      "step": 9258
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5323428598071698e-05,
+      "loss": 0.2283,
+      "step": 9259
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5322416824229046e-05,
+      "loss": 0.2444,
+      "step": 9260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5321404974360837e-05,
+      "loss": 0.267,
+      "step": 9261
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320393048481524e-05,
+      "loss": 0.2519,
+      "step": 9262
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531938104660556e-05,
+      "loss": 0.2544,
+      "step": 9263
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53183689687474e-05,
+      "loss": 0.2462,
+      "step": 9264
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53173568149215e-05,
+      "loss": 0.2392,
+      "step": 9265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531634458514232e-05,
+      "loss": 0.2334,
+      "step": 9266
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315332279424314e-05,
+      "loss": 0.2659,
+      "step": 9267
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314319897781946e-05,
+      "loss": 0.2469,
+      "step": 9268
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313307440229678e-05,
+      "loss": 0.2409,
+      "step": 9269
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531229490678197e-05,
+      "loss": 0.2365,
+      "step": 9270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311282297453286e-05,
+      "loss": 0.2391,
+      "step": 9271
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5310269612258088e-05,
+      "loss": 0.2342,
+      "step": 9272
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309256851210838e-05,
+      "loss": 0.2514,
+      "step": 9273
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530824401432601e-05,
+      "loss": 0.2176,
+      "step": 9274
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530723110161807e-05,
+      "loss": 0.2408,
+      "step": 9275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5306218113101482e-05,
+      "loss": 0.2382,
+      "step": 9276
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5305205048790718e-05,
+      "loss": 0.2312,
+      "step": 9277
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530419190870025e-05,
+      "loss": 0.2349,
+      "step": 9278
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303178692844547e-05,
+      "loss": 0.2717,
+      "step": 9279
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302165401238086e-05,
+      "loss": 0.2353,
+      "step": 9280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5301152033895335e-05,
+      "loss": 0.2204,
+      "step": 9281
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5300138590830773e-05,
+      "loss": 0.2339,
+      "step": 9282
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299125072058877e-05,
+      "loss": 0.2205,
+      "step": 9283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529811147759412e-05,
+      "loss": 0.257,
+      "step": 9284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5297097807450984e-05,
+      "loss": 0.2309,
+      "step": 9285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296084061643945e-05,
+      "loss": 0.2364,
+      "step": 9286
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5295070240187485e-05,
+      "loss": 0.2467,
+      "step": 9287
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5294056343096085e-05,
+      "loss": 0.2632,
+      "step": 9288
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529304237038423e-05,
+      "loss": 0.237,
+      "step": 9289
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5292028322066404e-05,
+      "loss": 0.2287,
+      "step": 9290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291014198157085e-05,
+      "loss": 0.2332,
+      "step": 9291
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528999999867077e-05,
+      "loss": 0.228,
+      "step": 9292
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288985723621932e-05,
+      "loss": 0.2357,
+      "step": 9293
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5287971373025067e-05,
+      "loss": 0.2648,
+      "step": 9294
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5286956946894663e-05,
+      "loss": 0.256,
+      "step": 9295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5285942445245214e-05,
+      "loss": 0.249,
+      "step": 9296
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284927868091203e-05,
+      "loss": 0.2303,
+      "step": 9297
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528391321544713e-05,
+      "loss": 0.2428,
+      "step": 9298
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528289848732748e-05,
+      "loss": 0.253,
+      "step": 9299
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5281883683746757e-05,
+      "loss": 0.2463,
+      "step": 9300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528086880471945e-05,
+      "loss": 0.2462,
+      "step": 9301
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527985385026006e-05,
+      "loss": 0.2102,
+      "step": 9302
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527883882038308e-05,
+      "loss": 0.2479,
+      "step": 9303
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527782371510301e-05,
+      "loss": 0.2381,
+      "step": 9304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276808534434348e-05,
+      "loss": 0.2146,
+      "step": 9305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5275793278391603e-05,
+      "loss": 0.2402,
+      "step": 9306
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5274777946989264e-05,
+      "loss": 0.2614,
+      "step": 9307
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273762540241848e-05,
+      "loss": 0.286,
+      "step": 9308
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527274705816385e-05,
+      "loss": 0.2174,
+      "step": 9309
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5271731500769778e-05,
+      "loss": 0.2625,
+      "step": 9310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5270715868074137e-05,
+      "loss": 0.2457,
+      "step": 9311
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5269700160091437e-05,
+      "loss": 0.2255,
+      "step": 9312
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268684376836184e-05,
+      "loss": 0.2224,
+      "step": 9313
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5267668518322888e-05,
+      "loss": 0.2322,
+      "step": 9314
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5266652584566056e-05,
+      "loss": 0.2233,
+      "step": 9315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5265636575580206e-05,
+      "loss": 0.2424,
+      "step": 9316
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526462049137985e-05,
+      "loss": 0.1992,
+      "step": 9317
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52636043319795e-05,
+      "loss": 0.2307,
+      "step": 9318
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262588097393667e-05,
+      "loss": 0.2596,
+      "step": 9319
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526157178763687e-05,
+      "loss": 0.2411,
+      "step": 9320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526055540272363e-05,
+      "loss": 0.2283,
+      "step": 9321
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259538942668458e-05,
+      "loss": 0.2329,
+      "step": 9322
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5258522407485879e-05,
+      "loss": 0.2333,
+      "step": 9323
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257505797190413e-05,
+      "loss": 0.2374,
+      "step": 9324
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256489111796577e-05,
+      "loss": 0.2307,
+      "step": 9325
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5255472351318897e-05,
+      "loss": 0.2382,
+      "step": 9326
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254455515771892e-05,
+      "loss": 0.2507,
+      "step": 9327
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253438605170088e-05,
+      "loss": 0.2441,
+      "step": 9328
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252421619528018e-05,
+      "loss": 0.2528,
+      "step": 9329
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52514045588602e-05,
+      "loss": 0.2466,
+      "step": 9330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250387423181164e-05,
+      "loss": 0.2529,
+      "step": 9331
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5249370212505438e-05,
+      "loss": 0.245,
+      "step": 9332
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5248352926847556e-05,
+      "loss": 0.2559,
+      "step": 9333
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5247335566222043e-05,
+      "loss": 0.2481,
+      "step": 9334
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246318130643439e-05,
+      "loss": 0.2686,
+      "step": 9335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245300620126266e-05,
+      "loss": 0.2597,
+      "step": 9336
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524428303468507e-05,
+      "loss": 0.2683,
+      "step": 9337
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243265374334377e-05,
+      "loss": 0.2463,
+      "step": 9338
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5242247639088727e-05,
+      "loss": 0.2586,
+      "step": 9339
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241229828962658e-05,
+      "loss": 0.2279,
+      "step": 9340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5240211943970708e-05,
+      "loss": 0.2386,
+      "step": 9341
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239193984127416e-05,
+      "loss": 0.2523,
+      "step": 9342
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523817594944732e-05,
+      "loss": 0.2431,
+      "step": 9343
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523715783994497e-05,
+      "loss": 0.279,
+      "step": 9344
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5236139655634899e-05,
+      "loss": 0.234,
+      "step": 9345
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5235121396531657e-05,
+      "loss": 0.2605,
+      "step": 9346
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234103062649786e-05,
+      "loss": 0.2357,
+      "step": 9347
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523308465400383e-05,
+      "loss": 0.2722,
+      "step": 9348
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523206617060834e-05,
+      "loss": 0.2414,
+      "step": 9349
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5231047612477863e-05,
+      "loss": 0.2203,
+      "step": 9350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230028979626948e-05,
+      "loss": 0.2395,
+      "step": 9351
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5229010272070146e-05,
+      "loss": 0.2293,
+      "step": 9352
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5227991489822005e-05,
+      "loss": 0.2373,
+      "step": 9353
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5226972632897079e-05,
+      "loss": 0.2695,
+      "step": 9354
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225953701309922e-05,
+      "loss": 0.232,
+      "step": 9355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522493469507509e-05,
+      "loss": 0.2406,
+      "step": 9356
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223915614207137e-05,
+      "loss": 0.2388,
+      "step": 9357
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522289645872062e-05,
+      "loss": 0.252,
+      "step": 9358
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221877228630092e-05,
+      "loss": 0.2709,
+      "step": 9359
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5220857923950121e-05,
+      "loss": 0.2412,
+      "step": 9360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5219838544695255e-05,
+      "loss": 0.2438,
+      "step": 9361
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5218819090880068e-05,
+      "loss": 0.2609,
+      "step": 9362
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5217799562519111e-05,
+      "loss": 0.2246,
+      "step": 9363
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216779959626956e-05,
+      "loss": 0.238,
+      "step": 9364
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5215760282218157e-05,
+      "loss": 0.2385,
+      "step": 9365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5214740530307285e-05,
+      "loss": 0.2339,
+      "step": 9366
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213720703908905e-05,
+      "loss": 0.256,
+      "step": 9367
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521270080303759e-05,
+      "loss": 0.2315,
+      "step": 9368
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211680827707898e-05,
+      "loss": 0.2585,
+      "step": 9369
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210660777934407e-05,
+      "loss": 0.2301,
+      "step": 9370
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520964065373168e-05,
+      "loss": 0.2382,
+      "step": 9371
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5208620455114296e-05,
+      "loss": 0.2363,
+      "step": 9372
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520760018209682e-05,
+      "loss": 0.2344,
+      "step": 9373
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5206579834693832e-05,
+      "loss": 0.2514,
+      "step": 9374
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205559412919905e-05,
+      "loss": 0.228,
+      "step": 9375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5204538916789613e-05,
+      "loss": 0.2727,
+      "step": 9376
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5203518346317534e-05,
+      "loss": 0.2675,
+      "step": 9377
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202497701518244e-05,
+      "loss": 0.2367,
+      "step": 9378
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5201476982406327e-05,
+      "loss": 0.2678,
+      "step": 9379
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200456188996355e-05,
+      "loss": 0.2596,
+      "step": 9380
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199435321302915e-05,
+      "loss": 0.2492,
+      "step": 9381
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198414379340588e-05,
+      "loss": 0.248,
+      "step": 9382
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5197393363123958e-05,
+      "loss": 0.2476,
+      "step": 9383
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196372272667606e-05,
+      "loss": 0.2282,
+      "step": 9384
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5195351107986121e-05,
+      "loss": 0.251,
+      "step": 9385
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194329869094087e-05,
+      "loss": 0.2262,
+      "step": 9386
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193308556006092e-05,
+      "loss": 0.2506,
+      "step": 9387
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192287168736723e-05,
+      "loss": 0.2274,
+      "step": 9388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519126570730057e-05,
+      "loss": 0.249,
+      "step": 9389
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519024417171223e-05,
+      "loss": 0.2394,
+      "step": 9390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189222561986288e-05,
+      "loss": 0.254,
+      "step": 9391
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5188200878137335e-05,
+      "loss": 0.2278,
+      "step": 9392
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187179120179969e-05,
+      "loss": 0.2366,
+      "step": 9393
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186157288128782e-05,
+      "loss": 0.2551,
+      "step": 9394
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5185135381998373e-05,
+      "loss": 0.2509,
+      "step": 9395
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184113401803338e-05,
+      "loss": 0.2271,
+      "step": 9396
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183091347558275e-05,
+      "loss": 0.2475,
+      "step": 9397
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518206921927778e-05,
+      "loss": 0.2497,
+      "step": 9398
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5181047016976456e-05,
+      "loss": 0.2498,
+      "step": 9399
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180024740668903e-05,
+      "loss": 0.2619,
+      "step": 9400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5179002390369725e-05,
+      "loss": 0.2622,
+      "step": 9401
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177979966093525e-05,
+      "loss": 0.2367,
+      "step": 9402
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176957467854907e-05,
+      "loss": 0.2661,
+      "step": 9403
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175934895668474e-05,
+      "loss": 0.2274,
+      "step": 9404
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5174912249548834e-05,
+      "loss": 0.2498,
+      "step": 9405
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173889529510598e-05,
+      "loss": 0.2608,
+      "step": 9406
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517286673556837e-05,
+      "loss": 0.2697,
+      "step": 9407
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517184386773676e-05,
+      "loss": 0.2251,
+      "step": 9408
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5170820926030381e-05,
+      "loss": 0.231,
+      "step": 9409
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169797910463841e-05,
+      "loss": 0.2535,
+      "step": 9410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516877482105176e-05,
+      "loss": 0.24,
+      "step": 9411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516775165780874e-05,
+      "loss": 0.266,
+      "step": 9412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166728420749408e-05,
+      "loss": 0.2363,
+      "step": 9413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165705109888376e-05,
+      "loss": 0.2426,
+      "step": 9414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164681725240255e-05,
+      "loss": 0.2399,
+      "step": 9415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163658266819675e-05,
+      "loss": 0.2711,
+      "step": 9416
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5162634734641244e-05,
+      "loss": 0.2548,
+      "step": 9417
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161611128719585e-05,
+      "loss": 0.2187,
+      "step": 9418
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160587449069321e-05,
+      "loss": 0.2304,
+      "step": 9419
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5159563695705075e-05,
+      "loss": 0.2734,
+      "step": 9420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158539868641468e-05,
+      "loss": 0.2477,
+      "step": 9421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157515967893125e-05,
+      "loss": 0.2391,
+      "step": 9422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5156491993474673e-05,
+      "loss": 0.2305,
+      "step": 9423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5155467945400739e-05,
+      "loss": 0.2071,
+      "step": 9424
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154443823685947e-05,
+      "loss": 0.2571,
+      "step": 9425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515341962834493e-05,
+      "loss": 0.2288,
+      "step": 9426
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515239535939231e-05,
+      "loss": 0.2257,
+      "step": 9427
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151371016842726e-05,
+      "loss": 0.2236,
+      "step": 9428
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5150346600710808e-05,
+      "loss": 0.2487,
+      "step": 9429
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5149322111011187e-05,
+      "loss": 0.2374,
+      "step": 9430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148297547758494e-05,
+      "loss": 0.2474,
+      "step": 9431
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5147272910967368e-05,
+      "loss": 0.2338,
+      "step": 9432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5146248200652444e-05,
+      "loss": 0.2264,
+      "step": 9433
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5145223416828362e-05,
+      "loss": 0.2307,
+      "step": 9434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144198559509756e-05,
+      "loss": 0.2095,
+      "step": 9435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143173628711264e-05,
+      "loss": 0.2264,
+      "step": 9436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514214862444753e-05,
+      "loss": 0.2443,
+      "step": 9437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141123546733192e-05,
+      "loss": 0.2419,
+      "step": 9438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5140098395582895e-05,
+      "loss": 0.2447,
+      "step": 9439
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513907317101128e-05,
+      "loss": 0.2397,
+      "step": 9440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5138047873032993e-05,
+      "loss": 0.2535,
+      "step": 9441
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137022501662677e-05,
+      "loss": 0.2567,
+      "step": 9442
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135997056914984e-05,
+      "loss": 0.2445,
+      "step": 9443
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5134971538804555e-05,
+      "loss": 0.2509,
+      "step": 9444
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513394594734604e-05,
+      "loss": 0.255,
+      "step": 9445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5132920282554093e-05,
+      "loss": 0.2322,
+      "step": 9446
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513189454444336e-05,
+      "loss": 0.2397,
+      "step": 9447
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130868733028493e-05,
+      "loss": 0.2323,
+      "step": 9448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129842848324148e-05,
+      "loss": 0.2459,
+      "step": 9449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128816890344973e-05,
+      "loss": 0.2402,
+      "step": 9450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512779085910563e-05,
+      "loss": 0.2446,
+      "step": 9451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126764754620769e-05,
+      "loss": 0.2314,
+      "step": 9452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512573857690505e-05,
+      "loss": 0.2336,
+      "step": 9453
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124712325973127e-05,
+      "loss": 0.2673,
+      "step": 9454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5123686001839666e-05,
+      "loss": 0.2395,
+      "step": 9455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5122659604519325e-05,
+      "loss": 0.2544,
+      "step": 9456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121633134026761e-05,
+      "loss": 0.2235,
+      "step": 9457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5120606590376637e-05,
+      "loss": 0.2233,
+      "step": 9458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5119579973583616e-05,
+      "loss": 0.2417,
+      "step": 9459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118553283662367e-05,
+      "loss": 0.2473,
+      "step": 9460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511752652062755e-05,
+      "loss": 0.2381,
+      "step": 9461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5116499684493836e-05,
+      "loss": 0.2415,
+      "step": 9462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115472775275888e-05,
+      "loss": 0.2506,
+      "step": 9463
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5114445792988375e-05,
+      "loss": 0.2451,
+      "step": 9464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5113418737645968e-05,
+      "loss": 0.2526,
+      "step": 9465
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112391609263336e-05,
+      "loss": 0.2302,
+      "step": 9466
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5111364407855155e-05,
+      "loss": 0.2469,
+      "step": 9467
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110337133436093e-05,
+      "loss": 0.2446,
+      "step": 9468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109309786020824e-05,
+      "loss": 0.243,
+      "step": 9469
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108282365624023e-05,
+      "loss": 0.2611,
+      "step": 9470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5107254872260366e-05,
+      "loss": 0.2198,
+      "step": 9471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510622730594453e-05,
+      "loss": 0.2588,
+      "step": 9472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105199666691194e-05,
+      "loss": 0.2222,
+      "step": 9473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5104171954515034e-05,
+      "loss": 0.2884,
+      "step": 9474
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103144169430733e-05,
+      "loss": 0.2469,
+      "step": 9475
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510211631145297e-05,
+      "loss": 0.2525,
+      "step": 9476
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101088380596425e-05,
+      "loss": 0.2415,
+      "step": 9477
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5100060376875782e-05,
+      "loss": 0.2356,
+      "step": 9478
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099032300305731e-05,
+      "loss": 0.2299,
+      "step": 9479
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5098004150900952e-05,
+      "loss": 0.2571,
+      "step": 9480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5096975928676125e-05,
+      "loss": 0.2649,
+      "step": 9481
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095947633645952e-05,
+      "loss": 0.2225,
+      "step": 9482
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5094919265825107e-05,
+      "loss": 0.2322,
+      "step": 9483
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093890825228289e-05,
+      "loss": 0.2534,
+      "step": 9484
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509286231187018e-05,
+      "loss": 0.2307,
+      "step": 9485
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091833725765477e-05,
+      "loss": 0.2702,
+      "step": 9486
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5090805066928872e-05,
+      "loss": 0.2247,
+      "step": 9487
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089776335375057e-05,
+      "loss": 0.2164,
+      "step": 9488
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088747531118728e-05,
+      "loss": 0.2237,
+      "step": 9489
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508771865417458e-05,
+      "loss": 0.2254,
+      "step": 9490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5086689704557307e-05,
+      "loss": 0.2429,
+      "step": 9491
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5085660682281606e-05,
+      "loss": 0.2619,
+      "step": 9492
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508463158736218e-05,
+      "loss": 0.2404,
+      "step": 9493
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5083602419813726e-05,
+      "loss": 0.2164,
+      "step": 9494
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082573179650948e-05,
+      "loss": 0.2508,
+      "step": 9495
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5081543866888544e-05,
+      "loss": 0.2432,
+      "step": 9496
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080514481541214e-05,
+      "loss": 0.2235,
+      "step": 9497
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079485023623668e-05,
+      "loss": 0.2452,
+      "step": 9498
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5078455493150608e-05,
+      "loss": 0.225,
+      "step": 9499
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5077425890136739e-05,
+      "loss": 0.2165,
+      "step": 9500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507639621459677e-05,
+      "loss": 0.2449,
+      "step": 9501
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5075366466545412e-05,
+      "loss": 0.2482,
+      "step": 9502
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074336645997366e-05,
+      "loss": 0.2586,
+      "step": 9503
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5073306752967343e-05,
+      "loss": 0.2437,
+      "step": 9504
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072276787470057e-05,
+      "loss": 0.2203,
+      "step": 9505
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5071246749520225e-05,
+      "loss": 0.2365,
+      "step": 9506
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070216639132551e-05,
+      "loss": 0.2465,
+      "step": 9507
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069186456321751e-05,
+      "loss": 0.2198,
+      "step": 9508
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068156201102548e-05,
+      "loss": 0.2313,
+      "step": 9509
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506712587348965e-05,
+      "loss": 0.2401,
+      "step": 9510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066095473497775e-05,
+      "loss": 0.2244,
+      "step": 9511
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065065001141646e-05,
+      "loss": 0.2315,
+      "step": 9512
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064034456435977e-05,
+      "loss": 0.2426,
+      "step": 9513
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063003839395491e-05,
+      "loss": 0.2357,
+      "step": 9514
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061973150034909e-05,
+      "loss": 0.2468,
+      "step": 9515
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5060942388368953e-05,
+      "loss": 0.2336,
+      "step": 9516
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5059911554412346e-05,
+      "loss": 0.2708,
+      "step": 9517
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5058880648179814e-05,
+      "loss": 0.247,
+      "step": 9518
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057849669686082e-05,
+      "loss": 0.2121,
+      "step": 9519
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056818618945876e-05,
+      "loss": 0.2188,
+      "step": 9520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5055787495973924e-05,
+      "loss": 0.2484,
+      "step": 9521
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5054756300784954e-05,
+      "loss": 0.2151,
+      "step": 9522
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053725033393698e-05,
+      "loss": 0.2345,
+      "step": 9523
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5052693693814884e-05,
+      "loss": 0.2382,
+      "step": 9524
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051662282063242e-05,
+      "loss": 0.2382,
+      "step": 9525
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050630798153507e-05,
+      "loss": 0.2423,
+      "step": 9526
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5049599242100415e-05,
+      "loss": 0.2246,
+      "step": 9527
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5048567613918699e-05,
+      "loss": 0.2453,
+      "step": 9528
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5047535913623095e-05,
+      "loss": 0.252,
+      "step": 9529
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5046504141228336e-05,
+      "loss": 0.2641,
+      "step": 9530
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045472296749162e-05,
+      "loss": 0.2653,
+      "step": 9531
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5044440380200316e-05,
+      "loss": 0.239,
+      "step": 9532
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043408391596534e-05,
+      "loss": 0.2204,
+      "step": 9533
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504237633095256e-05,
+      "loss": 0.2367,
+      "step": 9534
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041344198283132e-05,
+      "loss": 0.2425,
+      "step": 9535
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5040311993602994e-05,
+      "loss": 0.2658,
+      "step": 9536
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5039279716926892e-05,
+      "loss": 0.2223,
+      "step": 9537
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503824736826957e-05,
+      "loss": 0.2383,
+      "step": 9538
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037214947645774e-05,
+      "loss": 0.2338,
+      "step": 9539
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503618245507025e-05,
+      "loss": 0.2369,
+      "step": 9540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035149890557752e-05,
+      "loss": 0.2645,
+      "step": 9541
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503411725412302e-05,
+      "loss": 0.2456,
+      "step": 9542
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5033084545780807e-05,
+      "loss": 0.2386,
+      "step": 9543
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503205176554587e-05,
+      "loss": 0.2331,
+      "step": 9544
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031018913432958e-05,
+      "loss": 0.2474,
+      "step": 9545
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5029985989456818e-05,
+      "loss": 0.2452,
+      "step": 9546
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028952993632216e-05,
+      "loss": 0.2391,
+      "step": 9547
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.50279199259739e-05,
+      "loss": 0.233,
+      "step": 9548
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026886786496624e-05,
+      "loss": 0.2633,
+      "step": 9549
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025853575215151e-05,
+      "loss": 0.2282,
+      "step": 9550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024820292144237e-05,
+      "loss": 0.2299,
+      "step": 9551
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023786937298641e-05,
+      "loss": 0.2553,
+      "step": 9552
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022753510693126e-05,
+      "loss": 0.2677,
+      "step": 9553
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021720012342451e-05,
+      "loss": 0.2314,
+      "step": 9554
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020686442261382e-05,
+      "loss": 0.2079,
+      "step": 9555
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5019652800464677e-05,
+      "loss": 0.2661,
+      "step": 9556
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5018619086967103e-05,
+      "loss": 0.2447,
+      "step": 9557
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017585301783428e-05,
+      "loss": 0.2413,
+      "step": 9558
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5016551444928415e-05,
+      "loss": 0.258,
+      "step": 9559
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015517516416833e-05,
+      "loss": 0.2631,
+      "step": 9560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014483516263454e-05,
+      "loss": 0.2415,
+      "step": 9561
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5013449444483043e-05,
+      "loss": 0.209,
+      "step": 9562
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012415301090373e-05,
+      "loss": 0.2227,
+      "step": 9563
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011381086100213e-05,
+      "loss": 0.2168,
+      "step": 9564
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010346799527338e-05,
+      "loss": 0.2598,
+      "step": 9565
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500931244138652e-05,
+      "loss": 0.2254,
+      "step": 9566
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500827801169254e-05,
+      "loss": 0.2383,
+      "step": 9567
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500724351046017e-05,
+      "loss": 0.2593,
+      "step": 9568
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006208937704183e-05,
+      "loss": 0.2422,
+      "step": 9569
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005174293439359e-05,
+      "loss": 0.2535,
+      "step": 9570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004139577680479e-05,
+      "loss": 0.2435,
+      "step": 9571
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5003104790442322e-05,
+      "loss": 0.2376,
+      "step": 9572
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500206993173967e-05,
+      "loss": 0.2304,
+      "step": 9573
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5001035001587302e-05,
+      "loss": 0.2546,
+      "step": 9574
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2584,
+      "step": 9575
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4998964926992554e-05,
+      "loss": 0.2381,
+      "step": 9576
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4997929782579746e-05,
+      "loss": 0.2335,
+      "step": 9577
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4996894566776361e-05,
+      "loss": 0.2364,
+      "step": 9578
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995859279597186e-05,
+      "loss": 0.2538,
+      "step": 9579
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4994823921057009e-05,
+      "loss": 0.2466,
+      "step": 9580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993788491170622e-05,
+      "loss": 0.2009,
+      "step": 9581
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992752989952812e-05,
+      "loss": 0.2633,
+      "step": 9582
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991717417418372e-05,
+      "loss": 0.2447,
+      "step": 9583
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4990681773582094e-05,
+      "loss": 0.2299,
+      "step": 9584
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989646058458767e-05,
+      "loss": 0.2457,
+      "step": 9585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988610272063191e-05,
+      "loss": 0.2647,
+      "step": 9586
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987574414410157e-05,
+      "loss": 0.2412,
+      "step": 9587
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986538485514466e-05,
+      "loss": 0.2342,
+      "step": 9588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498550248539091e-05,
+      "loss": 0.2451,
+      "step": 9589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4984466414054293e-05,
+      "loss": 0.2681,
+      "step": 9590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983430271519406e-05,
+      "loss": 0.2664,
+      "step": 9591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982394057801055e-05,
+      "loss": 0.2473,
+      "step": 9592
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4981357772914042e-05,
+      "loss": 0.2676,
+      "step": 9593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980321416873169e-05,
+      "loss": 0.2392,
+      "step": 9594
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979284989693237e-05,
+      "loss": 0.215,
+      "step": 9595
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497824849138905e-05,
+      "loss": 0.2191,
+      "step": 9596
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4977211921975413e-05,
+      "loss": 0.2745,
+      "step": 9597
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976175281467138e-05,
+      "loss": 0.244,
+      "step": 9598
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975138569879026e-05,
+      "loss": 0.284,
+      "step": 9599
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4974101787225893e-05,
+      "loss": 0.2052,
+      "step": 9600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497306493352254e-05,
+      "loss": 0.2592,
+      "step": 9601
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497202800878378e-05,
+      "loss": 0.2345,
+      "step": 9602
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4970991013024426e-05,
+      "loss": 0.247,
+      "step": 9603
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496995394625929e-05,
+      "loss": 0.2352,
+      "step": 9604
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968916808503184e-05,
+      "loss": 0.246,
+      "step": 9605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967879599770927e-05,
+      "loss": 0.2408,
+      "step": 9606
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496684232007733e-05,
+      "loss": 0.2315,
+      "step": 9607
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496580496943721e-05,
+      "loss": 0.2415,
+      "step": 9608
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964767547865387e-05,
+      "loss": 0.2592,
+      "step": 9609
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4963730055376675e-05,
+      "loss": 0.2456,
+      "step": 9610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962692491985899e-05,
+      "loss": 0.2402,
+      "step": 9611
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961654857707877e-05,
+      "loss": 0.216,
+      "step": 9612
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4960617152557429e-05,
+      "loss": 0.2318,
+      "step": 9613
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959579376549382e-05,
+      "loss": 0.2419,
+      "step": 9614
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4958541529698555e-05,
+      "loss": 0.2514,
+      "step": 9615
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957503612019775e-05,
+      "loss": 0.2307,
+      "step": 9616
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4956465623527869e-05,
+      "loss": 0.2528,
+      "step": 9617
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955427564237658e-05,
+      "loss": 0.2453,
+      "step": 9618
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4954389434163976e-05,
+      "loss": 0.2464,
+      "step": 9619
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4953351233321652e-05,
+      "loss": 0.2322,
+      "step": 9620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.495231296172551e-05,
+      "loss": 0.235,
+      "step": 9621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4951274619390384e-05,
+      "loss": 0.2836,
+      "step": 9622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4950236206331105e-05,
+      "loss": 0.2384,
+      "step": 9623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4949197722562507e-05,
+      "loss": 0.2548,
+      "step": 9624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948159168099421e-05,
+      "loss": 0.2407,
+      "step": 9625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4947120542956684e-05,
+      "loss": 0.2626,
+      "step": 9626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946081847149134e-05,
+      "loss": 0.2661,
+      "step": 9627
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4945043080691605e-05,
+      "loss": 0.2483,
+      "step": 9628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944004243598933e-05,
+      "loss": 0.2565,
+      "step": 9629
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494296533588596e-05,
+      "loss": 0.2491,
+      "step": 9630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941926357567523e-05,
+      "loss": 0.2536,
+      "step": 9631
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940887308658463e-05,
+      "loss": 0.242,
+      "step": 9632
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4939848189173625e-05,
+      "loss": 0.2432,
+      "step": 9633
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938808999127852e-05,
+      "loss": 0.2261,
+      "step": 9634
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937769738535982e-05,
+      "loss": 0.247,
+      "step": 9635
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936730407412862e-05,
+      "loss": 0.2436,
+      "step": 9636
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935691005773343e-05,
+      "loss": 0.2169,
+      "step": 9637
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934651533632267e-05,
+      "loss": 0.2209,
+      "step": 9638
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4933611991004485e-05,
+      "loss": 0.2523,
+      "step": 9639
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4932572377904842e-05,
+      "loss": 0.2656,
+      "step": 9640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493153269434819e-05,
+      "loss": 0.2715,
+      "step": 9641
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493049294034938e-05,
+      "loss": 0.2402,
+      "step": 9642
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492945311592326e-05,
+      "loss": 0.2573,
+      "step": 9643
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928413221084693e-05,
+      "loss": 0.2311,
+      "step": 9644
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492737325584852e-05,
+      "loss": 0.2623,
+      "step": 9645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926333220229605e-05,
+      "loss": 0.2447,
+      "step": 9646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925293114242803e-05,
+      "loss": 0.2372,
+      "step": 9647
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4924252937902966e-05,
+      "loss": 0.2539,
+      "step": 9648
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923212691224956e-05,
+      "loss": 0.2424,
+      "step": 9649
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492217237422363e-05,
+      "loss": 0.2264,
+      "step": 9650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492113198691385e-05,
+      "loss": 0.2482,
+      "step": 9651
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920091529310474e-05,
+      "loss": 0.2386,
+      "step": 9652
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919051001428369e-05,
+      "loss": 0.2399,
+      "step": 9653
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491801040328239e-05,
+      "loss": 0.2379,
+      "step": 9654
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4916969734887406e-05,
+      "loss": 0.2312,
+      "step": 9655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915928996258284e-05,
+      "loss": 0.2741,
+      "step": 9656
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914888187409886e-05,
+      "loss": 0.2348,
+      "step": 9657
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913847308357079e-05,
+      "loss": 0.2338,
+      "step": 9658
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912806359114733e-05,
+      "loss": 0.2261,
+      "step": 9659
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491176533969772e-05,
+      "loss": 0.2755,
+      "step": 9660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4910724250120903e-05,
+      "loss": 0.2551,
+      "step": 9661
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909683090399157e-05,
+      "loss": 0.2571,
+      "step": 9662
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490864186054735e-05,
+      "loss": 0.2562,
+      "step": 9663
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490760056058036e-05,
+      "loss": 0.2669,
+      "step": 9664
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4906559190513056e-05,
+      "loss": 0.2245,
+      "step": 9665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905517750360321e-05,
+      "loss": 0.2127,
+      "step": 9666
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4904476240137024e-05,
+      "loss": 0.2438,
+      "step": 9667
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4903434659858043e-05,
+      "loss": 0.2507,
+      "step": 9668
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4902393009538256e-05,
+      "loss": 0.2447,
+      "step": 9669
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4901351289192544e-05,
+      "loss": 0.2448,
+      "step": 9670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900309498835785e-05,
+      "loss": 0.2393,
+      "step": 9671
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489926763848286e-05,
+      "loss": 0.2386,
+      "step": 9672
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4898225708148657e-05,
+      "loss": 0.2685,
+      "step": 9673
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897183707848048e-05,
+      "loss": 0.2355,
+      "step": 9674
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896141637595925e-05,
+      "loss": 0.2181,
+      "step": 9675
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4895099497407169e-05,
+      "loss": 0.2251,
+      "step": 9676
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4894057287296667e-05,
+      "loss": 0.264,
+      "step": 9677
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893015007279309e-05,
+      "loss": 0.2458,
+      "step": 9678
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891972657369977e-05,
+      "loss": 0.2394,
+      "step": 9679
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890930237583567e-05,
+      "loss": 0.257,
+      "step": 9680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889887747934964e-05,
+      "loss": 0.2431,
+      "step": 9681
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488884518843906e-05,
+      "loss": 0.272,
+      "step": 9682
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4887802559110747e-05,
+      "loss": 0.2475,
+      "step": 9683
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886759859964917e-05,
+      "loss": 0.2397,
+      "step": 9684
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885717091016467e-05,
+      "loss": 0.2431,
+      "step": 9685
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884674252280288e-05,
+      "loss": 0.2495,
+      "step": 9686
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488363134377128e-05,
+      "loss": 0.2452,
+      "step": 9687
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882588365504338e-05,
+      "loss": 0.2264,
+      "step": 9688
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4881545317494359e-05,
+      "loss": 0.25,
+      "step": 9689
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4880502199756244e-05,
+      "loss": 0.2597,
+      "step": 9690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879459012304888e-05,
+      "loss": 0.207,
+      "step": 9691
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4878415755155198e-05,
+      "loss": 0.2626,
+      "step": 9692
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4877372428322077e-05,
+      "loss": 0.2532,
+      "step": 9693
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487632903182042e-05,
+      "loss": 0.2533,
+      "step": 9694
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4875285565665138e-05,
+      "loss": 0.2216,
+      "step": 9695
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487424202987113e-05,
+      "loss": 0.2371,
+      "step": 9696
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873198424453309e-05,
+      "loss": 0.2326,
+      "step": 9697
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872154749426575e-05,
+      "loss": 0.2445,
+      "step": 9698
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487111100480584e-05,
+      "loss": 0.2483,
+      "step": 9699
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4870067190606015e-05,
+      "loss": 0.2168,
+      "step": 9700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869023306842002e-05,
+      "loss": 0.2467,
+      "step": 9701
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486797935352872e-05,
+      "loss": 0.224,
+      "step": 9702
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866935330681074e-05,
+      "loss": 0.2246,
+      "step": 9703
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4865891238313982e-05,
+      "loss": 0.2314,
+      "step": 9704
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4864847076442358e-05,
+      "loss": 0.2606,
+      "step": 9705
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4863802845081113e-05,
+      "loss": 0.2426,
+      "step": 9706
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4862758544245163e-05,
+      "loss": 0.2207,
+      "step": 9707
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4861714173949429e-05,
+      "loss": 0.2442,
+      "step": 9708
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4860669734208823e-05,
+      "loss": 0.2487,
+      "step": 9709
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.485962522503827e-05,
+      "loss": 0.2335,
+      "step": 9710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485858064645269e-05,
+      "loss": 0.286,
+      "step": 9711
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4857535998466997e-05,
+      "loss": 0.2565,
+      "step": 9712
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485649128109612e-05,
+      "loss": 0.2531,
+      "step": 9713
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855446494354975e-05,
+      "loss": 0.2629,
+      "step": 9714
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854401638258488e-05,
+      "loss": 0.2564,
+      "step": 9715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485335671282159e-05,
+      "loss": 0.2467,
+      "step": 9716
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4852311718059196e-05,
+      "loss": 0.2404,
+      "step": 9717
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485126665398624e-05,
+      "loss": 0.236,
+      "step": 9718
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850221520617652e-05,
+      "loss": 0.2327,
+      "step": 9719
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849176317968353e-05,
+      "loss": 0.2488,
+      "step": 9720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4848131046053275e-05,
+      "loss": 0.2374,
+      "step": 9721
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847085704887355e-05,
+      "loss": 0.2645,
+      "step": 9722
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846040294485518e-05,
+      "loss": 0.2439,
+      "step": 9723
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844994814862696e-05,
+      "loss": 0.2463,
+      "step": 9724
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4843949266033827e-05,
+      "loss": 0.2544,
+      "step": 9725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4842903648013847e-05,
+      "loss": 0.2351,
+      "step": 9726
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841857960817685e-05,
+      "loss": 0.2669,
+      "step": 9727
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840812204460284e-05,
+      "loss": 0.2388,
+      "step": 9728
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839766378956574e-05,
+      "loss": 0.2074,
+      "step": 9729
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838720484321503e-05,
+      "loss": 0.2257,
+      "step": 9730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837674520570003e-05,
+      "loss": 0.2348,
+      "step": 9731
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4836628487717022e-05,
+      "loss": 0.2411,
+      "step": 9732
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4835582385777493e-05,
+      "loss": 0.2825,
+      "step": 9733
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4834536214766364e-05,
+      "loss": 0.2544,
+      "step": 9734
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833489974698576e-05,
+      "loss": 0.2508,
+      "step": 9735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832443665589074e-05,
+      "loss": 0.2609,
+      "step": 9736
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4831397287452806e-05,
+      "loss": 0.2525,
+      "step": 9737
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830350840304717e-05,
+      "loss": 0.2148,
+      "step": 9738
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829304324159756e-05,
+      "loss": 0.2209,
+      "step": 9739
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4828257739032866e-05,
+      "loss": 0.2465,
+      "step": 9740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827211084939001e-05,
+      "loss": 0.2376,
+      "step": 9741
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4826164361893111e-05,
+      "loss": 0.238,
+      "step": 9742
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4825117569910146e-05,
+      "loss": 0.2518,
+      "step": 9743
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4824070709005063e-05,
+      "loss": 0.2574,
+      "step": 9744
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4823023779192806e-05,
+      "loss": 0.2179,
+      "step": 9745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482197678048834e-05,
+      "loss": 0.2501,
+      "step": 9746
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820929712906615e-05,
+      "loss": 0.2573,
+      "step": 9747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819882576462585e-05,
+      "loss": 0.2397,
+      "step": 9748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4818835371171213e-05,
+      "loss": 0.2581,
+      "step": 9749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4817788097047454e-05,
+      "loss": 0.2689,
+      "step": 9750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4816740754106266e-05,
+      "loss": 0.2556,
+      "step": 9751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4815693342362613e-05,
+      "loss": 0.2453,
+      "step": 9752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814645861831453e-05,
+      "loss": 0.2443,
+      "step": 9753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481359831252775e-05,
+      "loss": 0.2533,
+      "step": 9754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481255069446647e-05,
+      "loss": 0.2639,
+      "step": 9755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481150300766257e-05,
+      "loss": 0.2614,
+      "step": 9756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810455252131019e-05,
+      "loss": 0.231,
+      "step": 9757
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809407427886786e-05,
+      "loss": 0.2407,
+      "step": 9758
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808359534944833e-05,
+      "loss": 0.2374,
+      "step": 9759
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4807311573320134e-05,
+      "loss": 0.2448,
+      "step": 9760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806263543027652e-05,
+      "loss": 0.2174,
+      "step": 9761
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805215444082362e-05,
+      "loss": 0.2517,
+      "step": 9762
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4804167276499231e-05,
+      "loss": 0.2597,
+      "step": 9763
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803119040293233e-05,
+      "loss": 0.2519,
+      "step": 9764
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4802070735479341e-05,
+      "loss": 0.2408,
+      "step": 9765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4801022362072532e-05,
+      "loss": 0.2334,
+      "step": 9766
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799973920087775e-05,
+      "loss": 0.23,
+      "step": 9767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4798925409540049e-05,
+      "loss": 0.2364,
+      "step": 9768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797876830444331e-05,
+      "loss": 0.2457,
+      "step": 9769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4796828182815599e-05,
+      "loss": 0.2312,
+      "step": 9770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4795779466668836e-05,
+      "loss": 0.242,
+      "step": 9771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4794730682019014e-05,
+      "loss": 0.2288,
+      "step": 9772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4793681828881118e-05,
+      "loss": 0.2383,
+      "step": 9773
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792632907270128e-05,
+      "loss": 0.2338,
+      "step": 9774
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791583917201029e-05,
+      "loss": 0.2646,
+      "step": 9775
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790534858688803e-05,
+      "loss": 0.2937,
+      "step": 9776
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4789485731748441e-05,
+      "loss": 0.2308,
+      "step": 9777
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4788436536394919e-05,
+      "loss": 0.2491,
+      "step": 9778
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478738727264323e-05,
+      "loss": 0.2544,
+      "step": 9779
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478633794050836e-05,
+      "loss": 0.226,
+      "step": 9780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785288540005296e-05,
+      "loss": 0.2485,
+      "step": 9781
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784239071149031e-05,
+      "loss": 0.2472,
+      "step": 9782
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783189533954555e-05,
+      "loss": 0.2295,
+      "step": 9783
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4782139928436858e-05,
+      "loss": 0.2562,
+      "step": 9784
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781090254610932e-05,
+      "loss": 0.2717,
+      "step": 9785
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780040512491774e-05,
+      "loss": 0.2289,
+      "step": 9786
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4778990702094373e-05,
+      "loss": 0.2628,
+      "step": 9787
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4777940823433732e-05,
+      "loss": 0.2381,
+      "step": 9788
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477689087652484e-05,
+      "loss": 0.2192,
+      "step": 9789
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.47758408613827e-05,
+      "loss": 0.2448,
+      "step": 9790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774790778022307e-05,
+      "loss": 0.2356,
+      "step": 9791
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4773740626458666e-05,
+      "loss": 0.2278,
+      "step": 9792
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4772690406706769e-05,
+      "loss": 0.2106,
+      "step": 9793
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771640118781623e-05,
+      "loss": 0.2402,
+      "step": 9794
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477058976269823e-05,
+      "loss": 0.2367,
+      "step": 9795
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4769539338471587e-05,
+      "loss": 0.2277,
+      "step": 9796
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768488846116706e-05,
+      "loss": 0.2486,
+      "step": 9797
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4767438285648593e-05,
+      "loss": 0.2297,
+      "step": 9798
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766387657082249e-05,
+      "loss": 0.2475,
+      "step": 9799
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4765336960432684e-05,
+      "loss": 0.2509,
+      "step": 9800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764286195714903e-05,
+      "loss": 0.217,
+      "step": 9801
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4763235362943918e-05,
+      "loss": 0.2718,
+      "step": 9802
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762184462134739e-05,
+      "loss": 0.2712,
+      "step": 9803
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476113349330238e-05,
+      "loss": 0.2699,
+      "step": 9804
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760082456461846e-05,
+      "loss": 0.223,
+      "step": 9805
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759031351628154e-05,
+      "loss": 0.2312,
+      "step": 9806
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475798017881632e-05,
+      "loss": 0.2391,
+      "step": 9807
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756928938041355e-05,
+      "loss": 0.2619,
+      "step": 9808
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4755877629318276e-05,
+      "loss": 0.2324,
+      "step": 9809
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754826252662104e-05,
+      "loss": 0.2552,
+      "step": 9810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753774808087853e-05,
+      "loss": 0.2408,
+      "step": 9811
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4752723295610541e-05,
+      "loss": 0.2556,
+      "step": 9812
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475167171524519e-05,
+      "loss": 0.2368,
+      "step": 9813
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475062006700682e-05,
+      "loss": 0.2369,
+      "step": 9814
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4749568350910452e-05,
+      "loss": 0.2337,
+      "step": 9815
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748516566971112e-05,
+      "loss": 0.2209,
+      "step": 9816
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747464715203818e-05,
+      "loss": 0.2385,
+      "step": 9817
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4746412795623602e-05,
+      "loss": 0.2254,
+      "step": 9818
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4745360808245484e-05,
+      "loss": 0.2795,
+      "step": 9819
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474430875308449e-05,
+      "loss": 0.2667,
+      "step": 9820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743256630155656e-05,
+      "loss": 0.2393,
+      "step": 9821
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742204439473999e-05,
+      "loss": 0.2602,
+      "step": 9822
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741152181054554e-05,
+      "loss": 0.2632,
+      "step": 9823
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4740099854912353e-05,
+      "loss": 0.2306,
+      "step": 9824
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739047461062428e-05,
+      "loss": 0.24,
+      "step": 9825
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4737994999519807e-05,
+      "loss": 0.248,
+      "step": 9826
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736942470299527e-05,
+      "loss": 0.2509,
+      "step": 9827
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735889873416618e-05,
+      "loss": 0.2288,
+      "step": 9828
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4734837208886122e-05,
+      "loss": 0.2199,
+      "step": 9829
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4733784476723069e-05,
+      "loss": 0.2139,
+      "step": 9830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732731676942503e-05,
+      "loss": 0.2308,
+      "step": 9831
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731678809559454e-05,
+      "loss": 0.2378,
+      "step": 9832
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4730625874588968e-05,
+      "loss": 0.2537,
+      "step": 9833
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729572872046081e-05,
+      "loss": 0.2546,
+      "step": 9834
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728519801945836e-05,
+      "loss": 0.2611,
+      "step": 9835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727466664303274e-05,
+      "loss": 0.2587,
+      "step": 9836
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4726413459133442e-05,
+      "loss": 0.2441,
+      "step": 9837
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725360186451378e-05,
+      "loss": 0.2242,
+      "step": 9838
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4724306846272132e-05,
+      "loss": 0.259,
+      "step": 9839
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4723253438610745e-05,
+      "loss": 0.2683,
+      "step": 9840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722199963482266e-05,
+      "loss": 0.2308,
+      "step": 9841
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721146420901747e-05,
+      "loss": 0.226,
+      "step": 9842
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4720092810884232e-05,
+      "loss": 0.2535,
+      "step": 9843
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719039133444773e-05,
+      "loss": 0.2492,
+      "step": 9844
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4717985388598419e-05,
+      "loss": 0.2325,
+      "step": 9845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716931576360223e-05,
+      "loss": 0.2829,
+      "step": 9846
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715877696745236e-05,
+      "loss": 0.2362,
+      "step": 9847
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714823749768513e-05,
+      "loss": 0.2473,
+      "step": 9848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471376973544511e-05,
+      "loss": 0.2618,
+      "step": 9849
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471271565379008e-05,
+      "loss": 0.2759,
+      "step": 9850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4711661504818482e-05,
+      "loss": 0.2505,
+      "step": 9851
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471060728854537e-05,
+      "loss": 0.236,
+      "step": 9852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4709553004985806e-05,
+      "loss": 0.2544,
+      "step": 9853
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470849865415485e-05,
+      "loss": 0.2369,
+      "step": 9854
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707444236067558e-05,
+      "loss": 0.2327,
+      "step": 9855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706389750738995e-05,
+      "loss": 0.2096,
+      "step": 9856
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4705335198184219e-05,
+      "loss": 0.2506,
+      "step": 9857
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47042805784183e-05,
+      "loss": 0.2326,
+      "step": 9858
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703225891456297e-05,
+      "loss": 0.2576,
+      "step": 9859
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4702171137313279e-05,
+      "loss": 0.2424,
+      "step": 9860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701116316004307e-05,
+      "loss": 0.2564,
+      "step": 9861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4700061427544449e-05,
+      "loss": 0.2407,
+      "step": 9862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4699006471948778e-05,
+      "loss": 0.2349,
+      "step": 9863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697951449232362e-05,
+      "loss": 0.2427,
+      "step": 9864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4696896359410266e-05,
+      "loss": 0.2124,
+      "step": 9865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695841202497568e-05,
+      "loss": 0.2521,
+      "step": 9866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469478597850933e-05,
+      "loss": 0.2379,
+      "step": 9867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4693730687460632e-05,
+      "loss": 0.2643,
+      "step": 9868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692675329366548e-05,
+      "loss": 0.2457,
+      "step": 9869
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691619904242151e-05,
+      "loss": 0.2291,
+      "step": 9870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469056441210252e-05,
+      "loss": 0.2433,
+      "step": 9871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689508852962726e-05,
+      "loss": 0.2543,
+      "step": 9872
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688453226837849e-05,
+      "loss": 0.2648,
+      "step": 9873
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687397533742966e-05,
+      "loss": 0.2378,
+      "step": 9874
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468634177369316e-05,
+      "loss": 0.2239,
+      "step": 9875
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685285946703513e-05,
+      "loss": 0.2348,
+      "step": 9876
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4684230052789102e-05,
+      "loss": 0.2559,
+      "step": 9877
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683174091965009e-05,
+      "loss": 0.2127,
+      "step": 9878
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4682118064246318e-05,
+      "loss": 0.2255,
+      "step": 9879
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4681061969648118e-05,
+      "loss": 0.2436,
+      "step": 9880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680005808185489e-05,
+      "loss": 0.2625,
+      "step": 9881
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4678949579873521e-05,
+      "loss": 0.2427,
+      "step": 9882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677893284727296e-05,
+      "loss": 0.2586,
+      "step": 9883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676836922761908e-05,
+      "loss": 0.2597,
+      "step": 9884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4675780493992442e-05,
+      "loss": 0.2248,
+      "step": 9885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674723998433988e-05,
+      "loss": 0.2296,
+      "step": 9886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673667436101641e-05,
+      "loss": 0.2268,
+      "step": 9887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4672610807010493e-05,
+      "loss": 0.2661,
+      "step": 9888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4671554111175632e-05,
+      "loss": 0.231,
+      "step": 9889
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4670497348612154e-05,
+      "loss": 0.2206,
+      "step": 9890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4669440519335155e-05,
+      "loss": 0.2189,
+      "step": 9891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466838362335973e-05,
+      "loss": 0.219,
+      "step": 9892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4667326660700976e-05,
+      "loss": 0.2289,
+      "step": 9893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666269631373989e-05,
+      "loss": 0.2405,
+      "step": 9894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4665212535393869e-05,
+      "loss": 0.2639,
+      "step": 9895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664155372775715e-05,
+      "loss": 0.2193,
+      "step": 9896
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466309814353463e-05,
+      "loss": 0.2397,
+      "step": 9897
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662040847685712e-05,
+      "loss": 0.2471,
+      "step": 9898
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660983485244068e-05,
+      "loss": 0.2457,
+      "step": 9899
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4659926056224798e-05,
+      "loss": 0.2305,
+      "step": 9900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658868560643002e-05,
+      "loss": 0.2337,
+      "step": 9901
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4657810998513795e-05,
+      "loss": 0.2727,
+      "step": 9902
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656753369852278e-05,
+      "loss": 0.2219,
+      "step": 9903
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655695674673558e-05,
+      "loss": 0.2104,
+      "step": 9904
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654637912992744e-05,
+      "loss": 0.237,
+      "step": 9905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653580084824946e-05,
+      "loss": 0.2425,
+      "step": 9906
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652522190185269e-05,
+      "loss": 0.2857,
+      "step": 9907
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465146422908883e-05,
+      "loss": 0.2304,
+      "step": 9908
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465040620155074e-05,
+      "loss": 0.2274,
+      "step": 9909
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464934810758611e-05,
+      "loss": 0.227,
+      "step": 9910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4648289947210055e-05,
+      "loss": 0.2501,
+      "step": 9911
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.245,
+      "step": 9912
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646173427284125e-05,
+      "loss": 0.216,
+      "step": 9913
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645115067764487e-05,
+      "loss": 0.2457,
+      "step": 9914
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4644056641893887e-05,
+      "loss": 0.2339,
+      "step": 9915
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642998149687444e-05,
+      "loss": 0.2502,
+      "step": 9916
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641939591160282e-05,
+      "loss": 0.2424,
+      "step": 9917
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4640880966327515e-05,
+      "loss": 0.2513,
+      "step": 9918
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639822275204268e-05,
+      "loss": 0.2569,
+      "step": 9919
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463876351780566e-05,
+      "loss": 0.2514,
+      "step": 9920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4637704694146822e-05,
+      "loss": 0.2381,
+      "step": 9921
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463664580424287e-05,
+      "loss": 0.2423,
+      "step": 9922
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4635586848108938e-05,
+      "loss": 0.2323,
+      "step": 9923
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463452782576014e-05,
+      "loss": 0.2452,
+      "step": 9924
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633468737211613e-05,
+      "loss": 0.239,
+      "step": 9925
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4632409582478485e-05,
+      "loss": 0.2426,
+      "step": 9926
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631350361575878e-05,
+      "loss": 0.2344,
+      "step": 9927
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630291074518925e-05,
+      "loss": 0.2642,
+      "step": 9928
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462923172132276e-05,
+      "loss": 0.248,
+      "step": 9929
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4628172302002515e-05,
+      "loss": 0.2377,
+      "step": 9930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627112816573316e-05,
+      "loss": 0.2177,
+      "step": 9931
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626053265050304e-05,
+      "loss": 0.2397,
+      "step": 9932
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4624993647448609e-05,
+      "loss": 0.2373,
+      "step": 9933
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462393396378337e-05,
+      "loss": 0.2498,
+      "step": 9934
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462287421406972e-05,
+      "loss": 0.2685,
+      "step": 9935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621814398322798e-05,
+      "loss": 0.2479,
+      "step": 9936
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4620754516557747e-05,
+      "loss": 0.2433,
+      "step": 9937
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46196945687897e-05,
+      "loss": 0.2501,
+      "step": 9938
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46186345550338e-05,
+      "loss": 0.2321,
+      "step": 9939
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617574475305185e-05,
+      "loss": 0.2585,
+      "step": 9940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616514329619002e-05,
+      "loss": 0.2571,
+      "step": 9941
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615454117990394e-05,
+      "loss": 0.2562,
+      "step": 9942
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4614393840434502e-05,
+      "loss": 0.2454,
+      "step": 9943
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4613333496966473e-05,
+      "loss": 0.2266,
+      "step": 9944
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461227308760145e-05,
+      "loss": 0.2433,
+      "step": 9945
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4611212612354584e-05,
+      "loss": 0.227,
+      "step": 9946
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4610152071241019e-05,
+      "loss": 0.2476,
+      "step": 9947
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460909146427591e-05,
+      "loss": 0.2539,
+      "step": 9948
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46080307914744e-05,
+      "loss": 0.2279,
+      "step": 9949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4606970052851643e-05,
+      "loss": 0.2361,
+      "step": 9950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460590924842279e-05,
+      "loss": 0.212,
+      "step": 9951
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4604848378202992e-05,
+      "loss": 0.2606,
+      "step": 9952
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603787442207406e-05,
+      "loss": 0.2454,
+      "step": 9953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602726440451184e-05,
+      "loss": 0.2336,
+      "step": 9954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601665372949483e-05,
+      "loss": 0.2536,
+      "step": 9955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4600604239717457e-05,
+      "loss": 0.2593,
+      "step": 9956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599543040770263e-05,
+      "loss": 0.2544,
+      "step": 9957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598481776123061e-05,
+      "loss": 0.2517,
+      "step": 9958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597420445791012e-05,
+      "loss": 0.2435,
+      "step": 9959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4596359049789273e-05,
+      "loss": 0.2623,
+      "step": 9960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4595297588133004e-05,
+      "loss": 0.223,
+      "step": 9961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459423606083737e-05,
+      "loss": 0.2274,
+      "step": 9962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4593174467917534e-05,
+      "loss": 0.2447,
+      "step": 9963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592112809388657e-05,
+      "loss": 0.2269,
+      "step": 9964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591051085265907e-05,
+      "loss": 0.2385,
+      "step": 9965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589989295564447e-05,
+      "loss": 0.2359,
+      "step": 9966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4588927440299441e-05,
+      "loss": 0.2231,
+      "step": 9967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587865519486064e-05,
+      "loss": 0.229,
+      "step": 9968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586803533139482e-05,
+      "loss": 0.2356,
+      "step": 9969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458574148127486e-05,
+      "loss": 0.266,
+      "step": 9970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4584679363907372e-05,
+      "loss": 0.2457,
+      "step": 9971
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458361718105219e-05,
+      "loss": 0.2381,
+      "step": 9972
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4582554932724485e-05,
+      "loss": 0.2239,
+      "step": 9973
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581492618939428e-05,
+      "loss": 0.2447,
+      "step": 9974
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580430239712198e-05,
+      "loss": 0.2367,
+      "step": 9975
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4579367795057967e-05,
+      "loss": 0.2391,
+      "step": 9976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578305284991912e-05,
+      "loss": 0.2531,
+      "step": 9977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457724270952921e-05,
+      "loss": 0.2306,
+      "step": 9978
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4576180068685032e-05,
+      "loss": 0.2221,
+      "step": 9979
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575117362474569e-05,
+      "loss": 0.2496,
+      "step": 9980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4574054590912994e-05,
+      "loss": 0.2388,
+      "step": 9981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572991754015488e-05,
+      "loss": 0.2289,
+      "step": 9982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571928851797235e-05,
+      "loss": 0.237,
+      "step": 9983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.457086588427341e-05,
+      "loss": 0.2581,
+      "step": 9984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4569802851459207e-05,
+      "loss": 0.2472,
+      "step": 9985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568739753369802e-05,
+      "loss": 0.2597,
+      "step": 9986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567676590020388e-05,
+      "loss": 0.2288,
+      "step": 9987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566613361426142e-05,
+      "loss": 0.2424,
+      "step": 9988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4565550067602262e-05,
+      "loss": 0.249,
+      "step": 9989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4564486708563924e-05,
+      "loss": 0.2337,
+      "step": 9990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563423284326326e-05,
+      "loss": 0.2443,
+      "step": 9991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562359794904655e-05,
+      "loss": 0.2326,
+      "step": 9992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4561296240314102e-05,
+      "loss": 0.2326,
+      "step": 9993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456023262056986e-05,
+      "loss": 0.2533,
+      "step": 9994
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455916893568712e-05,
+      "loss": 0.2309,
+      "step": 9995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4558105185681078e-05,
+      "loss": 0.2249,
+      "step": 9996
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557041370566925e-05,
+      "loss": 0.2594,
+      "step": 9997
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455597749035986e-05,
+      "loss": 0.2361,
+      "step": 9998
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4554913545075077e-05,
+      "loss": 0.236,
+      "step": 9999
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553849534727778e-05,
+      "loss": 0.248,
+      "step": 10000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552785459333155e-05,
+      "loss": 0.2406,
+      "step": 10001
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4551721318906411e-05,
+      "loss": 0.2673,
+      "step": 10002
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550657113462745e-05,
+      "loss": 0.2542,
+      "step": 10003
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4549592843017359e-05,
+      "loss": 0.2485,
+      "step": 10004
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4548528507585456e-05,
+      "loss": 0.2226,
+      "step": 10005
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547464107182234e-05,
+      "loss": 0.2427,
+      "step": 10006
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546399641822904e-05,
+      "loss": 0.251,
+      "step": 10007
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4545335111522665e-05,
+      "loss": 0.1976,
+      "step": 10008
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544270516296731e-05,
+      "loss": 0.2263,
+      "step": 10009
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4543205856160298e-05,
+      "loss": 0.2694,
+      "step": 10010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542141131128579e-05,
+      "loss": 0.2511,
+      "step": 10011
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4541076341216782e-05,
+      "loss": 0.2292,
+      "step": 10012
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4540011486440118e-05,
+      "loss": 0.2495,
+      "step": 10013
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538946566813796e-05,
+      "loss": 0.2211,
+      "step": 10014
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4537881582353027e-05,
+      "loss": 0.2356,
+      "step": 10015
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536816533073024e-05,
+      "loss": 0.2395,
+      "step": 10016
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4535751418989e-05,
+      "loss": 0.2472,
+      "step": 10017
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4534686240116166e-05,
+      "loss": 0.2319,
+      "step": 10018
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533620996469745e-05,
+      "loss": 0.2459,
+      "step": 10019
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4532555688064947e-05,
+      "loss": 0.2295,
+      "step": 10020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4531490314916988e-05,
+      "loss": 0.2309,
+      "step": 10021
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453042487704109e-05,
+      "loss": 0.2381,
+      "step": 10022
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4529359374452466e-05,
+      "loss": 0.2474,
+      "step": 10023
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452829380716634e-05,
+      "loss": 0.234,
+      "step": 10024
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4527228175197934e-05,
+      "loss": 0.2484,
+      "step": 10025
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4526162478562466e-05,
+      "loss": 0.246,
+      "step": 10026
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452509671727516e-05,
+      "loss": 0.2293,
+      "step": 10027
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524030891351238e-05,
+      "loss": 0.2208,
+      "step": 10028
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522965000805929e-05,
+      "loss": 0.2727,
+      "step": 10029
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452189904565445e-05,
+      "loss": 0.2293,
+      "step": 10030
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520833025912036e-05,
+      "loss": 0.2301,
+      "step": 10031
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519766941593906e-05,
+      "loss": 0.2501,
+      "step": 10032
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4518700792715295e-05,
+      "loss": 0.2308,
+      "step": 10033
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4517634579291424e-05,
+      "loss": 0.2372,
+      "step": 10034
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516568301337533e-05,
+      "loss": 0.2554,
+      "step": 10035
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515501958868845e-05,
+      "loss": 0.2245,
+      "step": 10036
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514435551900594e-05,
+      "loss": 0.2308,
+      "step": 10037
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513369080448012e-05,
+      "loss": 0.2587,
+      "step": 10038
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512302544526333e-05,
+      "loss": 0.2089,
+      "step": 10039
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451123594415079e-05,
+      "loss": 0.2492,
+      "step": 10040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510169279336622e-05,
+      "loss": 0.213,
+      "step": 10041
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509102550099063e-05,
+      "loss": 0.2618,
+      "step": 10042
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4508035756453351e-05,
+      "loss": 0.2316,
+      "step": 10043
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4506968898414724e-05,
+      "loss": 0.2489,
+      "step": 10044
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505901975998418e-05,
+      "loss": 0.2239,
+      "step": 10045
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4504834989219674e-05,
+      "loss": 0.2291,
+      "step": 10046
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503767938093737e-05,
+      "loss": 0.2351,
+      "step": 10047
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4502700822635846e-05,
+      "loss": 0.2437,
+      "step": 10048
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4501633642861243e-05,
+      "loss": 0.2585,
+      "step": 10049
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4500566398785172e-05,
+      "loss": 0.2206,
+      "step": 10050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449949909042288e-05,
+      "loss": 0.2326,
+      "step": 10051
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4498431717789606e-05,
+      "loss": 0.2485,
+      "step": 10052
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497364280900606e-05,
+      "loss": 0.2254,
+      "step": 10053
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496296779771117e-05,
+      "loss": 0.2593,
+      "step": 10054
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4495229214416395e-05,
+      "loss": 0.2417,
+      "step": 10055
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4494161584851687e-05,
+      "loss": 0.249,
+      "step": 10056
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449309389109224e-05,
+      "loss": 0.2455,
+      "step": 10057
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4492026133153307e-05,
+      "loss": 0.2215,
+      "step": 10058
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490958311050145e-05,
+      "loss": 0.2534,
+      "step": 10059
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4489890424797998e-05,
+      "loss": 0.2487,
+      "step": 10060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488822474412125e-05,
+      "loss": 0.2251,
+      "step": 10061
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4487754459907782e-05,
+      "loss": 0.2472,
+      "step": 10062
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448668638130022e-05,
+      "loss": 0.2373,
+      "step": 10063
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485618238604695e-05,
+      "loss": 0.2399,
+      "step": 10064
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4484550031836469e-05,
+      "loss": 0.2497,
+      "step": 10065
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4483481761010798e-05,
+      "loss": 0.2415,
+      "step": 10066
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4482413426142941e-05,
+      "loss": 0.2524,
+      "step": 10067
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4481345027248162e-05,
+      "loss": 0.2822,
+      "step": 10068
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4480276564341714e-05,
+      "loss": 0.2491,
+      "step": 10069
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479208037438865e-05,
+      "loss": 0.2492,
+      "step": 10070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4478139446554875e-05,
+      "loss": 0.2567,
+      "step": 10071
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447707079170501e-05,
+      "loss": 0.2436,
+      "step": 10072
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4476002072904533e-05,
+      "loss": 0.2431,
+      "step": 10073
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474933290168713e-05,
+      "loss": 0.275,
+      "step": 10074
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473864443512813e-05,
+      "loss": 0.2382,
+      "step": 10075
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472795532952102e-05,
+      "loss": 0.223,
+      "step": 10076
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471726558501847e-05,
+      "loss": 0.225,
+      "step": 10077
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4470657520177318e-05,
+      "loss": 0.2457,
+      "step": 10078
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469588417993783e-05,
+      "loss": 0.2444,
+      "step": 10079
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4468519251966517e-05,
+      "loss": 0.2805,
+      "step": 10080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467450022110794e-05,
+      "loss": 0.2628,
+      "step": 10081
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446638072844188e-05,
+      "loss": 0.251,
+      "step": 10082
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465311370975052e-05,
+      "loss": 0.2194,
+      "step": 10083
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4464241949725584e-05,
+      "loss": 0.248,
+      "step": 10084
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4463172464708752e-05,
+      "loss": 0.2267,
+      "step": 10085
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462102915939835e-05,
+      "loss": 0.2537,
+      "step": 10086
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461033303434106e-05,
+      "loss": 0.2212,
+      "step": 10087
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445996362720685e-05,
+      "loss": 0.2183,
+      "step": 10088
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458893887273337e-05,
+      "loss": 0.2514,
+      "step": 10089
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457824083648855e-05,
+      "loss": 0.2525,
+      "step": 10090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445675421634868e-05,
+      "loss": 0.2387,
+      "step": 10091
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455684285388101e-05,
+      "loss": 0.254,
+      "step": 10092
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454614290782395e-05,
+      "loss": 0.2417,
+      "step": 10093
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453544232546843e-05,
+      "loss": 0.2546,
+      "step": 10094
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4452474110696738e-05,
+      "loss": 0.2318,
+      "step": 10095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445140392524736e-05,
+      "loss": 0.2377,
+      "step": 10096
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450333676213996e-05,
+      "loss": 0.2451,
+      "step": 10097
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4449263363611938e-05,
+      "loss": 0.2536,
+      "step": 10098
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4448192987456468e-05,
+      "loss": 0.2463,
+      "step": 10099
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444712254776288e-05,
+      "loss": 0.2487,
+      "step": 10100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4446052044546463e-05,
+      "loss": 0.2434,
+      "step": 10101
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4444981477822503e-05,
+      "loss": 0.2456,
+      "step": 10102
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443910847606302e-05,
+      "loss": 0.2345,
+      "step": 10103
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4442840153913145e-05,
+      "loss": 0.2568,
+      "step": 10104
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444176939675833e-05,
+      "loss": 0.238,
+      "step": 10105
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4440698576157145e-05,
+      "loss": 0.239,
+      "step": 10106
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439627692124894e-05,
+      "loss": 0.2381,
+      "step": 10107
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4438556744676873e-05,
+      "loss": 0.2526,
+      "step": 10108
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437485733828375e-05,
+      "loss": 0.2289,
+      "step": 10109
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4436414659594698e-05,
+      "loss": 0.2513,
+      "step": 10110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4435343521991143e-05,
+      "loss": 0.2387,
+      "step": 10111
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4434272321033012e-05,
+      "loss": 0.2218,
+      "step": 10112
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4433201056735602e-05,
+      "loss": 0.2426,
+      "step": 10113
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443212972911422e-05,
+      "loss": 0.2682,
+      "step": 10114
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431058338184167e-05,
+      "loss": 0.2312,
+      "step": 10115
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429986883960745e-05,
+      "loss": 0.2482,
+      "step": 10116
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442891536645926e-05,
+      "loss": 0.2304,
+      "step": 10117
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4427843785695018e-05,
+      "loss": 0.2492,
+      "step": 10118
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426772141683324e-05,
+      "loss": 0.225,
+      "step": 10119
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425700434439485e-05,
+      "loss": 0.2253,
+      "step": 10120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4424628663978813e-05,
+      "loss": 0.2638,
+      "step": 10121
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423556830316611e-05,
+      "loss": 0.2446,
+      "step": 10122
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4422484933468195e-05,
+      "loss": 0.2443,
+      "step": 10123
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4421412973448874e-05,
+      "loss": 0.2446,
+      "step": 10124
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442034095027396e-05,
+      "loss": 0.2401,
+      "step": 10125
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419268863958766e-05,
+      "loss": 0.2317,
+      "step": 10126
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4418196714518604e-05,
+      "loss": 0.2425,
+      "step": 10127
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441712450196879e-05,
+      "loss": 0.2337,
+      "step": 10128
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4416052226324642e-05,
+      "loss": 0.2382,
+      "step": 10129
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4414979887601471e-05,
+      "loss": 0.2511,
+      "step": 10130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413907485814598e-05,
+      "loss": 0.241,
+      "step": 10131
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4412835020979339e-05,
+      "loss": 0.2367,
+      "step": 10132
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411762493111017e-05,
+      "loss": 0.237,
+      "step": 10133
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410689902224947e-05,
+      "loss": 0.2337,
+      "step": 10134
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4409617248336453e-05,
+      "loss": 0.2237,
+      "step": 10135
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440854453146086e-05,
+      "loss": 0.2404,
+      "step": 10136
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407471751613482e-05,
+      "loss": 0.2454,
+      "step": 10137
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440639890880965e-05,
+      "loss": 0.2493,
+      "step": 10138
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4405326003064684e-05,
+      "loss": 0.2506,
+      "step": 10139
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4404253034393913e-05,
+      "loss": 0.2425,
+      "step": 10140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403180002812663e-05,
+      "loss": 0.2413,
+      "step": 10141
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440210690833626e-05,
+      "loss": 0.2511,
+      "step": 10142
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401033750980031e-05,
+      "loss": 0.2445,
+      "step": 10143
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4399960530759305e-05,
+      "loss": 0.238,
+      "step": 10144
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4398887247689413e-05,
+      "loss": 0.2407,
+      "step": 10145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4397813901785687e-05,
+      "loss": 0.2551,
+      "step": 10146
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4396740493063459e-05,
+      "loss": 0.2432,
+      "step": 10147
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439566702153806e-05,
+      "loss": 0.2438,
+      "step": 10148
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4394593487224824e-05,
+      "loss": 0.2445,
+      "step": 10149
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4393519890139083e-05,
+      "loss": 0.2471,
+      "step": 10150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4392446230296173e-05,
+      "loss": 0.2724,
+      "step": 10151
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4391372507711436e-05,
+      "loss": 0.2473,
+      "step": 10152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390298722400202e-05,
+      "loss": 0.2484,
+      "step": 10153
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389224874377812e-05,
+      "loss": 0.264,
+      "step": 10154
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388150963659605e-05,
+      "loss": 0.2481,
+      "step": 10155
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4387076990260917e-05,
+      "loss": 0.2435,
+      "step": 10156
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4386002954197096e-05,
+      "loss": 0.22,
+      "step": 10157
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.438492885548348e-05,
+      "loss": 0.2269,
+      "step": 10158
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383854694135408e-05,
+      "loss": 0.227,
+      "step": 10159
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4382780470168229e-05,
+      "loss": 0.2631,
+      "step": 10160
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381706183597284e-05,
+      "loss": 0.2436,
+      "step": 10161
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4380631834437915e-05,
+      "loss": 0.2284,
+      "step": 10162
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4379557422705475e-05,
+      "loss": 0.2421,
+      "step": 10163
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4378482948415309e-05,
+      "loss": 0.2697,
+      "step": 10164
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4377408411582761e-05,
+      "loss": 0.2352,
+      "step": 10165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4376333812223181e-05,
+      "loss": 0.2549,
+      "step": 10166
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375259150351925e-05,
+      "loss": 0.2439,
+      "step": 10167
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4374184425984335e-05,
+      "loss": 0.2301,
+      "step": 10168
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373109639135762e-05,
+      "loss": 0.2269,
+      "step": 10169
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372034789821569e-05,
+      "loss": 0.2177,
+      "step": 10170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370959878057097e-05,
+      "loss": 0.2468,
+      "step": 10171
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436988490385771e-05,
+      "loss": 0.2469,
+      "step": 10172
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4368809867238754e-05,
+      "loss": 0.2401,
+      "step": 10173
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436773476821559e-05,
+      "loss": 0.2611,
+      "step": 10174
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366659606803577e-05,
+      "loss": 0.216,
+      "step": 10175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4365584383018068e-05,
+      "loss": 0.2773,
+      "step": 10176
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364509096874419e-05,
+      "loss": 0.2373,
+      "step": 10177
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4363433748387996e-05,
+      "loss": 0.2751,
+      "step": 10178
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362358337574156e-05,
+      "loss": 0.2251,
+      "step": 10179
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361282864448265e-05,
+      "loss": 0.2233,
+      "step": 10180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360207329025677e-05,
+      "loss": 0.2413,
+      "step": 10181
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4359131731321761e-05,
+      "loss": 0.2408,
+      "step": 10182
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4358056071351877e-05,
+      "loss": 0.229,
+      "step": 10183
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435698034913139e-05,
+      "loss": 0.2415,
+      "step": 10184
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355904564675672e-05,
+      "loss": 0.2503,
+      "step": 10185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354828718000082e-05,
+      "loss": 0.2452,
+      "step": 10186
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4353752809119991e-05,
+      "loss": 0.2521,
+      "step": 10187
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4352676838050768e-05,
+      "loss": 0.2687,
+      "step": 10188
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351600804807777e-05,
+      "loss": 0.2477,
+      "step": 10189
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4350524709406392e-05,
+      "loss": 0.2543,
+      "step": 10190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4349448551861989e-05,
+      "loss": 0.2446,
+      "step": 10191
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434837233218993e-05,
+      "loss": 0.2526,
+      "step": 10192
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347296050405595e-05,
+      "loss": 0.2407,
+      "step": 10193
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4346219706524354e-05,
+      "loss": 0.2574,
+      "step": 10194
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345143300561581e-05,
+      "loss": 0.2559,
+      "step": 10195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4344066832532655e-05,
+      "loss": 0.2172,
+      "step": 10196
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434299030245295e-05,
+      "loss": 0.2633,
+      "step": 10197
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4341913710337843e-05,
+      "loss": 0.2253,
+      "step": 10198
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4340837056202712e-05,
+      "loss": 0.243,
+      "step": 10199
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4339760340062939e-05,
+      "loss": 0.2635,
+      "step": 10200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43386835619339e-05,
+      "loss": 0.2585,
+      "step": 10201
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337606721830974e-05,
+      "loss": 0.2442,
+      "step": 10202
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336529819769552e-05,
+      "loss": 0.2367,
+      "step": 10203
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335452855765004e-05,
+      "loss": 0.242,
+      "step": 10204
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4334375829832723e-05,
+      "loss": 0.2317,
+      "step": 10205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333298741988089e-05,
+      "loss": 0.2442,
+      "step": 10206
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433222159224649e-05,
+      "loss": 0.2451,
+      "step": 10207
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4331144380623309e-05,
+      "loss": 0.243,
+      "step": 10208
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330067107133933e-05,
+      "loss": 0.2469,
+      "step": 10209
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328989771793751e-05,
+      "loss": 0.2515,
+      "step": 10210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4327912374618151e-05,
+      "loss": 0.2423,
+      "step": 10211
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326834915622522e-05,
+      "loss": 0.2349,
+      "step": 10212
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432575739482226e-05,
+      "loss": 0.2355,
+      "step": 10213
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324679812232749e-05,
+      "loss": 0.2508,
+      "step": 10214
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323602167869384e-05,
+      "loss": 0.2282,
+      "step": 10215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322524461747558e-05,
+      "loss": 0.2533,
+      "step": 10216
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4321446693882664e-05,
+      "loss": 0.2174,
+      "step": 10217
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43203688642901e-05,
+      "loss": 0.2499,
+      "step": 10218
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431929097298526e-05,
+      "loss": 0.2407,
+      "step": 10219
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318213019983542e-05,
+      "loss": 0.236,
+      "step": 10220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317135005300337e-05,
+      "loss": 0.2377,
+      "step": 10221
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316056928951052e-05,
+      "loss": 0.2321,
+      "step": 10222
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4314978790951081e-05,
+      "loss": 0.2187,
+      "step": 10223
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313900591315829e-05,
+      "loss": 0.2654,
+      "step": 10224
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312822330060689e-05,
+      "loss": 0.2822,
+      "step": 10225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311744007201072e-05,
+      "loss": 0.2192,
+      "step": 10226
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310665622752373e-05,
+      "loss": 0.2434,
+      "step": 10227
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309587176730003e-05,
+      "loss": 0.2399,
+      "step": 10228
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4308508669149358e-05,
+      "loss": 0.233,
+      "step": 10229
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307430100025851e-05,
+      "loss": 0.2414,
+      "step": 10230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4306351469374887e-05,
+      "loss": 0.2359,
+      "step": 10231
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305272777211867e-05,
+      "loss": 0.2474,
+      "step": 10232
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304194023552206e-05,
+      "loss": 0.2654,
+      "step": 10233
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4303115208411313e-05,
+      "loss": 0.2458,
+      "step": 10234
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4302036331804593e-05,
+      "loss": 0.2602,
+      "step": 10235
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300957393747462e-05,
+      "loss": 0.2273,
+      "step": 10236
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4299878394255324e-05,
+      "loss": 0.2288,
+      "step": 10237
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4298799333343597e-05,
+      "loss": 0.2341,
+      "step": 10238
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.42977202110277e-05,
+      "loss": 0.2598,
+      "step": 10239
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4296641027323035e-05,
+      "loss": 0.2529,
+      "step": 10240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4295561782245024e-05,
+      "loss": 0.2555,
+      "step": 10241
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294482475809082e-05,
+      "loss": 0.2449,
+      "step": 10242
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293403108030625e-05,
+      "loss": 0.2446,
+      "step": 10243
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4292323678925073e-05,
+      "loss": 0.2263,
+      "step": 10244
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4291244188507843e-05,
+      "loss": 0.2482,
+      "step": 10245
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4290164636794353e-05,
+      "loss": 0.2474,
+      "step": 10246
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289085023800029e-05,
+      "loss": 0.2286,
+      "step": 10247
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288005349540283e-05,
+      "loss": 0.2316,
+      "step": 10248
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 0.2514,
+      "step": 10249
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285845817286232e-05,
+      "loss": 0.2476,
+      "step": 10250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4284765959322772e-05,
+      "loss": 0.2519,
+      "step": 10251
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4283686040155591e-05,
+      "loss": 0.2542,
+      "step": 10252
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282606059800115e-05,
+      "loss": 0.2629,
+      "step": 10253
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4281526018271761e-05,
+      "loss": 0.2355,
+      "step": 10254
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4280445915585968e-05,
+      "loss": 0.2302,
+      "step": 10255
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4279365751758156e-05,
+      "loss": 0.2348,
+      "step": 10256
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4278285526803757e-05,
+      "loss": 0.2222,
+      "step": 10257
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4277205240738205e-05,
+      "loss": 0.2503,
+      "step": 10258
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276124893576928e-05,
+      "loss": 0.2383,
+      "step": 10259
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275044485335354e-05,
+      "loss": 0.2205,
+      "step": 10260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4273964016028919e-05,
+      "loss": 0.2389,
+      "step": 10261
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4272883485673054e-05,
+      "loss": 0.2494,
+      "step": 10262
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42718028942832e-05,
+      "loss": 0.241,
+      "step": 10263
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270722241874784e-05,
+      "loss": 0.2536,
+      "step": 10264
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269641528463248e-05,
+      "loss": 0.219,
+      "step": 10265
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268560754064025e-05,
+      "loss": 0.2244,
+      "step": 10266
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4267479918692556e-05,
+      "loss": 0.2434,
+      "step": 10267
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4266399022364275e-05,
+      "loss": 0.2382,
+      "step": 10268
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4265318065094627e-05,
+      "loss": 0.2621,
+      "step": 10269
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426423704689905e-05,
+      "loss": 0.2649,
+      "step": 10270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263155967792984e-05,
+      "loss": 0.2494,
+      "step": 10271
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4262074827791875e-05,
+      "loss": 0.2458,
+      "step": 10272
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426099362691116e-05,
+      "loss": 0.2402,
+      "step": 10273
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4259912365166288e-05,
+      "loss": 0.2316,
+      "step": 10274
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258831042572702e-05,
+      "loss": 0.2458,
+      "step": 10275
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425774965914585e-05,
+      "loss": 0.2336,
+      "step": 10276
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4256668214901172e-05,
+      "loss": 0.2311,
+      "step": 10277
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4255586709854123e-05,
+      "loss": 0.2527,
+      "step": 10278
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254505144020147e-05,
+      "loss": 0.2445,
+      "step": 10279
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253423517414697e-05,
+      "loss": 0.2611,
+      "step": 10280
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252341830053218e-05,
+      "loss": 0.2403,
+      "step": 10281
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251260081951162e-05,
+      "loss": 0.2561,
+      "step": 10282
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4250178273123981e-05,
+      "loss": 0.2643,
+      "step": 10283
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4249096403587128e-05,
+      "loss": 0.238,
+      "step": 10284
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424801447335606e-05,
+      "loss": 0.2481,
+      "step": 10285
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246932482446228e-05,
+      "loss": 0.2803,
+      "step": 10286
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4245850430873085e-05,
+      "loss": 0.2431,
+      "step": 10287
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244768318652087e-05,
+      "loss": 0.2593,
+      "step": 10288
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4243686145798696e-05,
+      "loss": 0.2495,
+      "step": 10289
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4242603912328367e-05,
+      "loss": 0.2327,
+      "step": 10290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424152161825656e-05,
+      "loss": 0.2569,
+      "step": 10291
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240439263598734e-05,
+      "loss": 0.2517,
+      "step": 10292
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4239356848370345e-05,
+      "loss": 0.2233,
+      "step": 10293
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238274372586857e-05,
+      "loss": 0.2537,
+      "step": 10294
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237191836263735e-05,
+      "loss": 0.2618,
+      "step": 10295
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4236109239416443e-05,
+      "loss": 0.2228,
+      "step": 10296
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235026582060436e-05,
+      "loss": 0.2303,
+      "step": 10297
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4233943864211187e-05,
+      "loss": 0.265,
+      "step": 10298
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232861085884159e-05,
+      "loss": 0.2416,
+      "step": 10299
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4231778247094821e-05,
+      "loss": 0.2501,
+      "step": 10300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4230695347858634e-05,
+      "loss": 0.2225,
+      "step": 10301
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229612388191073e-05,
+      "loss": 0.2209,
+      "step": 10302
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42285293681076e-05,
+      "loss": 0.2317,
+      "step": 10303
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4227446287623693e-05,
+      "loss": 0.2546,
+      "step": 10304
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4226363146754816e-05,
+      "loss": 0.2501,
+      "step": 10305
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225279945516443e-05,
+      "loss": 0.2332,
+      "step": 10306
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224196683924047e-05,
+      "loss": 0.2325,
+      "step": 10307
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4223113361993105e-05,
+      "loss": 0.2463,
+      "step": 10308
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4222029979739083e-05,
+      "loss": 0.274,
+      "step": 10309
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4220946537177462e-05,
+      "loss": 0.2377,
+      "step": 10310
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219863034323713e-05,
+      "loss": 0.2301,
+      "step": 10311
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4218779471193323e-05,
+      "loss": 0.2573,
+      "step": 10312
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4217695847801757e-05,
+      "loss": 0.2526,
+      "step": 10313
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42166121641645e-05,
+      "loss": 0.2222,
+      "step": 10314
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215528420297031e-05,
+      "loss": 0.2358,
+      "step": 10315
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421444461621483e-05,
+      "loss": 0.2448,
+      "step": 10316
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213360751933376e-05,
+      "loss": 0.2328,
+      "step": 10317
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4212276827468157e-05,
+      "loss": 0.2542,
+      "step": 10318
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211192842834649e-05,
+      "loss": 0.2229,
+      "step": 10319
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210108798048339e-05,
+      "loss": 0.2385,
+      "step": 10320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420902469312471e-05,
+      "loss": 0.2483,
+      "step": 10321
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4207940528079246e-05,
+      "loss": 0.242,
+      "step": 10322
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206856302927438e-05,
+      "loss": 0.2454,
+      "step": 10323
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205772017684773e-05,
+      "loss": 0.2148,
+      "step": 10324
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204687672366735e-05,
+      "loss": 0.2372,
+      "step": 10325
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4203603266988814e-05,
+      "loss": 0.2298,
+      "step": 10326
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202518801566498e-05,
+      "loss": 0.2251,
+      "step": 10327
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420143427611528e-05,
+      "loss": 0.2267,
+      "step": 10328
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200349690650654e-05,
+      "loss": 0.239,
+      "step": 10329
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4199265045188107e-05,
+      "loss": 0.2357,
+      "step": 10330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198180339743136e-05,
+      "loss": 0.2341,
+      "step": 10331
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419709557433123e-05,
+      "loss": 0.2451,
+      "step": 10332
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419601074896789e-05,
+      "loss": 0.2412,
+      "step": 10333
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4194925863668605e-05,
+      "loss": 0.258,
+      "step": 10334
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419384091844888e-05,
+      "loss": 0.2374,
+      "step": 10335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4192755913324205e-05,
+      "loss": 0.2581,
+      "step": 10336
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191670848310081e-05,
+      "loss": 0.2353,
+      "step": 10337
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419058572342201e-05,
+      "loss": 0.2524,
+      "step": 10338
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4189500538675485e-05,
+      "loss": 0.2512,
+      "step": 10339
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4188415294086013e-05,
+      "loss": 0.2415,
+      "step": 10340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4187329989669094e-05,
+      "loss": 0.2352,
+      "step": 10341
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4186244625440228e-05,
+      "loss": 0.2619,
+      "step": 10342
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185159201414921e-05,
+      "loss": 0.245,
+      "step": 10343
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4184073717608678e-05,
+      "loss": 0.2536,
+      "step": 10344
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4182988174037e-05,
+      "loss": 0.2719,
+      "step": 10345
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41819025707154e-05,
+      "loss": 0.2441,
+      "step": 10346
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180816907659376e-05,
+      "loss": 0.2696,
+      "step": 10347
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179731184884442e-05,
+      "loss": 0.2525,
+      "step": 10348
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4178645402406103e-05,
+      "loss": 0.2384,
+      "step": 10349
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417755956023987e-05,
+      "loss": 0.2347,
+      "step": 10350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4176473658401254e-05,
+      "loss": 0.254,
+      "step": 10351
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175387696905769e-05,
+      "loss": 0.238,
+      "step": 10352
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417430167576892e-05,
+      "loss": 0.2429,
+      "step": 10353
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417321559500622e-05,
+      "loss": 0.2531,
+      "step": 10354
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172129454633188e-05,
+      "loss": 0.2513,
+      "step": 10355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4171043254665336e-05,
+      "loss": 0.2363,
+      "step": 10356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169956995118179e-05,
+      "loss": 0.2609,
+      "step": 10357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168870676007234e-05,
+      "loss": 0.2329,
+      "step": 10358
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4167784297348016e-05,
+      "loss": 0.2343,
+      "step": 10359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166697859156045e-05,
+      "loss": 0.229,
+      "step": 10360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165611361446838e-05,
+      "loss": 0.237,
+      "step": 10361
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4164524804235915e-05,
+      "loss": 0.2491,
+      "step": 10362
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4163438187538798e-05,
+      "loss": 0.2703,
+      "step": 10363
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4162351511371008e-05,
+      "loss": 0.2492,
+      "step": 10364
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161264775748067e-05,
+      "loss": 0.2291,
+      "step": 10365
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4160177980685496e-05,
+      "loss": 0.2435,
+      "step": 10366
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4159091126198819e-05,
+      "loss": 0.2603,
+      "step": 10367
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158004212303565e-05,
+      "loss": 0.1991,
+      "step": 10368
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156917239015254e-05,
+      "loss": 0.2435,
+      "step": 10369
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4155830206349413e-05,
+      "loss": 0.2342,
+      "step": 10370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4154743114321575e-05,
+      "loss": 0.2331,
+      "step": 10371
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415365596294726e-05,
+      "loss": 0.2574,
+      "step": 10372
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152568752242002e-05,
+      "loss": 0.218,
+      "step": 10373
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415148148222133e-05,
+      "loss": 0.2425,
+      "step": 10374
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4150394152900773e-05,
+      "loss": 0.2283,
+      "step": 10375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149306764295865e-05,
+      "loss": 0.2383,
+      "step": 10376
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4148219316422135e-05,
+      "loss": 0.2286,
+      "step": 10377
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414713180929512e-05,
+      "loss": 0.2386,
+      "step": 10378
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146044242930352e-05,
+      "loss": 0.247,
+      "step": 10379
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4144956617343366e-05,
+      "loss": 0.2538,
+      "step": 10380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143868932549696e-05,
+      "loss": 0.2273,
+      "step": 10381
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414278118856488e-05,
+      "loss": 0.2035,
+      "step": 10382
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4141693385404456e-05,
+      "loss": 0.2599,
+      "step": 10383
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140605523083963e-05,
+      "loss": 0.2409,
+      "step": 10384
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413951760161894e-05,
+      "loss": 0.2613,
+      "step": 10385
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413842962102492e-05,
+      "loss": 0.2372,
+      "step": 10386
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137341581317451e-05,
+      "loss": 0.2224,
+      "step": 10387
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4136253482512075e-05,
+      "loss": 0.2519,
+      "step": 10388
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4135165324624329e-05,
+      "loss": 0.2344,
+      "step": 10389
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134077107669764e-05,
+      "loss": 0.2311,
+      "step": 10390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4132988831663921e-05,
+      "loss": 0.2257,
+      "step": 10391
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413190049662234e-05,
+      "loss": 0.243,
+      "step": 10392
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413081210256057e-05,
+      "loss": 0.2353,
+      "step": 10393
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4129723649494158e-05,
+      "loss": 0.2296,
+      "step": 10394
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412863513743865e-05,
+      "loss": 0.2607,
+      "step": 10395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127546566409603e-05,
+      "loss": 0.2441,
+      "step": 10396
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126457936422555e-05,
+      "loss": 0.2295,
+      "step": 10397
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412536924749306e-05,
+      "loss": 0.258,
+      "step": 10398
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124280499636668e-05,
+      "loss": 0.2263,
+      "step": 10399
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4123191692868932e-05,
+      "loss": 0.2329,
+      "step": 10400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4122102827205409e-05,
+      "loss": 0.2409,
+      "step": 10401
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121013902661643e-05,
+      "loss": 0.2303,
+      "step": 10402
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4119924919253191e-05,
+      "loss": 0.2109,
+      "step": 10403
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118835876995616e-05,
+      "loss": 0.2498,
+      "step": 10404
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117746775904465e-05,
+      "loss": 0.2743,
+      "step": 10405
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116657615995296e-05,
+      "loss": 0.2307,
+      "step": 10406
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411556839728367e-05,
+      "loss": 0.2407,
+      "step": 10407
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4114479119785146e-05,
+      "loss": 0.2378,
+      "step": 10408
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4113389783515277e-05,
+      "loss": 0.2253,
+      "step": 10409
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112300388489625e-05,
+      "loss": 0.2447,
+      "step": 10410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411121093472376e-05,
+      "loss": 0.2438,
+      "step": 10411
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110121422233234e-05,
+      "loss": 0.2246,
+      "step": 10412
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4109031851033612e-05,
+      "loss": 0.2791,
+      "step": 10413
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107942221140456e-05,
+      "loss": 0.2244,
+      "step": 10414
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4106852532569335e-05,
+      "loss": 0.2432,
+      "step": 10415
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410576278533581e-05,
+      "loss": 0.2451,
+      "step": 10416
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410467297945545e-05,
+      "loss": 0.2257,
+      "step": 10417
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410358311494382e-05,
+      "loss": 0.2434,
+      "step": 10418
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102493191816486e-05,
+      "loss": 0.2294,
+      "step": 10419
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101403210089018e-05,
+      "loss": 0.2295,
+      "step": 10420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4100313169776987e-05,
+      "loss": 0.2594,
+      "step": 10421
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409922307089596e-05,
+      "loss": 0.2528,
+      "step": 10422
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4098132913461514e-05,
+      "loss": 0.2245,
+      "step": 10423
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4097042697489213e-05,
+      "loss": 0.2147,
+      "step": 10424
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095952422994634e-05,
+      "loss": 0.2429,
+      "step": 10425
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094862089993351e-05,
+      "loss": 0.2405,
+      "step": 10426
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093771698500935e-05,
+      "loss": 0.2662,
+      "step": 10427
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092681248532964e-05,
+      "loss": 0.2448,
+      "step": 10428
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091590740105017e-05,
+      "loss": 0.2441,
+      "step": 10429
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4090500173232666e-05,
+      "loss": 0.267,
+      "step": 10430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089409547931487e-05,
+      "loss": 0.2314,
+      "step": 10431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4088318864217062e-05,
+      "loss": 0.25,
+      "step": 10432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408722812210497e-05,
+      "loss": 0.2673,
+      "step": 10433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4086137321610794e-05,
+      "loss": 0.2376,
+      "step": 10434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4085046462750107e-05,
+      "loss": 0.2321,
+      "step": 10435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40839555455385e-05,
+      "loss": 0.2654,
+      "step": 10436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408286456999155e-05,
+      "loss": 0.2588,
+      "step": 10437
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4081773536124843e-05,
+      "loss": 0.2658,
+      "step": 10438
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080682443953962e-05,
+      "loss": 0.2392,
+      "step": 10439
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079591293494494e-05,
+      "loss": 0.2015,
+      "step": 10440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4078500084762022e-05,
+      "loss": 0.2466,
+      "step": 10441
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077408817772133e-05,
+      "loss": 0.2581,
+      "step": 10442
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4076317492540419e-05,
+      "loss": 0.2338,
+      "step": 10443
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4075226109082465e-05,
+      "loss": 0.2299,
+      "step": 10444
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407413466741386e-05,
+      "loss": 0.2708,
+      "step": 10445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4073043167550198e-05,
+      "loss": 0.2371,
+      "step": 10446
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071951609507065e-05,
+      "loss": 0.2531,
+      "step": 10447
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070859993300054e-05,
+      "loss": 0.2379,
+      "step": 10448
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406976831894476e-05,
+      "loss": 0.2667,
+      "step": 10449
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406867658645678e-05,
+      "loss": 0.2504,
+      "step": 10450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067584795851702e-05,
+      "loss": 0.2248,
+      "step": 10451
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4066492947145118e-05,
+      "loss": 0.2334,
+      "step": 10452
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4065401040352631e-05,
+      "loss": 0.261,
+      "step": 10453
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064309075489834e-05,
+      "loss": 0.2513,
+      "step": 10454
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063217052572329e-05,
+      "loss": 0.2305,
+      "step": 10455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062124971615713e-05,
+      "loss": 0.2416,
+      "step": 10456
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4061032832635583e-05,
+      "loss": 0.2451,
+      "step": 10457
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059940635647542e-05,
+      "loss": 0.2386,
+      "step": 10458
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058848380667187e-05,
+      "loss": 0.2296,
+      "step": 10459
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405775606771012e-05,
+      "loss": 0.2298,
+      "step": 10460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4056663696791948e-05,
+      "loss": 0.2441,
+      "step": 10461
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055571267928273e-05,
+      "loss": 0.2343,
+      "step": 10462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40544787811347e-05,
+      "loss": 0.2591,
+      "step": 10463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405338623642683e-05,
+      "loss": 0.2443,
+      "step": 10464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405229363382027e-05,
+      "loss": 0.2396,
+      "step": 10465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405120097333063e-05,
+      "loss": 0.2524,
+      "step": 10466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4050108254973515e-05,
+      "loss": 0.2206,
+      "step": 10467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4049015478764536e-05,
+      "loss": 0.2074,
+      "step": 10468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40479226447193e-05,
+      "loss": 0.2406,
+      "step": 10469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046829752853419e-05,
+      "loss": 0.2306,
+      "step": 10470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4045736803182499e-05,
+      "loss": 0.217,
+      "step": 10471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044643795722156e-05,
+      "loss": 0.2528,
+      "step": 10472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043550730488007e-05,
+      "loss": 0.2325,
+      "step": 10473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4042457607495654e-05,
+      "loss": 0.23,
+      "step": 10474
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041364426760719e-05,
+      "loss": 0.2438,
+      "step": 10475
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040271188298815e-05,
+      "loss": 0.2706,
+      "step": 10476
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4039177892125563e-05,
+      "loss": 0.2335,
+      "step": 10477
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403808453825657e-05,
+      "loss": 0.2412,
+      "step": 10478
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403699112670746e-05,
+      "loss": 0.2426,
+      "step": 10479
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035897657493849e-05,
+      "loss": 0.2526,
+      "step": 10480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4034804130631357e-05,
+      "loss": 0.2379,
+      "step": 10481
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4033710546135604e-05,
+      "loss": 0.2368,
+      "step": 10482
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032616904022209e-05,
+      "loss": 0.2819,
+      "step": 10483
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031523204306799e-05,
+      "loss": 0.2339,
+      "step": 10484
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030429447004992e-05,
+      "loss": 0.2466,
+      "step": 10485
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402933563213241e-05,
+      "loss": 0.2488,
+      "step": 10486
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402824175970468e-05,
+      "loss": 0.2295,
+      "step": 10487
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027147829737424e-05,
+      "loss": 0.2336,
+      "step": 10488
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026053842246272e-05,
+      "loss": 0.2385,
+      "step": 10489
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402495979724685e-05,
+      "loss": 0.2513,
+      "step": 10490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402386569475478e-05,
+      "loss": 0.2252,
+      "step": 10491
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022771534785693e-05,
+      "loss": 0.2488,
+      "step": 10492
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402167731735522e-05,
+      "loss": 0.2868,
+      "step": 10493
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402058304247899e-05,
+      "loss": 0.2276,
+      "step": 10494
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019488710172637e-05,
+      "loss": 0.2495,
+      "step": 10495
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018394320451786e-05,
+      "loss": 0.2525,
+      "step": 10496
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4017299873332071e-05,
+      "loss": 0.2139,
+      "step": 10497
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016205368829128e-05,
+      "loss": 0.2445,
+      "step": 10498
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4015110806958585e-05,
+      "loss": 0.264,
+      "step": 10499
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014016187736083e-05,
+      "loss": 0.2495,
+      "step": 10500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4012921511177257e-05,
+      "loss": 0.2425,
+      "step": 10501
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401182677729774e-05,
+      "loss": 0.2536,
+      "step": 10502
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010731986113175e-05,
+      "loss": 0.2398,
+      "step": 10503
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4009637137639193e-05,
+      "loss": 0.2401,
+      "step": 10504
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008542231891435e-05,
+      "loss": 0.2431,
+      "step": 10505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007447268885545e-05,
+      "loss": 0.2318,
+      "step": 10506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400635224863716e-05,
+      "loss": 0.2307,
+      "step": 10507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005257171161916e-05,
+      "loss": 0.2376,
+      "step": 10508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004162036475466e-05,
+      "loss": 0.2861,
+      "step": 10509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4003066844593445e-05,
+      "loss": 0.2172,
+      "step": 10510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001971595531499e-05,
+      "loss": 0.2294,
+      "step": 10511
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000876289305275e-05,
+      "loss": 0.2539,
+      "step": 10512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999780925930414e-05,
+      "loss": 0.2257,
+      "step": 10513
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3998685505422562e-05,
+      "loss": 0.2677,
+      "step": 10514
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399759002779737e-05,
+      "loss": 0.2107,
+      "step": 10515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3996494493070487e-05,
+      "loss": 0.2459,
+      "step": 10516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995398901257558e-05,
+      "loss": 0.2082,
+      "step": 10517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3994303252374233e-05,
+      "loss": 0.2418,
+      "step": 10518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3993207546436161e-05,
+      "loss": 0.246,
+      "step": 10519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992111783458996e-05,
+      "loss": 0.2318,
+      "step": 10520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991015963458388e-05,
+      "loss": 0.2333,
+      "step": 10521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3989920086449991e-05,
+      "loss": 0.27,
+      "step": 10522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3988824152449459e-05,
+      "loss": 0.2203,
+      "step": 10523
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3987728161472442e-05,
+      "loss": 0.2481,
+      "step": 10524
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986632113534602e-05,
+      "loss": 0.2648,
+      "step": 10525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985536008651588e-05,
+      "loss": 0.2315,
+      "step": 10526
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398443984683906e-05,
+      "loss": 0.2375,
+      "step": 10527
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398334362811268e-05,
+      "loss": 0.2653,
+      "step": 10528
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39822473524881e-05,
+      "loss": 0.2377,
+      "step": 10529
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398115101998098e-05,
+      "loss": 0.2305,
+      "step": 10530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980054630606982e-05,
+      "loss": 0.2201,
+      "step": 10531
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3978958184381767e-05,
+      "loss": 0.2261,
+      "step": 10532
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3977861681320997e-05,
+      "loss": 0.2666,
+      "step": 10533
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976765121440332e-05,
+      "loss": 0.2259,
+      "step": 10534
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3975668504755437e-05,
+      "loss": 0.235,
+      "step": 10535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974571831281977e-05,
+      "loss": 0.2346,
+      "step": 10536
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973475101035616e-05,
+      "loss": 0.2406,
+      "step": 10537
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972378314032017e-05,
+      "loss": 0.2278,
+      "step": 10538
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3971281470286852e-05,
+      "loss": 0.2283,
+      "step": 10539
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970184569815784e-05,
+      "loss": 0.2484,
+      "step": 10540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969087612634483e-05,
+      "loss": 0.2404,
+      "step": 10541
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967990598758617e-05,
+      "loss": 0.225,
+      "step": 10542
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3966893528203855e-05,
+      "loss": 0.2429,
+      "step": 10543
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3965796400985872e-05,
+      "loss": 0.2427,
+      "step": 10544
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3964699217120334e-05,
+      "loss": 0.234,
+      "step": 10545
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3963601976622916e-05,
+      "loss": 0.2431,
+      "step": 10546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396250467950929e-05,
+      "loss": 0.2563,
+      "step": 10547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396140732579513e-05,
+      "loss": 0.2476,
+      "step": 10548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3960309915496114e-05,
+      "loss": 0.2515,
+      "step": 10549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959212448627912e-05,
+      "loss": 0.2436,
+      "step": 10550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958114925206205e-05,
+      "loss": 0.2274,
+      "step": 10551
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3957017345246664e-05,
+      "loss": 0.2464,
+      "step": 10552
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3955919708764971e-05,
+      "loss": 0.2621,
+      "step": 10553
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3954822015776806e-05,
+      "loss": 0.2366,
+      "step": 10554
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953724266297845e-05,
+      "loss": 0.218,
+      "step": 10555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3952626460343774e-05,
+      "loss": 0.2528,
+      "step": 10556
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3951528597930268e-05,
+      "loss": 0.2423,
+      "step": 10557
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395043067907301e-05,
+      "loss": 0.2402,
+      "step": 10558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949332703787683e-05,
+      "loss": 0.2527,
+      "step": 10559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948234672089974e-05,
+      "loss": 0.2414,
+      "step": 10560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3947136583995564e-05,
+      "loss": 0.2304,
+      "step": 10561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394603843952014e-05,
+      "loss": 0.24,
+      "step": 10562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3944940238679384e-05,
+      "loss": 0.2297,
+      "step": 10563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943841981488984e-05,
+      "loss": 0.2405,
+      "step": 10564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942743667964631e-05,
+      "loss": 0.2487,
+      "step": 10565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941645298122011e-05,
+      "loss": 0.2702,
+      "step": 10566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940546871976815e-05,
+      "loss": 0.2378,
+      "step": 10567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3939448389544727e-05,
+      "loss": 0.263,
+      "step": 10568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3938349850841445e-05,
+      "loss": 0.2427,
+      "step": 10569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393725125588266e-05,
+      "loss": 0.2585,
+      "step": 10570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936152604684057e-05,
+      "loss": 0.2707,
+      "step": 10571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3935053897261336e-05,
+      "loss": 0.2609,
+      "step": 10572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3933955133630192e-05,
+      "loss": 0.2528,
+      "step": 10573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3932856313806313e-05,
+      "loss": 0.2401,
+      "step": 10574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931757437805403e-05,
+      "loss": 0.269,
+      "step": 10575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3930658505643149e-05,
+      "loss": 0.245,
+      "step": 10576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3929559517335253e-05,
+      "loss": 0.2613,
+      "step": 10577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928460472897418e-05,
+      "loss": 0.2676,
+      "step": 10578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927361372345334e-05,
+      "loss": 0.2366,
+      "step": 10579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926262215694704e-05,
+      "loss": 0.2615,
+      "step": 10580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925163002961231e-05,
+      "loss": 0.2455,
+      "step": 10581
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3924063734160613e-05,
+      "loss": 0.2451,
+      "step": 10582
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922964409308553e-05,
+      "loss": 0.2495,
+      "step": 10583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921865028420753e-05,
+      "loss": 0.2453,
+      "step": 10584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920765591512919e-05,
+      "loss": 0.2005,
+      "step": 10585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3919666098600753e-05,
+      "loss": 0.2415,
+      "step": 10586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918566549699959e-05,
+      "loss": 0.2509,
+      "step": 10587
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917466944826247e-05,
+      "loss": 0.2291,
+      "step": 10588
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391636728399532e-05,
+      "loss": 0.229,
+      "step": 10589
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3915267567222893e-05,
+      "loss": 0.206,
+      "step": 10590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914167794524663e-05,
+      "loss": 0.2353,
+      "step": 10591
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913067965916343e-05,
+      "loss": 0.2187,
+      "step": 10592
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391196808141365e-05,
+      "loss": 0.2441,
+      "step": 10593
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3910868141032285e-05,
+      "loss": 0.256,
+      "step": 10594
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390976814478797e-05,
+      "loss": 0.2268,
+      "step": 10595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908668092696406e-05,
+      "loss": 0.2283,
+      "step": 10596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3907567984773315e-05,
+      "loss": 0.2401,
+      "step": 10597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3906467821034405e-05,
+      "loss": 0.2289,
+      "step": 10598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905367601495394e-05,
+      "loss": 0.2413,
+      "step": 10599
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3904267326172001e-05,
+      "loss": 0.2149,
+      "step": 10600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3903166995079935e-05,
+      "loss": 0.2275,
+      "step": 10601
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902066608234919e-05,
+      "loss": 0.2438,
+      "step": 10602
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900966165652665e-05,
+      "loss": 0.2403,
+      "step": 10603
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3899865667348897e-05,
+      "loss": 0.2484,
+      "step": 10604
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898765113339331e-05,
+      "loss": 0.2129,
+      "step": 10605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3897664503639694e-05,
+      "loss": 0.2357,
+      "step": 10606
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3896563838265698e-05,
+      "loss": 0.2461,
+      "step": 10607
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895463117233073e-05,
+      "loss": 0.2275,
+      "step": 10608
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3894362340557535e-05,
+      "loss": 0.2417,
+      "step": 10609
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389326150825481e-05,
+      "loss": 0.2742,
+      "step": 10610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3892160620340625e-05,
+      "loss": 0.2546,
+      "step": 10611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.38910596768307e-05,
+      "loss": 0.2263,
+      "step": 10612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889958677740767e-05,
+      "loss": 0.2643,
+      "step": 10613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3888857623086548e-05,
+      "loss": 0.2417,
+      "step": 10614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3887756512883775e-05,
+      "loss": 0.2319,
+      "step": 10615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388665534714817e-05,
+      "loss": 0.2427,
+      "step": 10616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3885554125895466e-05,
+      "loss": 0.2595,
+      "step": 10617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3884452849141395e-05,
+      "loss": 0.2256,
+      "step": 10618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883351516901679e-05,
+      "loss": 0.2267,
+      "step": 10619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882250129192061e-05,
+      "loss": 0.2563,
+      "step": 10620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3881148686028265e-05,
+      "loss": 0.2521,
+      "step": 10621
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880047187426029e-05,
+      "loss": 0.2517,
+      "step": 10622
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3878945633401084e-05,
+      "loss": 0.2593,
+      "step": 10623
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3877844023969162e-05,
+      "loss": 0.2198,
+      "step": 10624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876742359146005e-05,
+      "loss": 0.2485,
+      "step": 10625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875640638947345e-05,
+      "loss": 0.2388,
+      "step": 10626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387453886338892e-05,
+      "loss": 0.2426,
+      "step": 10627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387343703248647e-05,
+      "loss": 0.2544,
+      "step": 10628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872335146255729e-05,
+      "loss": 0.2384,
+      "step": 10629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3871233204712438e-05,
+      "loss": 0.2506,
+      "step": 10630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870131207872337e-05,
+      "loss": 0.2163,
+      "step": 10631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3869029155751168e-05,
+      "loss": 0.2495,
+      "step": 10632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867927048364675e-05,
+      "loss": 0.2374,
+      "step": 10633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3866824885728595e-05,
+      "loss": 0.205,
+      "step": 10634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3865722667858679e-05,
+      "loss": 0.2403,
+      "step": 10635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864620394770663e-05,
+      "loss": 0.2541,
+      "step": 10636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863518066480292e-05,
+      "loss": 0.2313,
+      "step": 10637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386241568300332e-05,
+      "loss": 0.2364,
+      "step": 10638
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386131324435549e-05,
+      "loss": 0.2461,
+      "step": 10639
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3860210750552543e-05,
+      "loss": 0.2632,
+      "step": 10640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3859108201610236e-05,
+      "loss": 0.2143,
+      "step": 10641
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858005597544311e-05,
+      "loss": 0.23,
+      "step": 10642
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385690293837052e-05,
+      "loss": 0.2734,
+      "step": 10643
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855800224104617e-05,
+      "loss": 0.2534,
+      "step": 10644
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854697454762347e-05,
+      "loss": 0.2267,
+      "step": 10645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3853594630359464e-05,
+      "loss": 0.2529,
+      "step": 10646
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852491750911723e-05,
+      "loss": 0.2156,
+      "step": 10647
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3851388816434876e-05,
+      "loss": 0.2633,
+      "step": 10648
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850285826944677e-05,
+      "loss": 0.2355,
+      "step": 10649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849182782456887e-05,
+      "loss": 0.2465,
+      "step": 10650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848079682987253e-05,
+      "loss": 0.2654,
+      "step": 10651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846976528551533e-05,
+      "loss": 0.2393,
+      "step": 10652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3845873319165488e-05,
+      "loss": 0.2543,
+      "step": 10653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844770054844876e-05,
+      "loss": 0.2575,
+      "step": 10654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3843666735605456e-05,
+      "loss": 0.2234,
+      "step": 10655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3842563361462986e-05,
+      "loss": 0.2697,
+      "step": 10656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841459932433227e-05,
+      "loss": 0.2423,
+      "step": 10657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840356448531944e-05,
+      "loss": 0.2692,
+      "step": 10658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3839252909774892e-05,
+      "loss": 0.2487,
+      "step": 10659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383814931617784e-05,
+      "loss": 0.2389,
+      "step": 10660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837045667756552e-05,
+      "loss": 0.2498,
+      "step": 10661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3835941964526791e-05,
+      "loss": 0.2406,
+      "step": 10662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383483820650432e-05,
+      "loss": 0.2593,
+      "step": 10663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833734393704906e-05,
+      "loss": 0.248,
+      "step": 10664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832630526144318e-05,
+      "loss": 0.2261,
+      "step": 10665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3831526603838323e-05,
+      "loss": 0.2375,
+      "step": 10666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3830422626802691e-05,
+      "loss": 0.2307,
+      "step": 10667
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829318595053186e-05,
+      "loss": 0.2339,
+      "step": 10668
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3828214508605583e-05,
+      "loss": 0.2524,
+      "step": 10669
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827110367475646e-05,
+      "loss": 0.2451,
+      "step": 10670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826006171679156e-05,
+      "loss": 0.2176,
+      "step": 10671
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3824901921231883e-05,
+      "loss": 0.2399,
+      "step": 10672
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823797616149597e-05,
+      "loss": 0.2347,
+      "step": 10673
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3822693256448073e-05,
+      "loss": 0.2501,
+      "step": 10674
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821588842143086e-05,
+      "loss": 0.2472,
+      "step": 10675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3820484373250408e-05,
+      "loss": 0.2676,
+      "step": 10676
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381937984978582e-05,
+      "loss": 0.2453,
+      "step": 10677
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818275271765103e-05,
+      "loss": 0.2397,
+      "step": 10678
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3817170639204023e-05,
+      "loss": 0.2509,
+      "step": 10679
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3816065952118368e-05,
+      "loss": 0.2443,
+      "step": 10680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814961210523914e-05,
+      "loss": 0.2211,
+      "step": 10681
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3813856414436441e-05,
+      "loss": 0.2435,
+      "step": 10682
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381275156387173e-05,
+      "loss": 0.253,
+      "step": 10683
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3811646658845565e-05,
+      "loss": 0.2423,
+      "step": 10684
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810541699373727e-05,
+      "loss": 0.2495,
+      "step": 10685
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809436685471998e-05,
+      "loss": 0.2511,
+      "step": 10686
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3808331617156163e-05,
+      "loss": 0.2362,
+      "step": 10687
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3807226494442008e-05,
+      "loss": 0.2246,
+      "step": 10688
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3806121317345318e-05,
+      "loss": 0.24,
+      "step": 10689
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805016085881878e-05,
+      "loss": 0.2136,
+      "step": 10690
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803910800067473e-05,
+      "loss": 0.2393,
+      "step": 10691
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3802805459917896e-05,
+      "loss": 0.2402,
+      "step": 10692
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3801700065448933e-05,
+      "loss": 0.2323,
+      "step": 10693
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3800594616676377e-05,
+      "loss": 0.248,
+      "step": 10694
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799489113616016e-05,
+      "loss": 0.2519,
+      "step": 10695
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3798383556283638e-05,
+      "loss": 0.2349,
+      "step": 10696
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3797277944695038e-05,
+      "loss": 0.2181,
+      "step": 10697
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796172278866006e-05,
+      "loss": 0.2568,
+      "step": 10698
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379506655881234e-05,
+      "loss": 0.2262,
+      "step": 10699
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3793960784549833e-05,
+      "loss": 0.249,
+      "step": 10700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3792854956094274e-05,
+      "loss": 0.2491,
+      "step": 10701
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3791749073461464e-05,
+      "loss": 0.2579,
+      "step": 10702
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.37906431366672e-05,
+      "loss": 0.238,
+      "step": 10703
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3789537145727277e-05,
+      "loss": 0.2529,
+      "step": 10704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3788431100657494e-05,
+      "loss": 0.2433,
+      "step": 10705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3787325001473649e-05,
+      "loss": 0.2535,
+      "step": 10706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3786218848191544e-05,
+      "loss": 0.2362,
+      "step": 10707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3785112640826975e-05,
+      "loss": 0.2654,
+      "step": 10708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3784006379395744e-05,
+      "loss": 0.2505,
+      "step": 10709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782900063913655e-05,
+      "loss": 0.2256,
+      "step": 10710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3781793694396512e-05,
+      "loss": 0.2186,
+      "step": 10711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3780687270860116e-05,
+      "loss": 0.2245,
+      "step": 10712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377958079332027e-05,
+      "loss": 0.2243,
+      "step": 10713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3778474261792784e-05,
+      "loss": 0.2342,
+      "step": 10714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777367676293458e-05,
+      "loss": 0.2505,
+      "step": 10715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776261036838104e-05,
+      "loss": 0.2524,
+      "step": 10716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3775154343442521e-05,
+      "loss": 0.218,
+      "step": 10717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774047596122523e-05,
+      "loss": 0.2403,
+      "step": 10718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772940794893916e-05,
+      "loss": 0.2553,
+      "step": 10719
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771833939772519e-05,
+      "loss": 0.2675,
+      "step": 10720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3770727030774129e-05,
+      "loss": 0.2495,
+      "step": 10721
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3769620067914564e-05,
+      "loss": 0.2444,
+      "step": 10722
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768513051209636e-05,
+      "loss": 0.2525,
+      "step": 10723
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767405980675153e-05,
+      "loss": 0.2451,
+      "step": 10724
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766298856326933e-05,
+      "loss": 0.2373,
+      "step": 10725
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3765191678180792e-05,
+      "loss": 0.2382,
+      "step": 10726
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3764084446252541e-05,
+      "loss": 0.2543,
+      "step": 10727
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3762977160557998e-05,
+      "loss": 0.2713,
+      "step": 10728
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3761869821112975e-05,
+      "loss": 0.209,
+      "step": 10729
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760762427933295e-05,
+      "loss": 0.2285,
+      "step": 10730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759654981034771e-05,
+      "loss": 0.2437,
+      "step": 10731
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375854748043323e-05,
+      "loss": 0.2418,
+      "step": 10732
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757439926144483e-05,
+      "loss": 0.2283,
+      "step": 10733
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375633231818435e-05,
+      "loss": 0.2549,
+      "step": 10734
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3755224656568657e-05,
+      "loss": 0.2226,
+      "step": 10735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754116941313225e-05,
+      "loss": 0.2504,
+      "step": 10736
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753009172433878e-05,
+      "loss": 0.238,
+      "step": 10737
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3751901349946435e-05,
+      "loss": 0.239,
+      "step": 10738
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3750793473866724e-05,
+      "loss": 0.2448,
+      "step": 10739
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749685544210566e-05,
+      "loss": 0.2294,
+      "step": 10740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748577560993791e-05,
+      "loss": 0.2271,
+      "step": 10741
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3747469524232223e-05,
+      "loss": 0.2328,
+      "step": 10742
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374636143394169e-05,
+      "loss": 0.2368,
+      "step": 10743
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3745253290138022e-05,
+      "loss": 0.2297,
+      "step": 10744
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3744145092837045e-05,
+      "loss": 0.2527,
+      "step": 10745
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374303684205459e-05,
+      "loss": 0.2396,
+      "step": 10746
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741928537806481e-05,
+      "loss": 0.238,
+      "step": 10747
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740820180108556e-05,
+      "loss": 0.2261,
+      "step": 10748
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3739711768976647e-05,
+      "loss": 0.229,
+      "step": 10749
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3738603304426588e-05,
+      "loss": 0.226,
+      "step": 10750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737494786474203e-05,
+      "loss": 0.2505,
+      "step": 10751
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3736386215135336e-05,
+      "loss": 0.2378,
+      "step": 10752
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373527759042582e-05,
+      "loss": 0.2603,
+      "step": 10753
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3734168912361485e-05,
+      "loss": 0.2535,
+      "step": 10754
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733060180958174e-05,
+      "loss": 0.242,
+      "step": 10755
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731951396231719e-05,
+      "loss": 0.2642,
+      "step": 10756
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373084255819796e-05,
+      "loss": 0.2397,
+      "step": 10757
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729733666872736e-05,
+      "loss": 0.238,
+      "step": 10758
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728624722271889e-05,
+      "loss": 0.2083,
+      "step": 10759
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727515724411255e-05,
+      "loss": 0.2255,
+      "step": 10760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3726406673306679e-05,
+      "loss": 0.2307,
+      "step": 10761
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725297568973997e-05,
+      "loss": 0.2414,
+      "step": 10762
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3724188411429053e-05,
+      "loss": 0.2193,
+      "step": 10763
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3723079200687694e-05,
+      "loss": 0.2537,
+      "step": 10764
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721969936765763e-05,
+      "loss": 0.249,
+      "step": 10765
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3720860619679104e-05,
+      "loss": 0.2349,
+      "step": 10766
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719751249443562e-05,
+      "loss": 0.2232,
+      "step": 10767
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371864182607498e-05,
+      "loss": 0.2429,
+      "step": 10768
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3717532349589213e-05,
+      "loss": 0.2372,
+      "step": 10769
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716422820002101e-05,
+      "loss": 0.2501,
+      "step": 10770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37153132373295e-05,
+      "loss": 0.2107,
+      "step": 10771
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714203601587253e-05,
+      "loss": 0.2237,
+      "step": 10772
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713093912791213e-05,
+      "loss": 0.2385,
+      "step": 10773
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3711984170957227e-05,
+      "loss": 0.2459,
+      "step": 10774
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710874376101152e-05,
+      "loss": 0.2354,
+      "step": 10775
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709764528238837e-05,
+      "loss": 0.2512,
+      "step": 10776
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708654627386137e-05,
+      "loss": 0.2604,
+      "step": 10777
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707544673558906e-05,
+      "loss": 0.2466,
+      "step": 10778
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3706434666772999e-05,
+      "loss": 0.2533,
+      "step": 10779
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3705324607044266e-05,
+      "loss": 0.2552,
+      "step": 10780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3704214494388567e-05,
+      "loss": 0.2353,
+      "step": 10781
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370310432882176e-05,
+      "loss": 0.2339,
+      "step": 10782
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3701994110359705e-05,
+      "loss": 0.2402,
+      "step": 10783
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700883839018253e-05,
+      "loss": 0.2346,
+      "step": 10784
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3699773514813268e-05,
+      "loss": 0.2368,
+      "step": 10785
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3698663137760609e-05,
+      "loss": 0.2399,
+      "step": 10786
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697552707876138e-05,
+      "loss": 0.2334,
+      "step": 10787
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696442225175714e-05,
+      "loss": 0.2508,
+      "step": 10788
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.36953316896752e-05,
+      "loss": 0.2191,
+      "step": 10789
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369422110139046e-05,
+      "loss": 0.2159,
+      "step": 10790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3693110460337355e-05,
+      "loss": 0.2493,
+      "step": 10791
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3691999766531753e-05,
+      "loss": 0.2631,
+      "step": 10792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369088901998952e-05,
+      "loss": 0.2443,
+      "step": 10793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3689778220726517e-05,
+      "loss": 0.2435,
+      "step": 10794
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3688667368758612e-05,
+      "loss": 0.2502,
+      "step": 10795
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3687556464101675e-05,
+      "loss": 0.2266,
+      "step": 10796
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368644550677157e-05,
+      "loss": 0.2008,
+      "step": 10797
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368533449678417e-05,
+      "loss": 0.2268,
+      "step": 10798
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684223434155348e-05,
+      "loss": 0.2148,
+      "step": 10799
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3683112318900968e-05,
+      "loss": 0.2403,
+      "step": 10800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682001151036901e-05,
+      "loss": 0.2105,
+      "step": 10801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368088993057902e-05,
+      "loss": 0.2387,
+      "step": 10802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679778657543199e-05,
+      "loss": 0.2327,
+      "step": 10803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678667331945315e-05,
+      "loss": 0.2698,
+      "step": 10804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3677555953801233e-05,
+      "loss": 0.2405,
+      "step": 10805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3676444523126839e-05,
+      "loss": 0.2266,
+      "step": 10806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675333039938e-05,
+      "loss": 0.244,
+      "step": 10807
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674221504250595e-05,
+      "loss": 0.2251,
+      "step": 10808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36731099160805e-05,
+      "loss": 0.2426,
+      "step": 10809
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3671998275443598e-05,
+      "loss": 0.2132,
+      "step": 10810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3670886582355765e-05,
+      "loss": 0.2473,
+      "step": 10811
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3669774836832878e-05,
+      "loss": 0.2509,
+      "step": 10812
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668663038890819e-05,
+      "loss": 0.2133,
+      "step": 10813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667551188545469e-05,
+      "loss": 0.2617,
+      "step": 10814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366643928581271e-05,
+      "loss": 0.2188,
+      "step": 10815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3665327330708427e-05,
+      "loss": 0.2335,
+      "step": 10816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664215323248496e-05,
+      "loss": 0.2256,
+      "step": 10817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663103263448805e-05,
+      "loss": 0.245,
+      "step": 10818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661991151325244e-05,
+      "loss": 0.2504,
+      "step": 10819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3660878986893689e-05,
+      "loss": 0.2435,
+      "step": 10820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659766770170033e-05,
+      "loss": 0.2274,
+      "step": 10821
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3658654501170162e-05,
+      "loss": 0.2682,
+      "step": 10822
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657542179909956e-05,
+      "loss": 0.228,
+      "step": 10823
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3656429806405313e-05,
+      "loss": 0.2554,
+      "step": 10824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3655317380672122e-05,
+      "loss": 0.2425,
+      "step": 10825
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654204902726266e-05,
+      "loss": 0.2432,
+      "step": 10826
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653092372583643e-05,
+      "loss": 0.223,
+      "step": 10827
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651979790260137e-05,
+      "loss": 0.2406,
+      "step": 10828
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3650867155771646e-05,
+      "loss": 0.2473,
+      "step": 10829
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364975446913406e-05,
+      "loss": 0.2218,
+      "step": 10830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648641730363272e-05,
+      "loss": 0.2422,
+      "step": 10831
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647528939475183e-05,
+      "loss": 0.2368,
+      "step": 10832
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364641609648568e-05,
+      "loss": 0.2557,
+      "step": 10833
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3645303201410662e-05,
+      "loss": 0.2306,
+      "step": 10834
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3644190254266026e-05,
+      "loss": 0.2241,
+      "step": 10835
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643077255067667e-05,
+      "loss": 0.2377,
+      "step": 10836
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364196420383149e-05,
+      "loss": 0.2502,
+      "step": 10837
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640851100573385e-05,
+      "loss": 0.2366,
+      "step": 10838
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3639737945309258e-05,
+      "loss": 0.2645,
+      "step": 10839
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3638624738055004e-05,
+      "loss": 0.2199,
+      "step": 10840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3637511478826527e-05,
+      "loss": 0.2201,
+      "step": 10841
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3636398167639731e-05,
+      "loss": 0.2287,
+      "step": 10842
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635284804510515e-05,
+      "loss": 0.2305,
+      "step": 10843
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3634171389454785e-05,
+      "loss": 0.2308,
+      "step": 10844
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633057922488444e-05,
+      "loss": 0.2379,
+      "step": 10845
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631944403627396e-05,
+      "loss": 0.2537,
+      "step": 10846
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630830832887545e-05,
+      "loss": 0.2625,
+      "step": 10847
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36297172102848e-05,
+      "loss": 0.2609,
+      "step": 10848
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362860353583507e-05,
+      "loss": 0.2269,
+      "step": 10849
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3627489809554258e-05,
+      "loss": 0.2116,
+      "step": 10850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626376031458273e-05,
+      "loss": 0.2459,
+      "step": 10851
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362526220156303e-05,
+      "loss": 0.2507,
+      "step": 10852
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624148319884432e-05,
+      "loss": 0.2621,
+      "step": 10853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3623034386438392e-05,
+      "loss": 0.2271,
+      "step": 10854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3621920401240824e-05,
+      "loss": 0.2204,
+      "step": 10855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620806364307639e-05,
+      "loss": 0.2338,
+      "step": 10856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3619692275654746e-05,
+      "loss": 0.2088,
+      "step": 10857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618578135298067e-05,
+      "loss": 0.2475,
+      "step": 10858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3617463943253507e-05,
+      "loss": 0.2257,
+      "step": 10859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3616349699536985e-05,
+      "loss": 0.255,
+      "step": 10860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3615235404164419e-05,
+      "loss": 0.246,
+      "step": 10861
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3614121057151723e-05,
+      "loss": 0.2276,
+      "step": 10862
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3613006658514816e-05,
+      "loss": 0.2363,
+      "step": 10863
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3611892208269612e-05,
+      "loss": 0.2064,
+      "step": 10864
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361077770643204e-05,
+      "loss": 0.2199,
+      "step": 10865
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609663153018011e-05,
+      "loss": 0.2369,
+      "step": 10866
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3608548548043442e-05,
+      "loss": 0.2233,
+      "step": 10867
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3607433891524262e-05,
+      "loss": 0.2255,
+      "step": 10868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360631918347639e-05,
+      "loss": 0.2453,
+      "step": 10869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360520442391575e-05,
+      "loss": 0.24,
+      "step": 10870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604089612858262e-05,
+      "loss": 0.2452,
+      "step": 10871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602974750319856e-05,
+      "loss": 0.2086,
+      "step": 10872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601859836316446e-05,
+      "loss": 0.2337,
+      "step": 10873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600744870863967e-05,
+      "loss": 0.244,
+      "step": 10874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599629853978342e-05,
+      "loss": 0.2517,
+      "step": 10875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598514785675499e-05,
+      "loss": 0.2268,
+      "step": 10876
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3597399665971363e-05,
+      "loss": 0.2497,
+      "step": 10877
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596284494881866e-05,
+      "loss": 0.2313,
+      "step": 10878
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3595169272422935e-05,
+      "loss": 0.245,
+      "step": 10879
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3594053998610497e-05,
+      "loss": 0.2184,
+      "step": 10880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3592938673460488e-05,
+      "loss": 0.2235,
+      "step": 10881
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3591823296988837e-05,
+      "loss": 0.2528,
+      "step": 10882
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590707869211478e-05,
+      "loss": 0.2361,
+      "step": 10883
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3589592390144337e-05,
+      "loss": 0.2144,
+      "step": 10884
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3588476859803356e-05,
+      "loss": 0.2169,
+      "step": 10885
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587361278204468e-05,
+      "loss": 0.2308,
+      "step": 10886
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35862456453636e-05,
+      "loss": 0.2236,
+      "step": 10887
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35851299612967e-05,
+      "loss": 0.2384,
+      "step": 10888
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584014226019692e-05,
+      "loss": 0.2554,
+      "step": 10889
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3582898439548522e-05,
+      "loss": 0.2023,
+      "step": 10890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581782601899125e-05,
+      "loss": 0.2542,
+      "step": 10891
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358066671308744e-05,
+      "loss": 0.2523,
+      "step": 10892
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579550773129407e-05,
+      "loss": 0.2577,
+      "step": 10893
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3578434782040963e-05,
+      "loss": 0.242,
+      "step": 10894
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3577318739838052e-05,
+      "loss": 0.2409,
+      "step": 10895
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576202646536614e-05,
+      "loss": 0.2486,
+      "step": 10896
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3575086502152593e-05,
+      "loss": 0.2356,
+      "step": 10897
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573970306701932e-05,
+      "loss": 0.2609,
+      "step": 10898
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3572854060200574e-05,
+      "loss": 0.2491,
+      "step": 10899
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571737762664458e-05,
+      "loss": 0.2495,
+      "step": 10900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357062141410954e-05,
+      "loss": 0.2327,
+      "step": 10901
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569505014551758e-05,
+      "loss": 0.2352,
+      "step": 10902
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356838856400706e-05,
+      "loss": 0.2263,
+      "step": 10903
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3567272062491399e-05,
+      "loss": 0.2515,
+      "step": 10904
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3566155510020717e-05,
+      "loss": 0.2378,
+      "step": 10905
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565038906610963e-05,
+      "loss": 0.2267,
+      "step": 10906
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563922252278088e-05,
+      "loss": 0.2442,
+      "step": 10907
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3562805547038042e-05,
+      "loss": 0.2473,
+      "step": 10908
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356168879090678e-05,
+      "loss": 0.2736,
+      "step": 10909
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560571983900249e-05,
+      "loss": 0.2362,
+      "step": 10910
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559455126034403e-05,
+      "loss": 0.2323,
+      "step": 10911
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3558338217325192e-05,
+      "loss": 0.2445,
+      "step": 10912
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3557221257788576e-05,
+      "loss": 0.2475,
+      "step": 10913
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3556104247440504e-05,
+      "loss": 0.2485,
+      "step": 10914
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554987186296939e-05,
+      "loss": 0.2508,
+      "step": 10915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553870074373827e-05,
+      "loss": 0.2359,
+      "step": 10916
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3552752911687132e-05,
+      "loss": 0.2477,
+      "step": 10917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551635698252811e-05,
+      "loss": 0.2549,
+      "step": 10918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355051843408682e-05,
+      "loss": 0.256,
+      "step": 10919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549401119205119e-05,
+      "loss": 0.2503,
+      "step": 10920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354828375362367e-05,
+      "loss": 0.2422,
+      "step": 10921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3547166337358429e-05,
+      "loss": 0.2329,
+      "step": 10922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.2189,
+      "step": 10923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544931352840422e-05,
+      "loss": 0.2275,
+      "step": 10924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543813784619582e-05,
+      "loss": 0.2124,
+      "step": 10925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542696165778804e-05,
+      "loss": 0.2534,
+      "step": 10926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541578496334045e-05,
+      "loss": 0.2511,
+      "step": 10927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3540460776301277e-05,
+      "loss": 0.2351,
+      "step": 10928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353934300569646e-05,
+      "loss": 0.2292,
+      "step": 10929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3538225184535566e-05,
+      "loss": 0.2423,
+      "step": 10930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537107312834558e-05,
+      "loss": 0.2369,
+      "step": 10931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535989390609408e-05,
+      "loss": 0.2402,
+      "step": 10932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534871417876078e-05,
+      "loss": 0.2529,
+      "step": 10933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3533753394650542e-05,
+      "loss": 0.2728,
+      "step": 10934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3532635320948767e-05,
+      "loss": 0.2368,
+      "step": 10935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531517196786723e-05,
+      "loss": 0.2485,
+      "step": 10936
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3530399022180387e-05,
+      "loss": 0.238,
+      "step": 10937
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529280797145727e-05,
+      "loss": 0.2152,
+      "step": 10938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528162521698715e-05,
+      "loss": 0.2352,
+      "step": 10939
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3527044195855325e-05,
+      "loss": 0.2409,
+      "step": 10940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525925819631533e-05,
+      "loss": 0.235,
+      "step": 10941
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352480739304331e-05,
+      "loss": 0.2222,
+      "step": 10942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523688916106639e-05,
+      "loss": 0.2279,
+      "step": 10943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352257038883749e-05,
+      "loss": 0.2447,
+      "step": 10944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352145181125184e-05,
+      "loss": 0.2478,
+      "step": 10945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352033318336567e-05,
+      "loss": 0.2414,
+      "step": 10946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3519214505194955e-05,
+      "loss": 0.2442,
+      "step": 10947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351809577675568e-05,
+      "loss": 0.2407,
+      "step": 10948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516976998063818e-05,
+      "loss": 0.2566,
+      "step": 10949
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515858169135351e-05,
+      "loss": 0.2328,
+      "step": 10950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3514739289986267e-05,
+      "loss": 0.2211,
+      "step": 10951
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513620360632543e-05,
+      "loss": 0.2574,
+      "step": 10952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512501381090158e-05,
+      "loss": 0.2621,
+      "step": 10953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3511382351375104e-05,
+      "loss": 0.2243,
+      "step": 10954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3510263271503359e-05,
+      "loss": 0.241,
+      "step": 10955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350914414149091e-05,
+      "loss": 0.2386,
+      "step": 10956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3508024961353746e-05,
+      "loss": 0.2553,
+      "step": 10957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506905731107845e-05,
+      "loss": 0.2379,
+      "step": 10958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35057864507692e-05,
+      "loss": 0.2582,
+      "step": 10959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504667120353801e-05,
+      "loss": 0.2501,
+      "step": 10960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503547739877632e-05,
+      "loss": 0.2387,
+      "step": 10961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502428309356681e-05,
+      "loss": 0.2425,
+      "step": 10962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3501308828806942e-05,
+      "loss": 0.259,
+      "step": 10963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3500189298244409e-05,
+      "loss": 0.2327,
+      "step": 10964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499069717685064e-05,
+      "loss": 0.2592,
+      "step": 10965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3497950087144906e-05,
+      "loss": 0.2206,
+      "step": 10966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496830406639925e-05,
+      "loss": 0.2661,
+      "step": 10967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3495710676186116e-05,
+      "loss": 0.243,
+      "step": 10968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349459089579947e-05,
+      "loss": 0.2458,
+      "step": 10969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349347106549599e-05,
+      "loss": 0.2372,
+      "step": 10970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492351185291665e-05,
+      "loss": 0.2445,
+      "step": 10971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3491231255202491e-05,
+      "loss": 0.249,
+      "step": 10972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3490111275244466e-05,
+      "loss": 0.2445,
+      "step": 10973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348899124543359e-05,
+      "loss": 0.2209,
+      "step": 10974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348787116578586e-05,
+      "loss": 0.2437,
+      "step": 10975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486751036317277e-05,
+      "loss": 0.2406,
+      "step": 10976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3485630857043839e-05,
+      "loss": 0.2369,
+      "step": 10977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484510627981543e-05,
+      "loss": 0.233,
+      "step": 10978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34833903491464e-05,
+      "loss": 0.2494,
+      "step": 10979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3482270020554402e-05,
+      "loss": 0.2258,
+      "step": 10980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481149642221557e-05,
+      "loss": 0.268,
+      "step": 10981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348002921416387e-05,
+      "loss": 0.2214,
+      "step": 10982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478908736397343e-05,
+      "loss": 0.2622,
+      "step": 10983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347778820893798e-05,
+      "loss": 0.2338,
+      "step": 10984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347666763180179e-05,
+      "loss": 0.2332,
+      "step": 10985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475547005004774e-05,
+      "loss": 0.2437,
+      "step": 10986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3474426328562944e-05,
+      "loss": 0.245,
+      "step": 10987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473305602492307e-05,
+      "loss": 0.2522,
+      "step": 10988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3472184826808866e-05,
+      "loss": 0.2187,
+      "step": 10989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3471064001528637e-05,
+      "loss": 0.2471,
+      "step": 10990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469943126667628e-05,
+      "loss": 0.2479,
+      "step": 10991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346882220224185e-05,
+      "loss": 0.23,
+      "step": 10992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3467701228267311e-05,
+      "loss": 0.2306,
+      "step": 10993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466580204760026e-05,
+      "loss": 0.2288,
+      "step": 10994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3465459131736007e-05,
+      "loss": 0.2179,
+      "step": 10995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3464338009211268e-05,
+      "loss": 0.2416,
+      "step": 10996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463216837201824e-05,
+      "loss": 0.242,
+      "step": 10997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462095615723692e-05,
+      "loss": 0.2444,
+      "step": 10998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460974344792878e-05,
+      "loss": 0.2219,
+      "step": 10999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3459853024425407e-05,
+      "loss": 0.2609,
+      "step": 11000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3458731654637291e-05,
+      "loss": 0.2431,
+      "step": 11001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3457610235444551e-05,
+      "loss": 0.228,
+      "step": 11002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456488766863208e-05,
+      "loss": 0.2279,
+      "step": 11003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3455367248909274e-05,
+      "loss": 0.2275,
+      "step": 11004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454245681598775e-05,
+      "loss": 0.2231,
+      "step": 11005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453124064947724e-05,
+      "loss": 0.2569,
+      "step": 11006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452002398972152e-05,
+      "loss": 0.2398,
+      "step": 11007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450880683688072e-05,
+      "loss": 0.2428,
+      "step": 11008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3449758919111515e-05,
+      "loss": 0.243,
+      "step": 11009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3448637105258499e-05,
+      "loss": 0.2306,
+      "step": 11010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447515242145048e-05,
+      "loss": 0.2113,
+      "step": 11011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3446393329787187e-05,
+      "loss": 0.2125,
+      "step": 11012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344527136820094e-05,
+      "loss": 0.2585,
+      "step": 11013
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3444149357402338e-05,
+      "loss": 0.237,
+      "step": 11014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443027297407407e-05,
+      "loss": 0.2321,
+      "step": 11015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3441905188232172e-05,
+      "loss": 0.257,
+      "step": 11016
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440783029892659e-05,
+      "loss": 0.2342,
+      "step": 11017
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3439660822404898e-05,
+      "loss": 0.2349,
+      "step": 11018
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438538565784924e-05,
+      "loss": 0.2325,
+      "step": 11019
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437416260048766e-05,
+      "loss": 0.2461,
+      "step": 11020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436293905212447e-05,
+      "loss": 0.2531,
+      "step": 11021
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3435171501292008e-05,
+      "loss": 0.255,
+      "step": 11022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343404904830348e-05,
+      "loss": 0.2216,
+      "step": 11023
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343292654626289e-05,
+      "loss": 0.2536,
+      "step": 11024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431803995186276e-05,
+      "loss": 0.2765,
+      "step": 11025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430681395089675e-05,
+      "loss": 0.2473,
+      "step": 11026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342955874598912e-05,
+      "loss": 0.2347,
+      "step": 11027
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3428436047900645e-05,
+      "loss": 0.2552,
+      "step": 11028
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427313300840289e-05,
+      "loss": 0.2533,
+      "step": 11029
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342619050482409e-05,
+      "loss": 0.2407,
+      "step": 11030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425067659868084e-05,
+      "loss": 0.2517,
+      "step": 11031
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342394476598831e-05,
+      "loss": 0.2588,
+      "step": 11032
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422821823200808e-05,
+      "loss": 0.2701,
+      "step": 11033
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342169883152162e-05,
+      "loss": 0.2376,
+      "step": 11034
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3420575790966781e-05,
+      "loss": 0.2642,
+      "step": 11035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3419452701552343e-05,
+      "loss": 0.2346,
+      "step": 11036
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341832956329434e-05,
+      "loss": 0.2388,
+      "step": 11037
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417206376208812e-05,
+      "loss": 0.251,
+      "step": 11038
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3416083140311811e-05,
+      "loss": 0.2332,
+      "step": 11039
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414959855619377e-05,
+      "loss": 0.2202,
+      "step": 11040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413836522147553e-05,
+      "loss": 0.2329,
+      "step": 11041
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3412713139912392e-05,
+      "loss": 0.244,
+      "step": 11042
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411589708929935e-05,
+      "loss": 0.2215,
+      "step": 11043
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410466229216228e-05,
+      "loss": 0.2335,
+      "step": 11044
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3409342700787322e-05,
+      "loss": 0.2651,
+      "step": 11045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3408219123659265e-05,
+      "loss": 0.2335,
+      "step": 11046
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3407095497848104e-05,
+      "loss": 0.243,
+      "step": 11047
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405971823369892e-05,
+      "loss": 0.2499,
+      "step": 11048
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340484810024068e-05,
+      "loss": 0.2687,
+      "step": 11049
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3403724328476515e-05,
+      "loss": 0.2387,
+      "step": 11050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402600508093451e-05,
+      "loss": 0.2312,
+      "step": 11051
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401476639107541e-05,
+      "loss": 0.2454,
+      "step": 11052
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400352721534836e-05,
+      "loss": 0.2447,
+      "step": 11053
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3399228755391396e-05,
+      "loss": 0.2478,
+      "step": 11054
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398104740693272e-05,
+      "loss": 0.2249,
+      "step": 11055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3396980677456522e-05,
+      "loss": 0.2383,
+      "step": 11056
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3395856565697194e-05,
+      "loss": 0.239,
+      "step": 11057
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394732405431354e-05,
+      "loss": 0.2263,
+      "step": 11058
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3393608196675057e-05,
+      "loss": 0.2457,
+      "step": 11059
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392483939444358e-05,
+      "loss": 0.2293,
+      "step": 11060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391359633755321e-05,
+      "loss": 0.2539,
+      "step": 11061
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3390235279624e-05,
+      "loss": 0.2489,
+      "step": 11062
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389110877066463e-05,
+      "loss": 0.2549,
+      "step": 11063
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3387986426098763e-05,
+      "loss": 0.2593,
+      "step": 11064
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3386861926736968e-05,
+      "loss": 0.2242,
+      "step": 11065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3385737378997133e-05,
+      "loss": 0.2416,
+      "step": 11066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3384612782895326e-05,
+      "loss": 0.2511,
+      "step": 11067
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3383488138447612e-05,
+      "loss": 0.2233,
+      "step": 11068
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382363445670054e-05,
+      "loss": 0.2371,
+      "step": 11069
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381238704578718e-05,
+      "loss": 0.2512,
+      "step": 11070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3380113915189668e-05,
+      "loss": 0.2483,
+      "step": 11071
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337898907751897e-05,
+      "loss": 0.2388,
+      "step": 11072
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337786419158269e-05,
+      "loss": 0.2417,
+      "step": 11073
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33767392573969e-05,
+      "loss": 0.2635,
+      "step": 11074
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337561427497767e-05,
+      "loss": 0.2519,
+      "step": 11075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374489244341064e-05,
+      "loss": 0.2635,
+      "step": 11076
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3373364165503155e-05,
+      "loss": 0.2452,
+      "step": 11077
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3372239038480011e-05,
+      "loss": 0.2393,
+      "step": 11078
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3371113863287706e-05,
+      "loss": 0.26,
+      "step": 11079
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336998863994231e-05,
+      "loss": 0.2481,
+      "step": 11080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368863368459902e-05,
+      "loss": 0.2337,
+      "step": 11081
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3367738048856547e-05,
+      "loss": 0.2538,
+      "step": 11082
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336661268114832e-05,
+      "loss": 0.2713,
+      "step": 11083
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33654872653513e-05,
+      "loss": 0.2207,
+      "step": 11084
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336436180148156e-05,
+      "loss": 0.2334,
+      "step": 11085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3363236289555178e-05,
+      "loss": 0.2501,
+      "step": 11086
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3362110729588232e-05,
+      "loss": 0.2522,
+      "step": 11087
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3360985121596793e-05,
+      "loss": 0.2425,
+      "step": 11088
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359859465596947e-05,
+      "loss": 0.2553,
+      "step": 11089
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3358733761604765e-05,
+      "loss": 0.2359,
+      "step": 11090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3357608009636334e-05,
+      "loss": 0.2678,
+      "step": 11091
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356482209707733e-05,
+      "loss": 0.2252,
+      "step": 11092
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355356361835039e-05,
+      "loss": 0.2349,
+      "step": 11093
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354230466034333e-05,
+      "loss": 0.2265,
+      "step": 11094
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3353104522321705e-05,
+      "loss": 0.2323,
+      "step": 11095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335197853071323e-05,
+      "loss": 0.2492,
+      "step": 11096
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350852491224994e-05,
+      "loss": 0.2641,
+      "step": 11097
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349726403873086e-05,
+      "loss": 0.2548,
+      "step": 11098
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348600268673585e-05,
+      "loss": 0.2514,
+      "step": 11099
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347474085642579e-05,
+      "loss": 0.2494,
+      "step": 11100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3346347854796155e-05,
+      "loss": 0.1975,
+      "step": 11101
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345221576150403e-05,
+      "loss": 0.2415,
+      "step": 11102
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3344095249721404e-05,
+      "loss": 0.2395,
+      "step": 11103
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334296887552525e-05,
+      "loss": 0.2316,
+      "step": 11104
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3341842453578028e-05,
+      "loss": 0.2315,
+      "step": 11105
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340715983895831e-05,
+      "loss": 0.243,
+      "step": 11106
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333958946649475e-05,
+      "loss": 0.2218,
+      "step": 11107
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3338462901390874e-05,
+      "loss": 0.2228,
+      "step": 11108
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3337336288600297e-05,
+      "loss": 0.2431,
+      "step": 11109
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336209628139107e-05,
+      "loss": 0.2501,
+      "step": 11110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335082920023404e-05,
+      "loss": 0.2556,
+      "step": 11111
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3333956164269273e-05,
+      "loss": 0.2491,
+      "step": 11112
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332829360892819e-05,
+      "loss": 0.2345,
+      "step": 11113
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333170250991013e-05,
+      "loss": 0.2436,
+      "step": 11114
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3330575611337304e-05,
+      "loss": 0.2446,
+      "step": 11115
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329448665190438e-05,
+      "loss": 0.2414,
+      "step": 11116
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332832167148563e-05,
+      "loss": 0.2513,
+      "step": 11117
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327194630238976e-05,
+      "loss": 0.2242,
+      "step": 11118
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326067541466576e-05,
+      "loss": 0.2385,
+      "step": 11119
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3324940405184527e-05,
+      "loss": 0.23,
+      "step": 11120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323813221408934e-05,
+      "loss": 0.2441,
+      "step": 11121
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322685990155897e-05,
+      "loss": 0.2163,
+      "step": 11122
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3321558711441512e-05,
+      "loss": 0.2679,
+      "step": 11123
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3320431385281884e-05,
+      "loss": 0.2433,
+      "step": 11124
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331930401169312e-05,
+      "loss": 0.2382,
+      "step": 11125
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318176590691316e-05,
+      "loss": 0.2308,
+      "step": 11126
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331704912229258e-05,
+      "loss": 0.2559,
+      "step": 11127
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315921606513019e-05,
+      "loss": 0.2381,
+      "step": 11128
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3314794043368734e-05,
+      "loss": 0.2233,
+      "step": 11129
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3313666432875833e-05,
+      "loss": 0.231,
+      "step": 11130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312538775050426e-05,
+      "loss": 0.2336,
+      "step": 11131
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3311411069908613e-05,
+      "loss": 0.2618,
+      "step": 11132
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331028331746651e-05,
+      "loss": 0.2233,
+      "step": 11133
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330915551774022e-05,
+      "loss": 0.2521,
+      "step": 11134
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3308027670745858e-05,
+      "loss": 0.235,
+      "step": 11135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330689977649953e-05,
+      "loss": 0.2503,
+      "step": 11136
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3305771835017347e-05,
+      "loss": 0.2577,
+      "step": 11137
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3304643846315425e-05,
+      "loss": 0.2633,
+      "step": 11138
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3303515810409868e-05,
+      "loss": 0.2253,
+      "step": 11139
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302387727316796e-05,
+      "loss": 0.1995,
+      "step": 11140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330125959705232e-05,
+      "loss": 0.2346,
+      "step": 11141
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300131419632558e-05,
+      "loss": 0.2469,
+      "step": 11142
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299003195073618e-05,
+      "loss": 0.2446,
+      "step": 11143
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3297874923391618e-05,
+      "loss": 0.2419,
+      "step": 11144
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296746604602677e-05,
+      "loss": 0.2328,
+      "step": 11145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329561823872291e-05,
+      "loss": 0.232,
+      "step": 11146
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3294489825768437e-05,
+      "loss": 0.2175,
+      "step": 11147
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3293361365755373e-05,
+      "loss": 0.2622,
+      "step": 11148
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292232858699839e-05,
+      "loss": 0.245,
+      "step": 11149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3291104304617952e-05,
+      "loss": 0.2542,
+      "step": 11150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3289975703525833e-05,
+      "loss": 0.225,
+      "step": 11151
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3288847055439607e-05,
+      "loss": 0.2388,
+      "step": 11152
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328771836037539e-05,
+      "loss": 0.2455,
+      "step": 11153
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3286589618349309e-05,
+      "loss": 0.2334,
+      "step": 11154
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285460829377486e-05,
+      "loss": 0.2495,
+      "step": 11155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3284331993476042e-05,
+      "loss": 0.2543,
+      "step": 11156
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3283203110661102e-05,
+      "loss": 0.2259,
+      "step": 11157
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3282074180948794e-05,
+      "loss": 0.2439,
+      "step": 11158
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328094520435524e-05,
+      "loss": 0.2304,
+      "step": 11159
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279816180896568e-05,
+      "loss": 0.2129,
+      "step": 11160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278687110588908e-05,
+      "loss": 0.221,
+      "step": 11161
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3277557993448383e-05,
+      "loss": 0.2376,
+      "step": 11162
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327642882949112e-05,
+      "loss": 0.2637,
+      "step": 11163
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275299618733255e-05,
+      "loss": 0.2386,
+      "step": 11164
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3274170361190912e-05,
+      "loss": 0.2446,
+      "step": 11165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273041056880223e-05,
+      "loss": 0.2351,
+      "step": 11166
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327191170581732e-05,
+      "loss": 0.2349,
+      "step": 11167
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270782308018333e-05,
+      "loss": 0.2245,
+      "step": 11168
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269652863499395e-05,
+      "loss": 0.255,
+      "step": 11169
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326852337227664e-05,
+      "loss": 0.2352,
+      "step": 11170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32673938343662e-05,
+      "loss": 0.24,
+      "step": 11171
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266264249784209e-05,
+      "loss": 0.2616,
+      "step": 11172
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3265134618546805e-05,
+      "loss": 0.2456,
+      "step": 11173
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264004940670122e-05,
+      "loss": 0.2286,
+      "step": 11174
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262875216170297e-05,
+      "loss": 0.2675,
+      "step": 11175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261745445063462e-05,
+      "loss": 0.2419,
+      "step": 11176
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260615627365763e-05,
+      "loss": 0.2276,
+      "step": 11177
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3259485763093332e-05,
+      "loss": 0.2237,
+      "step": 11178
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325835585226231e-05,
+      "loss": 0.2274,
+      "step": 11179
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3257225894888839e-05,
+      "loss": 0.2226,
+      "step": 11180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256095890989058e-05,
+      "loss": 0.2271,
+      "step": 11181
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254965840579105e-05,
+      "loss": 0.2033,
+      "step": 11182
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253835743675125e-05,
+      "loss": 0.2097,
+      "step": 11183
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3252705600293259e-05,
+      "loss": 0.2311,
+      "step": 11184
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3251575410449649e-05,
+      "loss": 0.2424,
+      "step": 11185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250445174160442e-05,
+      "loss": 0.2149,
+      "step": 11186
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324931489144178e-05,
+      "loss": 0.2442,
+      "step": 11187
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324818456230981e-05,
+      "loss": 0.2354,
+      "step": 11188
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247054186780672e-05,
+      "loss": 0.2485,
+      "step": 11189
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3245923764870519e-05,
+      "loss": 0.2654,
+      "step": 11190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244793296595495e-05,
+      "loss": 0.2227,
+      "step": 11191
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3243662781971746e-05,
+      "loss": 0.265,
+      "step": 11192
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242532221015423e-05,
+      "loss": 0.2418,
+      "step": 11193
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241401613742677e-05,
+      "loss": 0.24,
+      "step": 11194
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240270960169653e-05,
+      "loss": 0.2228,
+      "step": 11195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32391402603125e-05,
+      "loss": 0.223,
+      "step": 11196
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3238009514187377e-05,
+      "loss": 0.2418,
+      "step": 11197
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236878721810427e-05,
+      "loss": 0.2567,
+      "step": 11198
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3235747883197806e-05,
+      "loss": 0.2348,
+      "step": 11199
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323461699836567e-05,
+      "loss": 0.2165,
+      "step": 11200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233486067330167e-05,
+      "loss": 0.2401,
+      "step": 11201
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232355090107456e-05,
+      "loss": 0.2597,
+      "step": 11202
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3231224066713687e-05,
+      "loss": 0.2356,
+      "step": 11203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230092997165018e-05,
+      "loss": 0.2629,
+      "step": 11204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3228961881477606e-05,
+      "loss": 0.243,
+      "step": 11205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322783071966761e-05,
+      "loss": 0.2338,
+      "step": 11206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3226699511751181e-05,
+      "loss": 0.2259,
+      "step": 11207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3225568257744486e-05,
+      "loss": 0.2541,
+      "step": 11208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3224436957663675e-05,
+      "loss": 0.2488,
+      "step": 11209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3223305611524912e-05,
+      "loss": 0.2271,
+      "step": 11210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222174219344354e-05,
+      "loss": 0.2147,
+      "step": 11211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221042781138169e-05,
+      "loss": 0.2448,
+      "step": 11212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219911296922513e-05,
+      "loss": 0.227,
+      "step": 11213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321877976671355e-05,
+      "loss": 0.248,
+      "step": 11214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3217648190527439e-05,
+      "loss": 0.2628,
+      "step": 11215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216516568380346e-05,
+      "loss": 0.2356,
+      "step": 11216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3215384900288436e-05,
+      "loss": 0.2506,
+      "step": 11217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214253186267876e-05,
+      "loss": 0.214,
+      "step": 11218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213121426334828e-05,
+      "loss": 0.2429,
+      "step": 11219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211989620505456e-05,
+      "loss": 0.2202,
+      "step": 11220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210857768795934e-05,
+      "loss": 0.2585,
+      "step": 11221
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3209725871222423e-05,
+      "loss": 0.2643,
+      "step": 11222
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320859392780109e-05,
+      "loss": 0.2362,
+      "step": 11223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207461938548112e-05,
+      "loss": 0.2086,
+      "step": 11224
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206329903479648e-05,
+      "loss": 0.2595,
+      "step": 11225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3205197822611876e-05,
+      "loss": 0.2318,
+      "step": 11226
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204065695960966e-05,
+      "loss": 0.2177,
+      "step": 11227
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202933523543084e-05,
+      "loss": 0.2497,
+      "step": 11228
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201801305374404e-05,
+      "loss": 0.2274,
+      "step": 11229
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3200669041471103e-05,
+      "loss": 0.243,
+      "step": 11230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319953673184935e-05,
+      "loss": 0.2302,
+      "step": 11231
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198404376525318e-05,
+      "loss": 0.2471,
+      "step": 11232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3197271975515186e-05,
+      "loss": 0.2426,
+      "step": 11233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3196139528835128e-05,
+      "loss": 0.267,
+      "step": 11234
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3195007036501315e-05,
+      "loss": 0.2423,
+      "step": 11235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319387449852993e-05,
+      "loss": 0.2564,
+      "step": 11236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3192741914937148e-05,
+      "loss": 0.2542,
+      "step": 11237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191609285739144e-05,
+      "loss": 0.2555,
+      "step": 11238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.31904766109521e-05,
+      "loss": 0.2275,
+      "step": 11239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189343890592194e-05,
+      "loss": 0.2331,
+      "step": 11240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3188211124675605e-05,
+      "loss": 0.2721,
+      "step": 11241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3187078313218517e-05,
+      "loss": 0.2288,
+      "step": 11242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185945456237106e-05,
+      "loss": 0.2261,
+      "step": 11243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3184812553747555e-05,
+      "loss": 0.2446,
+      "step": 11244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183679605766049e-05,
+      "loss": 0.2291,
+      "step": 11245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3182546612308773e-05,
+      "loss": 0.253,
+      "step": 11246
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3181413573391906e-05,
+      "loss": 0.2565,
+      "step": 11247
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.318028048903163e-05,
+      "loss": 0.2317,
+      "step": 11248
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179147359244136e-05,
+      "loss": 0.2735,
+      "step": 11249
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3178014184045607e-05,
+      "loss": 0.2392,
+      "step": 11250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317688096345223e-05,
+      "loss": 0.2374,
+      "step": 11251
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3175747697480192e-05,
+      "loss": 0.219,
+      "step": 11252
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317461438614568e-05,
+      "loss": 0.2486,
+      "step": 11253
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3173481029464883e-05,
+      "loss": 0.231,
+      "step": 11254
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317234762745399e-05,
+      "loss": 0.2514,
+      "step": 11255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171214180129188e-05,
+      "loss": 0.2292,
+      "step": 11256
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3170080687506668e-05,
+      "loss": 0.2612,
+      "step": 11257
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168947149602626e-05,
+      "loss": 0.2327,
+      "step": 11258
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167813566433248e-05,
+      "loss": 0.2457,
+      "step": 11259
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3166679938014728e-05,
+      "loss": 0.241,
+      "step": 11260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165546264363256e-05,
+      "loss": 0.2318,
+      "step": 11261
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3164412545495029e-05,
+      "loss": 0.2379,
+      "step": 11262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3163278781426244e-05,
+      "loss": 0.242,
+      "step": 11263
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162144972173086e-05,
+      "loss": 0.2401,
+      "step": 11264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3161011117751756e-05,
+      "loss": 0.2661,
+      "step": 11265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3159877218178451e-05,
+      "loss": 0.2443,
+      "step": 11266
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158743273469372e-05,
+      "loss": 0.2564,
+      "step": 11267
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3157609283640704e-05,
+      "loss": 0.2182,
+      "step": 11268
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3156475248708658e-05,
+      "loss": 0.2364,
+      "step": 11269
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155341168689422e-05,
+      "loss": 0.2503,
+      "step": 11270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31542070435992e-05,
+      "loss": 0.262,
+      "step": 11271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153072873454191e-05,
+      "loss": 0.242,
+      "step": 11272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31519386582706e-05,
+      "loss": 0.269,
+      "step": 11273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315080439806462e-05,
+      "loss": 0.2791,
+      "step": 11274
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149670092852462e-05,
+      "loss": 0.2436,
+      "step": 11275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314853574265032e-05,
+      "loss": 0.234,
+      "step": 11276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31474013474744e-05,
+      "loss": 0.2194,
+      "step": 11277
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146266907340907e-05,
+      "loss": 0.2299,
+      "step": 11278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145132422266048e-05,
+      "loss": 0.2115,
+      "step": 11279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143997892266023e-05,
+      "loss": 0.2383,
+      "step": 11280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142863317357038e-05,
+      "loss": 0.2252,
+      "step": 11281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31417286975553e-05,
+      "loss": 0.2256,
+      "step": 11282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3140594032877016e-05,
+      "loss": 0.2349,
+      "step": 11283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3139459323338397e-05,
+      "loss": 0.232,
+      "step": 11284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3138324568955649e-05,
+      "loss": 0.2349,
+      "step": 11285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313718976974498e-05,
+      "loss": 0.2238,
+      "step": 11286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136054925722598e-05,
+      "loss": 0.264,
+      "step": 11287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3134920036904717e-05,
+      "loss": 0.235,
+      "step": 11288
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133785103307543e-05,
+      "loss": 0.2519,
+      "step": 11289
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3132650124947292e-05,
+      "loss": 0.2434,
+      "step": 11290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3131515101840176e-05,
+      "loss": 0.2571,
+      "step": 11291
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3130380034002404e-05,
+      "loss": 0.2759,
+      "step": 11292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3129244921450192e-05,
+      "loss": 0.228,
+      "step": 11293
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128109764199752e-05,
+      "loss": 0.2598,
+      "step": 11294
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3126974562267298e-05,
+      "loss": 0.2155,
+      "step": 11295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125839315669053e-05,
+      "loss": 0.2391,
+      "step": 11296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3124704024421223e-05,
+      "loss": 0.2333,
+      "step": 11297
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312356868854003e-05,
+      "loss": 0.2529,
+      "step": 11298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122433308041688e-05,
+      "loss": 0.2276,
+      "step": 11299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121297882942419e-05,
+      "loss": 0.2405,
+      "step": 11300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3120162413258437e-05,
+      "loss": 0.2375,
+      "step": 11301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3119026899005966e-05,
+      "loss": 0.2563,
+      "step": 11302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3117891340201221e-05,
+      "loss": 0.2213,
+      "step": 11303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116755736860422e-05,
+      "loss": 0.2187,
+      "step": 11304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115620088999795e-05,
+      "loss": 0.2771,
+      "step": 11305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3114484396635559e-05,
+      "loss": 0.252,
+      "step": 11306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113348659783936e-05,
+      "loss": 0.2123,
+      "step": 11307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3112212878461148e-05,
+      "loss": 0.231,
+      "step": 11308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111077052683419e-05,
+      "loss": 0.2388,
+      "step": 11309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3109941182466974e-05,
+      "loss": 0.236,
+      "step": 11310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108805267828038e-05,
+      "loss": 0.233,
+      "step": 11311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107669308782835e-05,
+      "loss": 0.2334,
+      "step": 11312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3106533305347596e-05,
+      "loss": 0.2353,
+      "step": 11313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310539725753854e-05,
+      "loss": 0.2313,
+      "step": 11314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3104261165371896e-05,
+      "loss": 0.2122,
+      "step": 11315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103125028863896e-05,
+      "loss": 0.2252,
+      "step": 11316
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3101988848030767e-05,
+      "loss": 0.2312,
+      "step": 11317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310085262288874e-05,
+      "loss": 0.2555,
+      "step": 11318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099716353454041e-05,
+      "loss": 0.2594,
+      "step": 11319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3098580039742902e-05,
+      "loss": 0.242,
+      "step": 11320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3097443681771553e-05,
+      "loss": 0.262,
+      "step": 11321
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3096307279556229e-05,
+      "loss": 0.2204,
+      "step": 11322
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095170833113158e-05,
+      "loss": 0.2449,
+      "step": 11323
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309403434245858e-05,
+      "loss": 0.2408,
+      "step": 11324
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3092897807608722e-05,
+      "loss": 0.2185,
+      "step": 11325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309176122857982e-05,
+      "loss": 0.2238,
+      "step": 11326
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090624605388109e-05,
+      "loss": 0.2155,
+      "step": 11327
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3089487938049826e-05,
+      "loss": 0.2102,
+      "step": 11328
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088351226581204e-05,
+      "loss": 0.2446,
+      "step": 11329
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3087214470998487e-05,
+      "loss": 0.2119,
+      "step": 11330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3086077671317904e-05,
+      "loss": 0.2379,
+      "step": 11331
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3084940827555696e-05,
+      "loss": 0.2368,
+      "step": 11332
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3083803939728106e-05,
+      "loss": 0.2394,
+      "step": 11333
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082667007851366e-05,
+      "loss": 0.2506,
+      "step": 11334
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3081530031941723e-05,
+      "loss": 0.2452,
+      "step": 11335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3080393012015412e-05,
+      "loss": 0.2376,
+      "step": 11336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3079255948088674e-05,
+      "loss": 0.233,
+      "step": 11337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3078118840177756e-05,
+      "loss": 0.2357,
+      "step": 11338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076981688298898e-05,
+      "loss": 0.2413,
+      "step": 11339
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3075844492468342e-05,
+      "loss": 0.2274,
+      "step": 11340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3074707252702332e-05,
+      "loss": 0.2351,
+      "step": 11341
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073569969017115e-05,
+      "loss": 0.2434,
+      "step": 11342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072432641428931e-05,
+      "loss": 0.225,
+      "step": 11343
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307129526995403e-05,
+      "loss": 0.2408,
+      "step": 11344
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070157854608655e-05,
+      "loss": 0.2288,
+      "step": 11345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306902039540906e-05,
+      "loss": 0.265,
+      "step": 11346
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3067882892371487e-05,
+      "loss": 0.2499,
+      "step": 11347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066745345512179e-05,
+      "loss": 0.2169,
+      "step": 11348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3065607754847392e-05,
+      "loss": 0.2286,
+      "step": 11349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3064470120393373e-05,
+      "loss": 0.2205,
+      "step": 11350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3063332442166374e-05,
+      "loss": 0.2633,
+      "step": 11351
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062194720182645e-05,
+      "loss": 0.2324,
+      "step": 11352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3061056954458435e-05,
+      "loss": 0.2418,
+      "step": 11353
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059919145009996e-05,
+      "loss": 0.2294,
+      "step": 11354
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3058781291853583e-05,
+      "loss": 0.2434,
+      "step": 11355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057643395005446e-05,
+      "loss": 0.2455,
+      "step": 11356
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3056505454481845e-05,
+      "loss": 0.2302,
+      "step": 11357
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055367470299025e-05,
+      "loss": 0.257,
+      "step": 11358
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305422944247325e-05,
+      "loss": 0.2292,
+      "step": 11359
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053091371020769e-05,
+      "loss": 0.2286,
+      "step": 11360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305195325595784e-05,
+      "loss": 0.2334,
+      "step": 11361
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3050815097300724e-05,
+      "loss": 0.2403,
+      "step": 11362
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049676895065674e-05,
+      "loss": 0.2346,
+      "step": 11363
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304853864926895e-05,
+      "loss": 0.2643,
+      "step": 11364
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304740035992681e-05,
+      "loss": 0.237,
+      "step": 11365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3046262027055512e-05,
+      "loss": 0.2474,
+      "step": 11366
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045123650671316e-05,
+      "loss": 0.2164,
+      "step": 11367
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043985230790488e-05,
+      "loss": 0.2422,
+      "step": 11368
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3042846767429283e-05,
+      "loss": 0.2279,
+      "step": 11369
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041708260603965e-05,
+      "loss": 0.2378,
+      "step": 11370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3040569710330795e-05,
+      "loss": 0.2377,
+      "step": 11371
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039431116626043e-05,
+      "loss": 0.2416,
+      "step": 11372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3038292479505961e-05,
+      "loss": 0.2266,
+      "step": 11373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3037153798986824e-05,
+      "loss": 0.2339,
+      "step": 11374
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3036015075084891e-05,
+      "loss": 0.2419,
+      "step": 11375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034876307816429e-05,
+      "loss": 0.2465,
+      "step": 11376
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033737497197704e-05,
+      "loss": 0.2526,
+      "step": 11377
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3032598643244983e-05,
+      "loss": 0.2375,
+      "step": 11378
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3031459745974536e-05,
+      "loss": 0.2259,
+      "step": 11379
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030320805402627e-05,
+      "loss": 0.2176,
+      "step": 11380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029181821545526e-05,
+      "loss": 0.2282,
+      "step": 11381
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028042794419502e-05,
+      "loss": 0.2343,
+      "step": 11382
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026903724040826e-05,
+      "loss": 0.2323,
+      "step": 11383
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302576461042577e-05,
+      "loss": 0.2246,
+      "step": 11384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3024625453590603e-05,
+      "loss": 0.2389,
+      "step": 11385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023486253551596e-05,
+      "loss": 0.2578,
+      "step": 11386
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3022347010325017e-05,
+      "loss": 0.2342,
+      "step": 11387
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021207723927151e-05,
+      "loss": 0.2405,
+      "step": 11388
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020068394374261e-05,
+      "loss": 0.2266,
+      "step": 11389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301892902168263e-05,
+      "loss": 0.244,
+      "step": 11390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017789605868523e-05,
+      "loss": 0.2321,
+      "step": 11391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016650146948222e-05,
+      "loss": 0.2574,
+      "step": 11392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3015510644938001e-05,
+      "loss": 0.2312,
+      "step": 11393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014371099854137e-05,
+      "loss": 0.2576,
+      "step": 11394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3013231511712908e-05,
+      "loss": 0.2601,
+      "step": 11395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301209188053059e-05,
+      "loss": 0.2427,
+      "step": 11396
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3010952206323463e-05,
+      "loss": 0.2414,
+      "step": 11397
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300981248910781e-05,
+      "loss": 0.2294,
+      "step": 11398
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3008672728899901e-05,
+      "loss": 0.2203,
+      "step": 11399
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3007532925716024e-05,
+      "loss": 0.2529,
+      "step": 11400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300639307957246e-05,
+      "loss": 0.2362,
+      "step": 11401
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005253190485486e-05,
+      "loss": 0.2309,
+      "step": 11402
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3004113258471388e-05,
+      "loss": 0.2613,
+      "step": 11403
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002973283546444e-05,
+      "loss": 0.2525,
+      "step": 11404
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001833265726948e-05,
+      "loss": 0.2469,
+      "step": 11405
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000693205029173e-05,
+      "loss": 0.2413,
+      "step": 11406
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2999553101469409e-05,
+      "loss": 0.2762,
+      "step": 11407
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299841295506394e-05,
+      "loss": 0.2503,
+      "step": 11408
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997272765829049e-05,
+      "loss": 0.2231,
+      "step": 11409
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996132533781029e-05,
+      "loss": 0.2511,
+      "step": 11410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2994992258936161e-05,
+      "loss": 0.2528,
+      "step": 11411
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299385194131074e-05,
+      "loss": 0.2408,
+      "step": 11412
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2992711580921047e-05,
+      "loss": 0.237,
+      "step": 11413
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991571177783371e-05,
+      "loss": 0.2174,
+      "step": 11414
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2990430731914008e-05,
+      "loss": 0.2305,
+      "step": 11415
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298929024332924e-05,
+      "loss": 0.2381,
+      "step": 11416
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988149712045366e-05,
+      "loss": 0.2295,
+      "step": 11417
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2987009138078675e-05,
+      "loss": 0.2534,
+      "step": 11418
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2985868521445452e-05,
+      "loss": 0.2248,
+      "step": 11419
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2984727862161996e-05,
+      "loss": 0.2107,
+      "step": 11420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2983587160244602e-05,
+      "loss": 0.273,
+      "step": 11421
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982446415709558e-05,
+      "loss": 0.242,
+      "step": 11422
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981305628573165e-05,
+      "loss": 0.2236,
+      "step": 11423
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298016479885171e-05,
+      "loss": 0.237,
+      "step": 11424
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2979023926561498e-05,
+      "loss": 0.2279,
+      "step": 11425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.297788301171882e-05,
+      "loss": 0.2453,
+      "step": 11426
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2976742054339968e-05,
+      "loss": 0.2241,
+      "step": 11427
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297560105444125e-05,
+      "loss": 0.2592,
+      "step": 11428
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2974460012038957e-05,
+      "loss": 0.2304,
+      "step": 11429
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297331892714939e-05,
+      "loss": 0.2213,
+      "step": 11430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972177799788847e-05,
+      "loss": 0.2459,
+      "step": 11431
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297103662997363e-05,
+      "loss": 0.2365,
+      "step": 11432
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2969895417720035e-05,
+      "loss": 0.2331,
+      "step": 11433
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2968754163044369e-05,
+      "loss": 0.2289,
+      "step": 11434
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967612865962933e-05,
+      "loss": 0.248,
+      "step": 11435
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966471526492023e-05,
+      "loss": 0.2657,
+      "step": 11436
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965330144647947e-05,
+      "loss": 0.2342,
+      "step": 11437
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2964188720447012e-05,
+      "loss": 0.2531,
+      "step": 11438
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963047253905513e-05,
+      "loss": 0.2345,
+      "step": 11439
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2961905745039764e-05,
+      "loss": 0.2324,
+      "step": 11440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960764193866063e-05,
+      "loss": 0.2615,
+      "step": 11441
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959622600400718e-05,
+      "loss": 0.2235,
+      "step": 11442
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2958480964660038e-05,
+      "loss": 0.2284,
+      "step": 11443
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295733928666033e-05,
+      "loss": 0.2215,
+      "step": 11444
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2956197566417901e-05,
+      "loss": 0.2413,
+      "step": 11445
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2955055803949058e-05,
+      "loss": 0.2176,
+      "step": 11446
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295391399927011e-05,
+      "loss": 0.2569,
+      "step": 11447
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2952772152397366e-05,
+      "loss": 0.2206,
+      "step": 11448
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2951630263347136e-05,
+      "loss": 0.2503,
+      "step": 11449
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950488332135735e-05,
+      "loss": 0.2349,
+      "step": 11450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2949346358779476e-05,
+      "loss": 0.2341,
+      "step": 11451
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948204343294664e-05,
+      "loss": 0.2423,
+      "step": 11452
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294706228569761e-05,
+      "loss": 0.2685,
+      "step": 11453
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2945920186004634e-05,
+      "loss": 0.2347,
+      "step": 11454
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944778044232047e-05,
+      "loss": 0.2552,
+      "step": 11455
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943635860396166e-05,
+      "loss": 0.2192,
+      "step": 11456
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29424936345133e-05,
+      "loss": 0.2444,
+      "step": 11457
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2941351366599773e-05,
+      "loss": 0.2648,
+      "step": 11458
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940209056671891e-05,
+      "loss": 0.2692,
+      "step": 11459
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293906670474598e-05,
+      "loss": 0.2495,
+      "step": 11460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293792431083835e-05,
+      "loss": 0.2228,
+      "step": 11461
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936781874965325e-05,
+      "loss": 0.2195,
+      "step": 11462
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293563939714322e-05,
+      "loss": 0.2522,
+      "step": 11463
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934496877388359e-05,
+      "loss": 0.2169,
+      "step": 11464
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933354315717055e-05,
+      "loss": 0.2364,
+      "step": 11465
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293221171214563e-05,
+      "loss": 0.2451,
+      "step": 11466
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2931069066690406e-05,
+      "loss": 0.2306,
+      "step": 11467
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292992637936771e-05,
+      "loss": 0.2336,
+      "step": 11468
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928783650193855e-05,
+      "loss": 0.244,
+      "step": 11469
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927640879185168e-05,
+      "loss": 0.2621,
+      "step": 11470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926498066357976e-05,
+      "loss": 0.2206,
+      "step": 11471
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925355211728596e-05,
+      "loss": 0.2106,
+      "step": 11472
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292421231531336e-05,
+      "loss": 0.2314,
+      "step": 11473
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2923069377128586e-05,
+      "loss": 0.2322,
+      "step": 11474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29219263971906e-05,
+      "loss": 0.2382,
+      "step": 11475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920783375515735e-05,
+      "loss": 0.2333,
+      "step": 11476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2919640312120315e-05,
+      "loss": 0.234,
+      "step": 11477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2918497207020663e-05,
+      "loss": 0.2659,
+      "step": 11478
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917354060233116e-05,
+      "loss": 0.2358,
+      "step": 11479
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916210871773993e-05,
+      "loss": 0.259,
+      "step": 11480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915067641659629e-05,
+      "loss": 0.2442,
+      "step": 11481
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2913924369906351e-05,
+      "loss": 0.248,
+      "step": 11482
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912781056530495e-05,
+      "loss": 0.2488,
+      "step": 11483
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2911637701548387e-05,
+      "loss": 0.2641,
+      "step": 11484
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2910494304976362e-05,
+      "loss": 0.2728,
+      "step": 11485
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290935086683075e-05,
+      "loss": 0.2516,
+      "step": 11486
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908207387127883e-05,
+      "loss": 0.273,
+      "step": 11487
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907063865884095e-05,
+      "loss": 0.2536,
+      "step": 11488
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905920303115725e-05,
+      "loss": 0.2353,
+      "step": 11489
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2904776698839098e-05,
+      "loss": 0.2464,
+      "step": 11490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2903633053070561e-05,
+      "loss": 0.2247,
+      "step": 11491
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290248936582644e-05,
+      "loss": 0.2377,
+      "step": 11492
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2901345637123078e-05,
+      "loss": 0.2451,
+      "step": 11493
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2900201866976807e-05,
+      "loss": 0.2611,
+      "step": 11494
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899058055403968e-05,
+      "loss": 0.2696,
+      "step": 11495
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897914202420899e-05,
+      "loss": 0.2282,
+      "step": 11496
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896770308043937e-05,
+      "loss": 0.2447,
+      "step": 11497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2895626372289422e-05,
+      "loss": 0.2529,
+      "step": 11498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2894482395173695e-05,
+      "loss": 0.2278,
+      "step": 11499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2893338376713095e-05,
+      "loss": 0.2403,
+      "step": 11500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2892194316923969e-05,
+      "loss": 0.236,
+      "step": 11501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289105021582265e-05,
+      "loss": 0.2249,
+      "step": 11502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2889906073425484e-05,
+      "loss": 0.2367,
+      "step": 11503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888761889748818e-05,
+      "loss": 0.2304,
+      "step": 11504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2887617664808989e-05,
+      "loss": 0.2371,
+      "step": 11505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2886473398622349e-05,
+      "loss": 0.2896,
+      "step": 11506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2885329091205233e-05,
+      "loss": 0.2273,
+      "step": 11507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288418474257399e-05,
+      "loss": 0.2334,
+      "step": 11508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883040352744968e-05,
+      "loss": 0.2306,
+      "step": 11509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2881895921734517e-05,
+      "loss": 0.2571,
+      "step": 11510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2880751449558974e-05,
+      "loss": 0.2337,
+      "step": 11511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.28796069362347e-05,
+      "loss": 0.251,
+      "step": 11512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2878462381778027e-05,
+      "loss": 0.2499,
+      "step": 11513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877317786205317e-05,
+      "loss": 0.2569,
+      "step": 11514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876173149532913e-05,
+      "loss": 0.2604,
+      "step": 11515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2875028471777166e-05,
+      "loss": 0.2312,
+      "step": 11516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2873883752954432e-05,
+      "loss": 0.2355,
+      "step": 11517
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2872738993081058e-05,
+      "loss": 0.2227,
+      "step": 11518
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2871594192173391e-05,
+      "loss": 0.2354,
+      "step": 11519
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2870449350247789e-05,
+      "loss": 0.2292,
+      "step": 11520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2869304467320604e-05,
+      "loss": 0.2558,
+      "step": 11521
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868159543408191e-05,
+      "loss": 0.2558,
+      "step": 11522
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867014578526902e-05,
+      "loss": 0.2454,
+      "step": 11523
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2865869572693094e-05,
+      "loss": 0.2432,
+      "step": 11524
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864724525923117e-05,
+      "loss": 0.2287,
+      "step": 11525
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863579438233331e-05,
+      "loss": 0.2294,
+      "step": 11526
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862434309640092e-05,
+      "loss": 0.2091,
+      "step": 11527
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2861289140159761e-05,
+      "loss": 0.2298,
+      "step": 11528
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2860143929808686e-05,
+      "loss": 0.2423,
+      "step": 11529
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2858998678603235e-05,
+      "loss": 0.2207,
+      "step": 11530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285785338655976e-05,
+      "loss": 0.2335,
+      "step": 11531
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856708053694625e-05,
+      "loss": 0.2407,
+      "step": 11532
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2855562680024185e-05,
+      "loss": 0.2476,
+      "step": 11533
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854417265564807e-05,
+      "loss": 0.2266,
+      "step": 11534
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2853271810332846e-05,
+      "loss": 0.2484,
+      "step": 11535
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2852126314344667e-05,
+      "loss": 0.2483,
+      "step": 11536
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850980777616634e-05,
+      "loss": 0.224,
+      "step": 11537
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2849835200165104e-05,
+      "loss": 0.2231,
+      "step": 11538
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848689582006444e-05,
+      "loss": 0.2222,
+      "step": 11539
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284754392315702e-05,
+      "loss": 0.246,
+      "step": 11540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2846398223633194e-05,
+      "loss": 0.2284,
+      "step": 11541
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2845252483451332e-05,
+      "loss": 0.2157,
+      "step": 11542
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844106702627798e-05,
+      "loss": 0.2482,
+      "step": 11543
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842960881178962e-05,
+      "loss": 0.2283,
+      "step": 11544
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841815019121192e-05,
+      "loss": 0.2557,
+      "step": 11545
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840669116470848e-05,
+      "loss": 0.2419,
+      "step": 11546
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839523173244302e-05,
+      "loss": 0.2247,
+      "step": 11547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838377189457924e-05,
+      "loss": 0.2334,
+      "step": 11548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2837231165128083e-05,
+      "loss": 0.2479,
+      "step": 11549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283608510027115e-05,
+      "loss": 0.257,
+      "step": 11550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2834938994903494e-05,
+      "loss": 0.2362,
+      "step": 11551
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2833792849041485e-05,
+      "loss": 0.225,
+      "step": 11552
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832646662701494e-05,
+      "loss": 0.2463,
+      "step": 11553
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831500435899894e-05,
+      "loss": 0.2336,
+      "step": 11554
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830354168653061e-05,
+      "loss": 0.2313,
+      "step": 11555
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2829207860977366e-05,
+      "loss": 0.2648,
+      "step": 11556
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2828061512889183e-05,
+      "loss": 0.25,
+      "step": 11557
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826915124404881e-05,
+      "loss": 0.2249,
+      "step": 11558
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2825768695540843e-05,
+      "loss": 0.2134,
+      "step": 11559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824622226313442e-05,
+      "loss": 0.2406,
+      "step": 11560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2823475716739053e-05,
+      "loss": 0.2187,
+      "step": 11561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822329166834056e-05,
+      "loss": 0.25,
+      "step": 11562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2821182576614825e-05,
+      "loss": 0.2314,
+      "step": 11563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820035946097738e-05,
+      "loss": 0.2458,
+      "step": 11564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2818889275299174e-05,
+      "loss": 0.2077,
+      "step": 11565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2817742564235512e-05,
+      "loss": 0.2368,
+      "step": 11566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816595812923134e-05,
+      "loss": 0.2013,
+      "step": 11567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2815449021378417e-05,
+      "loss": 0.2347,
+      "step": 11568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2814302189617742e-05,
+      "loss": 0.2092,
+      "step": 11569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813155317657496e-05,
+      "loss": 0.2499,
+      "step": 11570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2812008405514051e-05,
+      "loss": 0.2328,
+      "step": 11571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28108614532038e-05,
+      "loss": 0.2436,
+      "step": 11572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280971446074312e-05,
+      "loss": 0.2408,
+      "step": 11573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808567428148393e-05,
+      "loss": 0.2219,
+      "step": 11574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807420355436006e-05,
+      "loss": 0.2424,
+      "step": 11575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280627324262235e-05,
+      "loss": 0.2168,
+      "step": 11576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2805126089723798e-05,
+      "loss": 0.2336,
+      "step": 11577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2803978896756745e-05,
+      "loss": 0.2385,
+      "step": 11578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2802831663737574e-05,
+      "loss": 0.2263,
+      "step": 11579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2801684390682674e-05,
+      "loss": 0.2448,
+      "step": 11580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800537077608428e-05,
+      "loss": 0.2534,
+      "step": 11581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279938972453123e-05,
+      "loss": 0.2658,
+      "step": 11582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798242331467469e-05,
+      "loss": 0.2521,
+      "step": 11583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279709489843353e-05,
+      "loss": 0.248,
+      "step": 11584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795947425445805e-05,
+      "loss": 0.2434,
+      "step": 11585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794799912520685e-05,
+      "loss": 0.2458,
+      "step": 11586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279365235967456e-05,
+      "loss": 0.2602,
+      "step": 11587
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2792504766923823e-05,
+      "loss": 0.2567,
+      "step": 11588
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791357134284865e-05,
+      "loss": 0.2394,
+      "step": 11589
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790209461774082e-05,
+      "loss": 0.2485,
+      "step": 11590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2789061749407861e-05,
+      "loss": 0.2446,
+      "step": 11591
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.27879139972026e-05,
+      "loss": 0.2341,
+      "step": 11592
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2786766205174695e-05,
+      "loss": 0.2346,
+      "step": 11593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278561837334054e-05,
+      "loss": 0.2357,
+      "step": 11594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2784470501716528e-05,
+      "loss": 0.2169,
+      "step": 11595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278332259031906e-05,
+      "loss": 0.2517,
+      "step": 11596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 0.2397,
+      "step": 11597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2781026648269332e-05,
+      "loss": 0.2403,
+      "step": 11598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779878617649868e-05,
+      "loss": 0.2329,
+      "step": 11599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277873054732254e-05,
+      "loss": 0.2237,
+      "step": 11600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2777582437303738e-05,
+      "loss": 0.2943,
+      "step": 11601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2776434287609871e-05,
+      "loss": 0.2327,
+      "step": 11602
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277528609825733e-05,
+      "loss": 0.2322,
+      "step": 11603
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2774137869262524e-05,
+      "loss": 0.2427,
+      "step": 11604
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772989600641847e-05,
+      "loss": 0.2323,
+      "step": 11605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277184129241171e-05,
+      "loss": 0.2198,
+      "step": 11606
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2770692944588508e-05,
+      "loss": 0.2419,
+      "step": 11607
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2769544557188644e-05,
+      "loss": 0.2316,
+      "step": 11608
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276839613022853e-05,
+      "loss": 0.2277,
+      "step": 11609
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2767247663724557e-05,
+      "loss": 0.2334,
+      "step": 11610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2766099157693144e-05,
+      "loss": 0.2266,
+      "step": 11611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2764950612150684e-05,
+      "loss": 0.253,
+      "step": 11612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2763802027113587e-05,
+      "loss": 0.2375,
+      "step": 11613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2762653402598262e-05,
+      "loss": 0.2552,
+      "step": 11614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2761504738621113e-05,
+      "loss": 0.2346,
+      "step": 11615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2760356035198553e-05,
+      "loss": 0.2397,
+      "step": 11616
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759207292346987e-05,
+      "loss": 0.2341,
+      "step": 11617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758058510082818e-05,
+      "loss": 0.211,
+      "step": 11618
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2756909688422461e-05,
+      "loss": 0.2605,
+      "step": 11619
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755760827382325e-05,
+      "loss": 0.267,
+      "step": 11620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2754611926978817e-05,
+      "loss": 0.2291,
+      "step": 11621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2753462987228357e-05,
+      "loss": 0.2402,
+      "step": 11622
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275231400814735e-05,
+      "loss": 0.2312,
+      "step": 11623
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751164989752204e-05,
+      "loss": 0.2374,
+      "step": 11624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2750015932059339e-05,
+      "loss": 0.2363,
+      "step": 11625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2748866835085166e-05,
+      "loss": 0.2444,
+      "step": 11626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2747717698846093e-05,
+      "loss": 0.2405,
+      "step": 11627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2746568523358546e-05,
+      "loss": 0.2243,
+      "step": 11628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745419308638933e-05,
+      "loss": 0.2483,
+      "step": 11629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744270054703667e-05,
+      "loss": 0.2489,
+      "step": 11630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2743120761569166e-05,
+      "loss": 0.2337,
+      "step": 11631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2741971429251847e-05,
+      "loss": 0.2437,
+      "step": 11632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274082205776813e-05,
+      "loss": 0.2454,
+      "step": 11633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2739672647134426e-05,
+      "loss": 0.2365,
+      "step": 11634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2738523197367164e-05,
+      "loss": 0.2278,
+      "step": 11635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273737370848275e-05,
+      "loss": 0.2311,
+      "step": 11636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273622418049761e-05,
+      "loss": 0.2302,
+      "step": 11637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735074613428162e-05,
+      "loss": 0.2289,
+      "step": 11638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733925007290831e-05,
+      "loss": 0.2467,
+      "step": 11639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732775362102032e-05,
+      "loss": 0.2291,
+      "step": 11640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731625677878193e-05,
+      "loss": 0.2539,
+      "step": 11641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2730475954635728e-05,
+      "loss": 0.2159,
+      "step": 11642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2729326192391065e-05,
+      "loss": 0.2542,
+      "step": 11643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2728176391160627e-05,
+      "loss": 0.2231,
+      "step": 11644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2727026550960839e-05,
+      "loss": 0.2353,
+      "step": 11645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272587667180812e-05,
+      "loss": 0.2461,
+      "step": 11646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2724726753718897e-05,
+      "loss": 0.2069,
+      "step": 11647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27235767967096e-05,
+      "loss": 0.2326,
+      "step": 11648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2722426800796652e-05,
+      "loss": 0.2409,
+      "step": 11649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272127676599648e-05,
+      "loss": 0.2584,
+      "step": 11650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2720126692325507e-05,
+      "loss": 0.2341,
+      "step": 11651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2718976579800167e-05,
+      "loss": 0.263,
+      "step": 11652
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717826428436881e-05,
+      "loss": 0.2134,
+      "step": 11653
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2716676238252086e-05,
+      "loss": 0.2342,
+      "step": 11654
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2715526009262209e-05,
+      "loss": 0.2269,
+      "step": 11655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2714375741483678e-05,
+      "loss": 0.2493,
+      "step": 11656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2713225434932921e-05,
+      "loss": 0.2568,
+      "step": 11657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712075089626372e-05,
+      "loss": 0.24,
+      "step": 11658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2710924705580462e-05,
+      "loss": 0.2179,
+      "step": 11659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709774282811625e-05,
+      "loss": 0.2483,
+      "step": 11660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2708623821336294e-05,
+      "loss": 0.2125,
+      "step": 11661
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27074733211709e-05,
+      "loss": 0.2291,
+      "step": 11662
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2706322782331877e-05,
+      "loss": 0.2315,
+      "step": 11663
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2705172204835658e-05,
+      "loss": 0.2107,
+      "step": 11664
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270402158869868e-05,
+      "loss": 0.2487,
+      "step": 11665
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2702870933937379e-05,
+      "loss": 0.2438,
+      "step": 11666
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270172024056819e-05,
+      "loss": 0.2325,
+      "step": 11667
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270056950860755e-05,
+      "loss": 0.2389,
+      "step": 11668
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2699418738071891e-05,
+      "loss": 0.2526,
+      "step": 11669
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2698267928977659e-05,
+      "loss": 0.2578,
+      "step": 11670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697117081341288e-05,
+      "loss": 0.2314,
+      "step": 11671
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695966195179217e-05,
+      "loss": 0.2512,
+      "step": 11672
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2694815270507885e-05,
+      "loss": 0.2203,
+      "step": 11673
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693664307343731e-05,
+      "loss": 0.221,
+      "step": 11674
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.26925133057032e-05,
+      "loss": 0.2231,
+      "step": 11675
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2691362265602727e-05,
+      "loss": 0.2388,
+      "step": 11676
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2690211187058757e-05,
+      "loss": 0.253,
+      "step": 11677
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268906007008773e-05,
+      "loss": 0.2291,
+      "step": 11678
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268790891470609e-05,
+      "loss": 0.2388,
+      "step": 11679
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2686757720930281e-05,
+      "loss": 0.2193,
+      "step": 11680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2685606488776748e-05,
+      "loss": 0.2558,
+      "step": 11681
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684455218261929e-05,
+      "loss": 0.2671,
+      "step": 11682
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2683303909402274e-05,
+      "loss": 0.2822,
+      "step": 11683
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2682152562214225e-05,
+      "loss": 0.2439,
+      "step": 11684
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268100117671423e-05,
+      "loss": 0.2464,
+      "step": 11685
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2679849752918734e-05,
+      "loss": 0.2386,
+      "step": 11686
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2678698290844188e-05,
+      "loss": 0.2483,
+      "step": 11687
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2677546790507036e-05,
+      "loss": 0.2376,
+      "step": 11688
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676395251923727e-05,
+      "loss": 0.2344,
+      "step": 11689
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2675243675110708e-05,
+      "loss": 0.2574,
+      "step": 11690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267409206008443e-05,
+      "loss": 0.2518,
+      "step": 11691
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2672940406861342e-05,
+      "loss": 0.2555,
+      "step": 11692
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2671788715457894e-05,
+      "loss": 0.2188,
+      "step": 11693
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2670636985890542e-05,
+      "loss": 0.2347,
+      "step": 11694
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669485218175728e-05,
+      "loss": 0.2269,
+      "step": 11695
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.266833341232991e-05,
+      "loss": 0.2311,
+      "step": 11696
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667181568369536e-05,
+      "loss": 0.2318,
+      "step": 11697
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2666029686311063e-05,
+      "loss": 0.2358,
+      "step": 11698
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664877766170947e-05,
+      "loss": 0.2355,
+      "step": 11699
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663725807965635e-05,
+      "loss": 0.2097,
+      "step": 11700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2662573811711588e-05,
+      "loss": 0.2579,
+      "step": 11701
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661421777425253e-05,
+      "loss": 0.2387,
+      "step": 11702
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660269705123094e-05,
+      "loss": 0.2542,
+      "step": 11703
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2659117594821564e-05,
+      "loss": 0.2363,
+      "step": 11704
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657965446537119e-05,
+      "loss": 0.2196,
+      "step": 11705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656813260286217e-05,
+      "loss": 0.2385,
+      "step": 11706
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2655661036085318e-05,
+      "loss": 0.2264,
+      "step": 11707
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654508773950878e-05,
+      "loss": 0.2498,
+      "step": 11708
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653356473899357e-05,
+      "loss": 0.2555,
+      "step": 11709
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265220413594721e-05,
+      "loss": 0.2506,
+      "step": 11710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2651051760110906e-05,
+      "loss": 0.2487,
+      "step": 11711
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2649899346406898e-05,
+      "loss": 0.2462,
+      "step": 11712
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2648746894851649e-05,
+      "loss": 0.2392,
+      "step": 11713
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647594405461625e-05,
+      "loss": 0.2466,
+      "step": 11714
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264644187825328e-05,
+      "loss": 0.2405,
+      "step": 11715
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2645289313243086e-05,
+      "loss": 0.2126,
+      "step": 11716
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2644136710447497e-05,
+      "loss": 0.2588,
+      "step": 11717
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642984069882981e-05,
+      "loss": 0.2203,
+      "step": 11718
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641831391566004e-05,
+      "loss": 0.2403,
+      "step": 11719
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264067867551303e-05,
+      "loss": 0.2532,
+      "step": 11720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2639525921740523e-05,
+      "loss": 0.2416,
+      "step": 11721
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2638373130264953e-05,
+      "loss": 0.2357,
+      "step": 11722
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637220301102782e-05,
+      "loss": 0.2297,
+      "step": 11723
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2636067434270476e-05,
+      "loss": 0.2577,
+      "step": 11724
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634914529784505e-05,
+      "loss": 0.2457,
+      "step": 11725
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263376158766134e-05,
+      "loss": 0.238,
+      "step": 11726
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2632608607917447e-05,
+      "loss": 0.2439,
+      "step": 11727
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2631455590569295e-05,
+      "loss": 0.2566,
+      "step": 11728
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2630302535633352e-05,
+      "loss": 0.2227,
+      "step": 11729
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629149443126089e-05,
+      "loss": 0.2654,
+      "step": 11730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262799631306398e-05,
+      "loss": 0.226,
+      "step": 11731
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2626843145463492e-05,
+      "loss": 0.2421,
+      "step": 11732
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2625689940341102e-05,
+      "loss": 0.2176,
+      "step": 11733
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2624536697713281e-05,
+      "loss": 0.2348,
+      "step": 11734
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2623383417596498e-05,
+      "loss": 0.2481,
+      "step": 11735
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622230100007226e-05,
+      "loss": 0.2559,
+      "step": 11736
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621076744961944e-05,
+      "loss": 0.2282,
+      "step": 11737
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2619923352477125e-05,
+      "loss": 0.2491,
+      "step": 11738
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618769922569245e-05,
+      "loss": 0.2322,
+      "step": 11739
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2617616455254776e-05,
+      "loss": 0.2641,
+      "step": 11740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616462950550199e-05,
+      "loss": 0.2465,
+      "step": 11741
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615309408471985e-05,
+      "loss": 0.236,
+      "step": 11742
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614155829036616e-05,
+      "loss": 0.2218,
+      "step": 11743
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261300221226057e-05,
+      "loss": 0.2241,
+      "step": 11744
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261184855816032e-05,
+      "loss": 0.2751,
+      "step": 11745
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2610694866752348e-05,
+      "loss": 0.2258,
+      "step": 11746
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609541138053137e-05,
+      "loss": 0.2661,
+      "step": 11747
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260838737207916e-05,
+      "loss": 0.2357,
+      "step": 11748
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607233568846902e-05,
+      "loss": 0.25,
+      "step": 11749
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606079728372844e-05,
+      "loss": 0.2433,
+      "step": 11750
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2604925850673466e-05,
+      "loss": 0.235,
+      "step": 11751
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2603771935765248e-05,
+      "loss": 0.2545,
+      "step": 11752
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2602617983664676e-05,
+      "loss": 0.2375,
+      "step": 11753
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601463994388235e-05,
+      "loss": 0.2334,
+      "step": 11754
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2600309967952404e-05,
+      "loss": 0.2325,
+      "step": 11755
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599155904373669e-05,
+      "loss": 0.238,
+      "step": 11756
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259800180366851e-05,
+      "loss": 0.2636,
+      "step": 11757
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2596847665853423e-05,
+      "loss": 0.2272,
+      "step": 11758
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2595693490944883e-05,
+      "loss": 0.2475,
+      "step": 11759
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2594539278959384e-05,
+      "loss": 0.2231,
+      "step": 11760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259338502991341e-05,
+      "loss": 0.2213,
+      "step": 11761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592230743823444e-05,
+      "loss": 0.2232,
+      "step": 11762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2591076420705979e-05,
+      "loss": 0.2248,
+      "step": 11763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2589922060577502e-05,
+      "loss": 0.24,
+      "step": 11764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588767663454502e-05,
+      "loss": 0.2378,
+      "step": 11765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587613229353468e-05,
+      "loss": 0.2339,
+      "step": 11766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2586458758290893e-05,
+      "loss": 0.2251,
+      "step": 11767
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2585304250283263e-05,
+      "loss": 0.2433,
+      "step": 11768
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258414970534707e-05,
+      "loss": 0.2825,
+      "step": 11769
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2582995123498804e-05,
+      "loss": 0.2397,
+      "step": 11770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581840504754965e-05,
+      "loss": 0.2358,
+      "step": 11771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258068584913204e-05,
+      "loss": 0.2644,
+      "step": 11772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2579531156646518e-05,
+      "loss": 0.2479,
+      "step": 11773
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.25783764273149e-05,
+      "loss": 0.2361,
+      "step": 11774
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577221661153676e-05,
+      "loss": 0.2252,
+      "step": 11775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257606685817934e-05,
+      "loss": 0.2153,
+      "step": 11776
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2574912018408393e-05,
+      "loss": 0.2615,
+      "step": 11777
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573757141857323e-05,
+      "loss": 0.2482,
+      "step": 11778
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2572602228542631e-05,
+      "loss": 0.2295,
+      "step": 11779
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571447278480818e-05,
+      "loss": 0.245,
+      "step": 11780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257029229168837e-05,
+      "loss": 0.1973,
+      "step": 11781
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2569137268181792e-05,
+      "loss": 0.2088,
+      "step": 11782
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567982207977585e-05,
+      "loss": 0.2473,
+      "step": 11783
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.256682711109224e-05,
+      "loss": 0.2461,
+      "step": 11784
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2565671977542264e-05,
+      "loss": 0.244,
+      "step": 11785
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2564516807344151e-05,
+      "loss": 0.2579,
+      "step": 11786
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563361600514409e-05,
+      "loss": 0.2516,
+      "step": 11787
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2562206357069534e-05,
+      "loss": 0.2405,
+      "step": 11788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2561051077026024e-05,
+      "loss": 0.2283,
+      "step": 11789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2559895760400385e-05,
+      "loss": 0.2624,
+      "step": 11790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2558740407209121e-05,
+      "loss": 0.2346,
+      "step": 11791
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557585017468732e-05,
+      "loss": 0.2303,
+      "step": 11792
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556429591195727e-05,
+      "loss": 0.2304,
+      "step": 11793
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2555274128406606e-05,
+      "loss": 0.2287,
+      "step": 11794
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255411862911787e-05,
+      "loss": 0.2408,
+      "step": 11795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255296309334603e-05,
+      "loss": 0.2211,
+      "step": 11796
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551807521107588e-05,
+      "loss": 0.2196,
+      "step": 11797
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2550651912419057e-05,
+      "loss": 0.2506,
+      "step": 11798
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549496267296937e-05,
+      "loss": 0.2502,
+      "step": 11799
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548340585757736e-05,
+      "loss": 0.2357,
+      "step": 11800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2547184867817964e-05,
+      "loss": 0.2211,
+      "step": 11801
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546029113494127e-05,
+      "loss": 0.2388,
+      "step": 11802
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2544873322802735e-05,
+      "loss": 0.2421,
+      "step": 11803
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25437174957603e-05,
+      "loss": 0.2358,
+      "step": 11804
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542561632383325e-05,
+      "loss": 0.2386,
+      "step": 11805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2541405732688328e-05,
+      "loss": 0.2364,
+      "step": 11806
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540249796691817e-05,
+      "loss": 0.2275,
+      "step": 11807
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2539093824410302e-05,
+      "loss": 0.237,
+      "step": 11808
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2537937815860295e-05,
+      "loss": 0.224,
+      "step": 11809
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536781771058311e-05,
+      "loss": 0.2267,
+      "step": 11810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2535625690020861e-05,
+      "loss": 0.2478,
+      "step": 11811
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2534469572764457e-05,
+      "loss": 0.2511,
+      "step": 11812
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2533313419305617e-05,
+      "loss": 0.2418,
+      "step": 11813
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532157229660853e-05,
+      "loss": 0.2129,
+      "step": 11814
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531001003846682e-05,
+      "loss": 0.2125,
+      "step": 11815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2529844741879616e-05,
+      "loss": 0.2463,
+      "step": 11816
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528688443776172e-05,
+      "loss": 0.2253,
+      "step": 11817
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2527532109552869e-05,
+      "loss": 0.2162,
+      "step": 11818
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2526375739226225e-05,
+      "loss": 0.2162,
+      "step": 11819
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2525219332812752e-05,
+      "loss": 0.2449,
+      "step": 11820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2524062890328974e-05,
+      "loss": 0.2508,
+      "step": 11821
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522906411791406e-05,
+      "loss": 0.2737,
+      "step": 11822
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252174989721657e-05,
+      "loss": 0.243,
+      "step": 11823
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2520593346620982e-05,
+      "loss": 0.2333,
+      "step": 11824
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519436760021163e-05,
+      "loss": 0.2077,
+      "step": 11825
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251828013743364e-05,
+      "loss": 0.2519,
+      "step": 11826
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2517123478874929e-05,
+      "loss": 0.2256,
+      "step": 11827
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515966784361549e-05,
+      "loss": 0.2536,
+      "step": 11828
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514810053910025e-05,
+      "loss": 0.2314,
+      "step": 11829
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513653287536882e-05,
+      "loss": 0.2658,
+      "step": 11830
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512496485258638e-05,
+      "loss": 0.2363,
+      "step": 11831
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511339647091825e-05,
+      "loss": 0.2214,
+      "step": 11832
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2510182773052961e-05,
+      "loss": 0.2374,
+      "step": 11833
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509025863158573e-05,
+      "loss": 0.2164,
+      "step": 11834
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2507868917425182e-05,
+      "loss": 0.2628,
+      "step": 11835
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506711935869317e-05,
+      "loss": 0.2278,
+      "step": 11836
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2505554918507507e-05,
+      "loss": 0.2206,
+      "step": 11837
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250439786535628e-05,
+      "loss": 0.2372,
+      "step": 11838
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2503240776432155e-05,
+      "loss": 0.246,
+      "step": 11839
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502083651751668e-05,
+      "loss": 0.23,
+      "step": 11840
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500926491331345e-05,
+      "loss": 0.2242,
+      "step": 11841
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499769295187711e-05,
+      "loss": 0.2524,
+      "step": 11842
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24986120633373e-05,
+      "loss": 0.2311,
+      "step": 11843
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2497454795796642e-05,
+      "loss": 0.2263,
+      "step": 11844
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496297492582264e-05,
+      "loss": 0.2741,
+      "step": 11845
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24951401537107e-05,
+      "loss": 0.2253,
+      "step": 11846
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493982779198481e-05,
+      "loss": 0.2375,
+      "step": 11847
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492825369062138e-05,
+      "loss": 0.2397,
+      "step": 11848
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2491667923318204e-05,
+      "loss": 0.2189,
+      "step": 11849
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490510441983212e-05,
+      "loss": 0.2076,
+      "step": 11850
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2489352925073694e-05,
+      "loss": 0.237,
+      "step": 11851
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2488195372606187e-05,
+      "loss": 0.2365,
+      "step": 11852
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487037784597224e-05,
+      "loss": 0.2418,
+      "step": 11853
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485880161063339e-05,
+      "loss": 0.2567,
+      "step": 11854
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2484722502021072e-05,
+      "loss": 0.2547,
+      "step": 11855
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483564807486953e-05,
+      "loss": 0.2566,
+      "step": 11856
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2482407077477521e-05,
+      "loss": 0.2309,
+      "step": 11857
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2481249312009312e-05,
+      "loss": 0.2347,
+      "step": 11858
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480091511098867e-05,
+      "loss": 0.2474,
+      "step": 11859
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2478933674762721e-05,
+      "loss": 0.248,
+      "step": 11860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477775803017411e-05,
+      "loss": 0.2162,
+      "step": 11861
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476617895879479e-05,
+      "loss": 0.2147,
+      "step": 11862
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475459953365466e-05,
+      "loss": 0.2296,
+      "step": 11863
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2474301975491909e-05,
+      "loss": 0.2336,
+      "step": 11864
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2473143962275351e-05,
+      "loss": 0.2474,
+      "step": 11865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2471985913732332e-05,
+      "loss": 0.2526,
+      "step": 11866
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247082782987939e-05,
+      "loss": 0.2342,
+      "step": 11867
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469669710733073e-05,
+      "loss": 0.2227,
+      "step": 11868
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246851155630992e-05,
+      "loss": 0.2265,
+      "step": 11869
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2467353366626473e-05,
+      "loss": 0.2154,
+      "step": 11870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2466195141699282e-05,
+      "loss": 0.2299,
+      "step": 11871
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2465036881544888e-05,
+      "loss": 0.2601,
+      "step": 11872
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246387858617983e-05,
+      "loss": 0.2254,
+      "step": 11873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2462720255620659e-05,
+      "loss": 0.2351,
+      "step": 11874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246156188988392e-05,
+      "loss": 0.2682,
+      "step": 11875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2460403488986159e-05,
+      "loss": 0.2181,
+      "step": 11876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.245924505294392e-05,
+      "loss": 0.2372,
+      "step": 11877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458086581773752e-05,
+      "loss": 0.2596,
+      "step": 11878
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2456928075492207e-05,
+      "loss": 0.2458,
+      "step": 11879
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455769534115824e-05,
+      "loss": 0.2292,
+      "step": 11880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.245461095766116e-05,
+      "loss": 0.2285,
+      "step": 11881
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453452346144758e-05,
+      "loss": 0.2283,
+      "step": 11882
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2452293699583173e-05,
+      "loss": 0.232,
+      "step": 11883
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451135017992951e-05,
+      "loss": 0.2548,
+      "step": 11884
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2449976301390642e-05,
+      "loss": 0.2323,
+      "step": 11885
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448817549792804e-05,
+      "loss": 0.2523,
+      "step": 11886
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2447658763215982e-05,
+      "loss": 0.2422,
+      "step": 11887
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244649994167673e-05,
+      "loss": 0.2484,
+      "step": 11888
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.24453410851916e-05,
+      "loss": 0.2379,
+      "step": 11889
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444182193777146e-05,
+      "loss": 0.2443,
+      "step": 11890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244302326744992e-05,
+      "loss": 0.2533,
+      "step": 11891
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441864306226482e-05,
+      "loss": 0.2373,
+      "step": 11892
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244070531012338e-05,
+      "loss": 0.2254,
+      "step": 11893
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2439546279157171e-05,
+      "loss": 0.2173,
+      "step": 11894
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243838721334441e-05,
+      "loss": 0.2382,
+      "step": 11895
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2437228112701654e-05,
+      "loss": 0.2449,
+      "step": 11896
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243606897724546e-05,
+      "loss": 0.2529,
+      "step": 11897
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434909806992388e-05,
+      "loss": 0.2519,
+      "step": 11898
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2433750601958993e-05,
+      "loss": 0.2344,
+      "step": 11899
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243259136216183e-05,
+      "loss": 0.231,
+      "step": 11900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243143208761746e-05,
+      "loss": 0.2214,
+      "step": 11901
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243027277834244e-05,
+      "loss": 0.2533,
+      "step": 11902
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2429113434353335e-05,
+      "loss": 0.2359,
+      "step": 11903
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2427954055666705e-05,
+      "loss": 0.2388,
+      "step": 11904
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2426794642299105e-05,
+      "loss": 0.2346,
+      "step": 11905
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425635194267099e-05,
+      "loss": 0.2215,
+      "step": 11906
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424475711587245e-05,
+      "loss": 0.2537,
+      "step": 11907
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423316194276111e-05,
+      "loss": 0.2429,
+      "step": 11908
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242215664235026e-05,
+      "loss": 0.2253,
+      "step": 11909
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2420997055826248e-05,
+      "loss": 0.2233,
+      "step": 11910
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419837434720643e-05,
+      "loss": 0.2505,
+      "step": 11911
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418677779050012e-05,
+      "loss": 0.2173,
+      "step": 11912
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417518088830914e-05,
+      "loss": 0.2412,
+      "step": 11913
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2416358364079915e-05,
+      "loss": 0.2405,
+      "step": 11914
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415198604813585e-05,
+      "loss": 0.2387,
+      "step": 11915
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414038811048486e-05,
+      "loss": 0.2408,
+      "step": 11916
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2412878982801184e-05,
+      "loss": 0.1942,
+      "step": 11917
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411719120088246e-05,
+      "loss": 0.2115,
+      "step": 11918
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410559222926247e-05,
+      "loss": 0.2121,
+      "step": 11919
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409399291331743e-05,
+      "loss": 0.2303,
+      "step": 11920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2408239325321314e-05,
+      "loss": 0.2413,
+      "step": 11921
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240707932491152e-05,
+      "loss": 0.2464,
+      "step": 11922
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2405919290118933e-05,
+      "loss": 0.2225,
+      "step": 11923
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2404759220960123e-05,
+      "loss": 0.2646,
+      "step": 11924
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2403599117451665e-05,
+      "loss": 0.229,
+      "step": 11925
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402438979610125e-05,
+      "loss": 0.2318,
+      "step": 11926
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2401278807452078e-05,
+      "loss": 0.2328,
+      "step": 11927
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240011860099409e-05,
+      "loss": 0.2307,
+      "step": 11928
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2398958360252738e-05,
+      "loss": 0.2476,
+      "step": 11929
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397798085244594e-05,
+      "loss": 0.2564,
+      "step": 11930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396637775986235e-05,
+      "loss": 0.2513,
+      "step": 11931
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395477432494229e-05,
+      "loss": 0.2344,
+      "step": 11932
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394317054785152e-05,
+      "loss": 0.2603,
+      "step": 11933
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.2595,
+      "step": 11934
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2391996196782086e-05,
+      "loss": 0.2216,
+      "step": 11935
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239083571652125e-05,
+      "loss": 0.2039,
+      "step": 11936
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2389675202109648e-05,
+      "loss": 0.2157,
+      "step": 11937
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388514653563854e-05,
+      "loss": 0.2183,
+      "step": 11938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2387354070900444e-05,
+      "loss": 0.2637,
+      "step": 11939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386193454135999e-05,
+      "loss": 0.2271,
+      "step": 11940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2385032803287096e-05,
+      "loss": 0.2615,
+      "step": 11941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383872118370312e-05,
+      "loss": 0.2298,
+      "step": 11942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2382711399402234e-05,
+      "loss": 0.2478,
+      "step": 11943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2381550646399431e-05,
+      "loss": 0.2125,
+      "step": 11944
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380389859378494e-05,
+      "loss": 0.2074,
+      "step": 11945
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2379229038355994e-05,
+      "loss": 0.2282,
+      "step": 11946
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2378068183348515e-05,
+      "loss": 0.2301,
+      "step": 11947
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376907294372644e-05,
+      "loss": 0.2349,
+      "step": 11948
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2375746371444957e-05,
+      "loss": 0.2118,
+      "step": 11949
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2374585414582039e-05,
+      "loss": 0.206,
+      "step": 11950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373424423800473e-05,
+      "loss": 0.2438,
+      "step": 11951
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2372263399116841e-05,
+      "loss": 0.244,
+      "step": 11952
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371102340547731e-05,
+      "loss": 0.2256,
+      "step": 11953
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2369941248109725e-05,
+      "loss": 0.217,
+      "step": 11954
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368780121819409e-05,
+      "loss": 0.2291,
+      "step": 11955
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2367618961693364e-05,
+      "loss": 0.2376,
+      "step": 11956
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2366457767748183e-05,
+      "loss": 0.2728,
+      "step": 11957
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236529654000045e-05,
+      "loss": 0.2584,
+      "step": 11958
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2364135278466748e-05,
+      "loss": 0.2617,
+      "step": 11959
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2362973983163676e-05,
+      "loss": 0.2381,
+      "step": 11960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361812654107806e-05,
+      "loss": 0.2521,
+      "step": 11961
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360651291315735e-05,
+      "loss": 0.2268,
+      "step": 11962
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2359489894804053e-05,
+      "loss": 0.2195,
+      "step": 11963
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235832846458935e-05,
+      "loss": 0.2433,
+      "step": 11964
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357167000688213e-05,
+      "loss": 0.253,
+      "step": 11965
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2356005503117232e-05,
+      "loss": 0.2432,
+      "step": 11966
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2354843971892998e-05,
+      "loss": 0.2169,
+      "step": 11967
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2353682407032104e-05,
+      "loss": 0.2536,
+      "step": 11968
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.235252080855114e-05,
+      "loss": 0.2307,
+      "step": 11969
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2351359176466702e-05,
+      "loss": 0.2537,
+      "step": 11970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2350197510795381e-05,
+      "loss": 0.2332,
+      "step": 11971
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2349035811553767e-05,
+      "loss": 0.2593,
+      "step": 11972
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2347874078758455e-05,
+      "loss": 0.2502,
+      "step": 11973
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346712312426042e-05,
+      "loss": 0.2461,
+      "step": 11974
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2345550512573122e-05,
+      "loss": 0.2597,
+      "step": 11975
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344388679216288e-05,
+      "loss": 0.2221,
+      "step": 11976
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2343226812372137e-05,
+      "loss": 0.2451,
+      "step": 11977
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342064912057268e-05,
+      "loss": 0.2355,
+      "step": 11978
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2340902978288271e-05,
+      "loss": 0.228,
+      "step": 11979
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2339741011081748e-05,
+      "loss": 0.2224,
+      "step": 11980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338579010454297e-05,
+      "loss": 0.235,
+      "step": 11981
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337416976422514e-05,
+      "loss": 0.227,
+      "step": 11982
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2336254909002996e-05,
+      "loss": 0.22,
+      "step": 11983
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2335092808212346e-05,
+      "loss": 0.2247,
+      "step": 11984
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233393067406716e-05,
+      "loss": 0.2509,
+      "step": 11985
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332768506584038e-05,
+      "loss": 0.21,
+      "step": 11986
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2331606305779587e-05,
+      "loss": 0.2751,
+      "step": 11987
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.23304440716704e-05,
+      "loss": 0.2392,
+      "step": 11988
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2329281804273081e-05,
+      "loss": 0.234,
+      "step": 11989
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232811950360423e-05,
+      "loss": 0.2488,
+      "step": 11990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326957169680455e-05,
+      "loss": 0.2363,
+      "step": 11991
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325794802518354e-05,
+      "loss": 0.2436,
+      "step": 11992
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2324632402134533e-05,
+      "loss": 0.249,
+      "step": 11993
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323469968545592e-05,
+      "loss": 0.2583,
+      "step": 11994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322307501768137e-05,
+      "loss": 0.2241,
+      "step": 11995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2321145001818772e-05,
+      "loss": 0.2469,
+      "step": 11996
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319982468714107e-05,
+      "loss": 0.2307,
+      "step": 11997
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2318819902470744e-05,
+      "loss": 0.247,
+      "step": 11998
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317657303105288e-05,
+      "loss": 0.2282,
+      "step": 11999
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2316494670634345e-05,
+      "loss": 0.2518,
+      "step": 12000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315332005074525e-05,
+      "loss": 0.2265,
+      "step": 12001
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2314169306442434e-05,
+      "loss": 0.2129,
+      "step": 12002
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2313006574754683e-05,
+      "loss": 0.2397,
+      "step": 12003
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311843810027877e-05,
+      "loss": 0.2732,
+      "step": 12004
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310681012278625e-05,
+      "loss": 0.2328,
+      "step": 12005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309518181523537e-05,
+      "loss": 0.2616,
+      "step": 12006
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2308355317779222e-05,
+      "loss": 0.2583,
+      "step": 12007
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.230719242106229e-05,
+      "loss": 0.2333,
+      "step": 12008
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306029491389359e-05,
+      "loss": 0.2166,
+      "step": 12009
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304866528777031e-05,
+      "loss": 0.2207,
+      "step": 12010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303703533241925e-05,
+      "loss": 0.2381,
+      "step": 12011
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302540504800646e-05,
+      "loss": 0.2062,
+      "step": 12012
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2301377443469813e-05,
+      "loss": 0.2304,
+      "step": 12013
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2300214349266036e-05,
+      "loss": 0.2313,
+      "step": 12014
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299051222205929e-05,
+      "loss": 0.2498,
+      "step": 12015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2297888062306107e-05,
+      "loss": 0.2487,
+      "step": 12016
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2296724869583185e-05,
+      "loss": 0.2331,
+      "step": 12017
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295561644053775e-05,
+      "loss": 0.2396,
+      "step": 12018
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2294398385734496e-05,
+      "loss": 0.227,
+      "step": 12019
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293235094641964e-05,
+      "loss": 0.249,
+      "step": 12020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2292071770792794e-05,
+      "loss": 0.2443,
+      "step": 12021
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2290908414203602e-05,
+      "loss": 0.2432,
+      "step": 12022
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2289745024891007e-05,
+      "loss": 0.2164,
+      "step": 12023
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2288581602871627e-05,
+      "loss": 0.2182,
+      "step": 12024
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287418148162078e-05,
+      "loss": 0.2383,
+      "step": 12025
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2286254660778985e-05,
+      "loss": 0.2602,
+      "step": 12026
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.228509114073896e-05,
+      "loss": 0.2465,
+      "step": 12027
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283927588058626e-05,
+      "loss": 0.2163,
+      "step": 12028
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.22827640027546e-05,
+      "loss": 0.2551,
+      "step": 12029
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281600384843514e-05,
+      "loss": 0.2203,
+      "step": 12030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280436734341973e-05,
+      "loss": 0.2202,
+      "step": 12031
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279273051266612e-05,
+      "loss": 0.2134,
+      "step": 12032
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2278109335634044e-05,
+      "loss": 0.2453,
+      "step": 12033
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2276945587460896e-05,
+      "loss": 0.1927,
+      "step": 12034
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227578180676379e-05,
+      "loss": 0.2196,
+      "step": 12035
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227461799355935e-05,
+      "loss": 0.2511,
+      "step": 12036
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273454147864202e-05,
+      "loss": 0.2181,
+      "step": 12037
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2272290269694962e-05,
+      "loss": 0.227,
+      "step": 12038
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271126359068264e-05,
+      "loss": 0.2547,
+      "step": 12039
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269962416000729e-05,
+      "loss": 0.2278,
+      "step": 12040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2268798440508984e-05,
+      "loss": 0.2646,
+      "step": 12041
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226763443260966e-05,
+      "loss": 0.2272,
+      "step": 12042
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266470392319374e-05,
+      "loss": 0.254,
+      "step": 12043
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226530631965476e-05,
+      "loss": 0.2108,
+      "step": 12044
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264142214632441e-05,
+      "loss": 0.2365,
+      "step": 12045
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226297807726905e-05,
+      "loss": 0.2187,
+      "step": 12046
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261813907581213e-05,
+      "loss": 0.221,
+      "step": 12047
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260649705585562e-05,
+      "loss": 0.2193,
+      "step": 12048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225948547129872e-05,
+      "loss": 0.2458,
+      "step": 12049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2258321204737325e-05,
+      "loss": 0.2304,
+      "step": 12050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2257156905918e-05,
+      "loss": 0.253,
+      "step": 12051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225599257485738e-05,
+      "loss": 0.2077,
+      "step": 12052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254828211572096e-05,
+      "loss": 0.2276,
+      "step": 12053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225366381607878e-05,
+      "loss": 0.268,
+      "step": 12054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252499388394062e-05,
+      "loss": 0.241,
+      "step": 12055
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2251334928534575e-05,
+      "loss": 0.2365,
+      "step": 12056
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2250170436516957e-05,
+      "loss": 0.2287,
+      "step": 12057
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2249005912357837e-05,
+      "loss": 0.2284,
+      "step": 12058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.224784135607385e-05,
+      "loss": 0.221,
+      "step": 12059
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224667676768163e-05,
+      "loss": 0.2189,
+      "step": 12060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2245512147197815e-05,
+      "loss": 0.2501,
+      "step": 12061
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2244347494639034e-05,
+      "loss": 0.2432,
+      "step": 12062
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2243182810021935e-05,
+      "loss": 0.2382,
+      "step": 12063
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.224201809336314e-05,
+      "loss": 0.2378,
+      "step": 12064
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240853344679295e-05,
+      "loss": 0.2118,
+      "step": 12065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2239688563987033e-05,
+      "loss": 0.2645,
+      "step": 12066
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238523751302996e-05,
+      "loss": 0.2335,
+      "step": 12067
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237358906643817e-05,
+      "loss": 0.2023,
+      "step": 12068
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223619403002614e-05,
+      "loss": 0.2499,
+      "step": 12069
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2235029121466601e-05,
+      "loss": 0.2211,
+      "step": 12070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233864180981838e-05,
+      "loss": 0.2616,
+      "step": 12071
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232699208588495e-05,
+      "loss": 0.2329,
+      "step": 12072
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223153420430321e-05,
+      "loss": 0.2397,
+      "step": 12073
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2230369168142624e-05,
+      "loss": 0.2461,
+      "step": 12074
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2229204100123381e-05,
+      "loss": 0.2375,
+      "step": 12075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222803900026212e-05,
+      "loss": 0.2216,
+      "step": 12076
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2226873868575486e-05,
+      "loss": 0.2251,
+      "step": 12077
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225708705080118e-05,
+      "loss": 0.2582,
+      "step": 12078
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2224543509792662e-05,
+      "loss": 0.2451,
+      "step": 12079
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2223378282729759e-05,
+      "loss": 0.2397,
+      "step": 12080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2222213023908056e-05,
+      "loss": 0.2264,
+      "step": 12081
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221047733344198e-05,
+      "loss": 0.257,
+      "step": 12082
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2219882411054829e-05,
+      "loss": 0.247,
+      "step": 12083
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218717057056592e-05,
+      "loss": 0.2105,
+      "step": 12084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2217551671366134e-05,
+      "loss": 0.2517,
+      "step": 12085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2216386254000108e-05,
+      "loss": 0.2543,
+      "step": 12086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215220804975152e-05,
+      "loss": 0.2435,
+      "step": 12087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2214055324307917e-05,
+      "loss": 0.2385,
+      "step": 12088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221288981201505e-05,
+      "loss": 0.2324,
+      "step": 12089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211724268113202e-05,
+      "loss": 0.2191,
+      "step": 12090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2210558692619015e-05,
+      "loss": 0.2488,
+      "step": 12091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2209393085549146e-05,
+      "loss": 0.2146,
+      "step": 12092
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220822744692024e-05,
+      "loss": 0.2476,
+      "step": 12093
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2207061776748947e-05,
+      "loss": 0.2557,
+      "step": 12094
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220589607505192e-05,
+      "loss": 0.2175,
+      "step": 12095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2204730341845808e-05,
+      "loss": 0.2192,
+      "step": 12096
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203564577147262e-05,
+      "loss": 0.2581,
+      "step": 12097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202398780972937e-05,
+      "loss": 0.244,
+      "step": 12098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2201232953339482e-05,
+      "loss": 0.2101,
+      "step": 12099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2200067094263547e-05,
+      "loss": 0.2292,
+      "step": 12100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198901203761792e-05,
+      "loss": 0.217,
+      "step": 12101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197735281850868e-05,
+      "loss": 0.223,
+      "step": 12102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196569328547429e-05,
+      "loss": 0.2383,
+      "step": 12103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219540334386813e-05,
+      "loss": 0.1934,
+      "step": 12104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2194237327829618e-05,
+      "loss": 0.2254,
+      "step": 12105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2193071280448561e-05,
+      "loss": 0.2383,
+      "step": 12106
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191905201741608e-05,
+      "loss": 0.2453,
+      "step": 12107
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2190739091725416e-05,
+      "loss": 0.2444,
+      "step": 12108
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2189572950416643e-05,
+      "loss": 0.2374,
+      "step": 12109
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2188406777831948e-05,
+      "loss": 0.2308,
+      "step": 12110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2187240573987982e-05,
+      "loss": 0.228,
+      "step": 12111
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2186074338901408e-05,
+      "loss": 0.2271,
+      "step": 12112
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2184908072588882e-05,
+      "loss": 0.2052,
+      "step": 12113
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183741775067068e-05,
+      "loss": 0.212,
+      "step": 12114
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218257544635262e-05,
+      "loss": 0.2448,
+      "step": 12115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2181409086462204e-05,
+      "loss": 0.2715,
+      "step": 12116
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2180242695412471e-05,
+      "loss": 0.2454,
+      "step": 12117
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217907627322009e-05,
+      "loss": 0.2556,
+      "step": 12118
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177909819901717e-05,
+      "loss": 0.2257,
+      "step": 12119
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217674333547402e-05,
+      "loss": 0.2201,
+      "step": 12120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175576819953651e-05,
+      "loss": 0.2312,
+      "step": 12121
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2174410273357282e-05,
+      "loss": 0.2515,
+      "step": 12122
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173243695701575e-05,
+      "loss": 0.2472,
+      "step": 12123
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172077087003186e-05,
+      "loss": 0.216,
+      "step": 12124
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217091044727879e-05,
+      "loss": 0.2477,
+      "step": 12125
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216974377654504e-05,
+      "loss": 0.2347,
+      "step": 12126
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2168577074818608e-05,
+      "loss": 0.2706,
+      "step": 12127
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2167410342116156e-05,
+      "loss": 0.2025,
+      "step": 12128
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166243578454353e-05,
+      "loss": 0.2479,
+      "step": 12129
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216507678384986e-05,
+      "loss": 0.2212,
+      "step": 12130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2163909958319348e-05,
+      "loss": 0.2436,
+      "step": 12131
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162743101879484e-05,
+      "loss": 0.226,
+      "step": 12132
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216157621454693e-05,
+      "loss": 0.2243,
+      "step": 12133
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216040929633836e-05,
+      "loss": 0.232,
+      "step": 12134
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2159242347270439e-05,
+      "loss": 0.2307,
+      "step": 12135
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158075367359839e-05,
+      "loss": 0.2346,
+      "step": 12136
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156908356623225e-05,
+      "loss": 0.2361,
+      "step": 12137
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2155741315077268e-05,
+      "loss": 0.2811,
+      "step": 12138
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2154574242738639e-05,
+      "loss": 0.2407,
+      "step": 12139
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2153407139624007e-05,
+      "loss": 0.2216,
+      "step": 12140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2152240005750047e-05,
+      "loss": 0.2287,
+      "step": 12141
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2151072841133427e-05,
+      "loss": 0.2156,
+      "step": 12142
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2149905645790817e-05,
+      "loss": 0.2641,
+      "step": 12143
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.214873841973889e-05,
+      "loss": 0.223,
+      "step": 12144
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.214757116299432e-05,
+      "loss": 0.236,
+      "step": 12145
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146403875573783e-05,
+      "loss": 0.224,
+      "step": 12146
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2145236557493949e-05,
+      "loss": 0.2528,
+      "step": 12147
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2144069208771492e-05,
+      "loss": 0.236,
+      "step": 12148
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2142901829423091e-05,
+      "loss": 0.203,
+      "step": 12149
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141734419465412e-05,
+      "loss": 0.2471,
+      "step": 12150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2140566978915136e-05,
+      "loss": 0.2363,
+      "step": 12151
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2139399507788939e-05,
+      "loss": 0.2633,
+      "step": 12152
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138232006103498e-05,
+      "loss": 0.2278,
+      "step": 12153
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137064473875487e-05,
+      "loss": 0.2571,
+      "step": 12154
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135896911121584e-05,
+      "loss": 0.2565,
+      "step": 12155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134729317858467e-05,
+      "loss": 0.2557,
+      "step": 12156
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213356169410281e-05,
+      "loss": 0.2409,
+      "step": 12157
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2132394039871302e-05,
+      "loss": 0.2128,
+      "step": 12158
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213122635518061e-05,
+      "loss": 0.265,
+      "step": 12159
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2130058640047417e-05,
+      "loss": 0.2122,
+      "step": 12160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2128890894488406e-05,
+      "loss": 0.2365,
+      "step": 12161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127723118520254e-05,
+      "loss": 0.2557,
+      "step": 12162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126555312159641e-05,
+      "loss": 0.2171,
+      "step": 12163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2125387475423251e-05,
+      "loss": 0.2367,
+      "step": 12164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2124219608327764e-05,
+      "loss": 0.2287,
+      "step": 12165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123051710889858e-05,
+      "loss": 0.2253,
+      "step": 12166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212188378312622e-05,
+      "loss": 0.2153,
+      "step": 12167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120715825053533e-05,
+      "loss": 0.2392,
+      "step": 12168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211954783668848e-05,
+      "loss": 0.2369,
+      "step": 12169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2118379818047742e-05,
+      "loss": 0.2639,
+      "step": 12170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117211769148003e-05,
+      "loss": 0.2425,
+      "step": 12171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2116043690005949e-05,
+      "loss": 0.208,
+      "step": 12172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114875580638266e-05,
+      "loss": 0.2228,
+      "step": 12173
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2113707441061638e-05,
+      "loss": 0.2464,
+      "step": 12174
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2112539271292748e-05,
+      "loss": 0.2229,
+      "step": 12175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111371071348288e-05,
+      "loss": 0.2283,
+      "step": 12176
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2110202841244938e-05,
+      "loss": 0.2511,
+      "step": 12177
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210903458099939e-05,
+      "loss": 0.2095,
+      "step": 12178
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107866290628329e-05,
+      "loss": 0.2336,
+      "step": 12179
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2106697970148446e-05,
+      "loss": 0.2045,
+      "step": 12180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2105529619576426e-05,
+      "loss": 0.2321,
+      "step": 12181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104361238928959e-05,
+      "loss": 0.2421,
+      "step": 12182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103192828222733e-05,
+      "loss": 0.2159,
+      "step": 12183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210202438747444e-05,
+      "loss": 0.2546,
+      "step": 12184
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100855916700767e-05,
+      "loss": 0.2302,
+      "step": 12185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2099687415918408e-05,
+      "loss": 0.2174,
+      "step": 12186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209851888514405e-05,
+      "loss": 0.2285,
+      "step": 12187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097350324394388e-05,
+      "loss": 0.2006,
+      "step": 12188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2096181733686113e-05,
+      "loss": 0.2464,
+      "step": 12189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095013113035914e-05,
+      "loss": 0.2221,
+      "step": 12190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2093844462460488e-05,
+      "loss": 0.2368,
+      "step": 12191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092675781976526e-05,
+      "loss": 0.2358,
+      "step": 12192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2091507071600718e-05,
+      "loss": 0.2341,
+      "step": 12193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090338331349764e-05,
+      "loss": 0.2497,
+      "step": 12194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2089169561240358e-05,
+      "loss": 0.2139,
+      "step": 12195
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208800076128919e-05,
+      "loss": 0.243,
+      "step": 12196
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086831931512956e-05,
+      "loss": 0.2273,
+      "step": 12197
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2085663071928355e-05,
+      "loss": 0.2311,
+      "step": 12198
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208449418255208e-05,
+      "loss": 0.264,
+      "step": 12199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208332526340083e-05,
+      "loss": 0.2514,
+      "step": 12200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082156314491298e-05,
+      "loss": 0.2137,
+      "step": 12201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080987335840186e-05,
+      "loss": 0.2248,
+      "step": 12202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079818327464193e-05,
+      "loss": 0.2234,
+      "step": 12203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207864928938001e-05,
+      "loss": 0.241,
+      "step": 12204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2077480221604337e-05,
+      "loss": 0.2391,
+      "step": 12205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2076311124153876e-05,
+      "loss": 0.2323,
+      "step": 12206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2075141997045328e-05,
+      "loss": 0.2247,
+      "step": 12207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073972840295391e-05,
+      "loss": 0.2183,
+      "step": 12208
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072803653920766e-05,
+      "loss": 0.2461,
+      "step": 12209
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207163443793815e-05,
+      "loss": 0.2668,
+      "step": 12210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2070465192364245e-05,
+      "loss": 0.2154,
+      "step": 12211
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069295917215756e-05,
+      "loss": 0.2476,
+      "step": 12212
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2068126612509384e-05,
+      "loss": 0.2405,
+      "step": 12213
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2066957278261828e-05,
+      "loss": 0.261,
+      "step": 12214
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2065787914489797e-05,
+      "loss": 0.2317,
+      "step": 12215
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2064618521209988e-05,
+      "loss": 0.2319,
+      "step": 12216
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2063449098439106e-05,
+      "loss": 0.2408,
+      "step": 12217
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2062279646193857e-05,
+      "loss": 0.2219,
+      "step": 12218
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2061110164490948e-05,
+      "loss": 0.2471,
+      "step": 12219
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059940653347077e-05,
+      "loss": 0.2383,
+      "step": 12220
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058771112778955e-05,
+      "loss": 0.242,
+      "step": 12221
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2057601542803286e-05,
+      "loss": 0.243,
+      "step": 12222
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056431943436776e-05,
+      "loss": 0.2364,
+      "step": 12223
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2055262314696127e-05,
+      "loss": 0.2521,
+      "step": 12224
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2054092656598055e-05,
+      "loss": 0.2189,
+      "step": 12225
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2052922969159262e-05,
+      "loss": 0.2327,
+      "step": 12226
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051753252396457e-05,
+      "loss": 0.2422,
+      "step": 12227
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2050583506326348e-05,
+      "loss": 0.2408,
+      "step": 12228
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2049413730965643e-05,
+      "loss": 0.2389,
+      "step": 12229
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2048243926331054e-05,
+      "loss": 0.2422,
+      "step": 12230
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2047074092439286e-05,
+      "loss": 0.2378,
+      "step": 12231
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204590422930705e-05,
+      "loss": 0.2478,
+      "step": 12232
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204473433695106e-05,
+      "loss": 0.2341,
+      "step": 12233
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2043564415388023e-05,
+      "loss": 0.2402,
+      "step": 12234
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2042394464634655e-05,
+      "loss": 0.2225,
+      "step": 12235
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2041224484707663e-05,
+      "loss": 0.2691,
+      "step": 12236
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204005447562376e-05,
+      "loss": 0.2369,
+      "step": 12237
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2038884437399656e-05,
+      "loss": 0.2315,
+      "step": 12238
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2037714370052069e-05,
+      "loss": 0.2368,
+      "step": 12239
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2036544273597708e-05,
+      "loss": 0.224,
+      "step": 12240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035374148053291e-05,
+      "loss": 0.2488,
+      "step": 12241
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2034203993435532e-05,
+      "loss": 0.2216,
+      "step": 12242
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203303380976114e-05,
+      "loss": 0.2523,
+      "step": 12243
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203186359704683e-05,
+      "loss": 0.2249,
+      "step": 12244
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030693355309324e-05,
+      "loss": 0.2022,
+      "step": 12245
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2029523084565335e-05,
+      "loss": 0.2129,
+      "step": 12246
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2028352784831578e-05,
+      "loss": 0.256,
+      "step": 12247
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202718245612477e-05,
+      "loss": 0.2109,
+      "step": 12248
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2026012098461628e-05,
+      "loss": 0.2654,
+      "step": 12249
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024841711858868e-05,
+      "loss": 0.243,
+      "step": 12250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2023671296333208e-05,
+      "loss": 0.2481,
+      "step": 12251
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2022500851901371e-05,
+      "loss": 0.2371,
+      "step": 12252
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021330378580071e-05,
+      "loss": 0.2247,
+      "step": 12253
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202015987638603e-05,
+      "loss": 0.2043,
+      "step": 12254
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018989345335962e-05,
+      "loss": 0.2449,
+      "step": 12255
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2017818785446594e-05,
+      "loss": 0.2415,
+      "step": 12256
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.201664819673464e-05,
+      "loss": 0.2129,
+      "step": 12257
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2015477579216827e-05,
+      "loss": 0.2495,
+      "step": 12258
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014306932909871e-05,
+      "loss": 0.2472,
+      "step": 12259
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2013136257830494e-05,
+      "loss": 0.2091,
+      "step": 12260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011965553995423e-05,
+      "loss": 0.234,
+      "step": 12261
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2010794821421373e-05,
+      "loss": 0.2249,
+      "step": 12262
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2009624060125074e-05,
+      "loss": 0.2585,
+      "step": 12263
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008453270123244e-05,
+      "loss": 0.2528,
+      "step": 12264
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2007282451432606e-05,
+      "loss": 0.2204,
+      "step": 12265
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200611160406989e-05,
+      "loss": 0.2463,
+      "step": 12266
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004940728051813e-05,
+      "loss": 0.245,
+      "step": 12267
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2003769823395107e-05,
+      "loss": 0.2339,
+      "step": 12268
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2002598890116494e-05,
+      "loss": 0.236,
+      "step": 12269
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001427928232699e-05,
+      "loss": 0.2418,
+      "step": 12270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 0.2424,
+      "step": 12271
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1999085918716464e-05,
+      "loss": 0.2195,
+      "step": 12272
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199791487111748e-05,
+      "loss": 0.2311,
+      "step": 12273
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1996743794980224e-05,
+      "loss": 0.2605,
+      "step": 12274
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199557269032142e-05,
+      "loss": 0.2145,
+      "step": 12275
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994401557157796e-05,
+      "loss": 0.2377,
+      "step": 12276
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1993230395506081e-05,
+      "loss": 0.2025,
+      "step": 12277
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1992059205383004e-05,
+      "loss": 0.2143,
+      "step": 12278
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1990887986805295e-05,
+      "loss": 0.204,
+      "step": 12279
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1989716739789682e-05,
+      "loss": 0.2518,
+      "step": 12280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1988545464352901e-05,
+      "loss": 0.2924,
+      "step": 12281
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987374160511673e-05,
+      "loss": 0.2525,
+      "step": 12282
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1986202828282736e-05,
+      "loss": 0.2227,
+      "step": 12283
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1985031467682819e-05,
+      "loss": 0.2583,
+      "step": 12284
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1983860078728656e-05,
+      "loss": 0.2664,
+      "step": 12285
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982688661436975e-05,
+      "loss": 0.2651,
+      "step": 12286
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198151721582451e-05,
+      "loss": 0.2634,
+      "step": 12287
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980345741907996e-05,
+      "loss": 0.231,
+      "step": 12288
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1979174239704164e-05,
+      "loss": 0.2318,
+      "step": 12289
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1978002709229748e-05,
+      "loss": 0.1962,
+      "step": 12290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976831150501487e-05,
+      "loss": 0.2253,
+      "step": 12291
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1975659563536108e-05,
+      "loss": 0.2273,
+      "step": 12292
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1974487948350349e-05,
+      "loss": 0.2331,
+      "step": 12293
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197331630496095e-05,
+      "loss": 0.1935,
+      "step": 12294
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197214463338464e-05,
+      "loss": 0.2431,
+      "step": 12295
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970972933638159e-05,
+      "loss": 0.2268,
+      "step": 12296
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969801205738241e-05,
+      "loss": 0.2833,
+      "step": 12297
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968629449701623e-05,
+      "loss": 0.2278,
+      "step": 12298
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967457665545046e-05,
+      "loss": 0.2364,
+      "step": 12299
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1966285853285248e-05,
+      "loss": 0.2767,
+      "step": 12300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196511401293896e-05,
+      "loss": 0.2316,
+      "step": 12301
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196394214452293e-05,
+      "loss": 0.2253,
+      "step": 12302
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1962770248053891e-05,
+      "loss": 0.2365,
+      "step": 12303
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1961598323548582e-05,
+      "loss": 0.2495,
+      "step": 12304
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960426371023747e-05,
+      "loss": 0.2392,
+      "step": 12305
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959254390496122e-05,
+      "loss": 0.2186,
+      "step": 12306
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958082381982453e-05,
+      "loss": 0.2472,
+      "step": 12307
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956910345499477e-05,
+      "loss": 0.2255,
+      "step": 12308
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1955738281063933e-05,
+      "loss": 0.2455,
+      "step": 12309
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954566188692567e-05,
+      "loss": 0.1956,
+      "step": 12310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195339406840212e-05,
+      "loss": 0.2299,
+      "step": 12311
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952221920209333e-05,
+      "loss": 0.2601,
+      "step": 12312
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1951049744130953e-05,
+      "loss": 0.2486,
+      "step": 12313
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1949877540183723e-05,
+      "loss": 0.2123,
+      "step": 12314
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194870530838438e-05,
+      "loss": 0.2566,
+      "step": 12315
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1947533048749676e-05,
+      "loss": 0.2293,
+      "step": 12316
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946360761296349e-05,
+      "loss": 0.2168,
+      "step": 12317
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194518844604115e-05,
+      "loss": 0.2307,
+      "step": 12318
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1944016103000822e-05,
+      "loss": 0.2501,
+      "step": 12319
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942843732192111e-05,
+      "loss": 0.2428,
+      "step": 12320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1941671333631763e-05,
+      "loss": 0.2348,
+      "step": 12321
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1940498907336523e-05,
+      "loss": 0.2276,
+      "step": 12322
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1939326453323138e-05,
+      "loss": 0.2316,
+      "step": 12323
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1938153971608361e-05,
+      "loss": 0.2201,
+      "step": 12324
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1936981462208933e-05,
+      "loss": 0.23,
+      "step": 12325
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935808925141603e-05,
+      "loss": 0.2116,
+      "step": 12326
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934636360423127e-05,
+      "loss": 0.2121,
+      "step": 12327
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933463768070244e-05,
+      "loss": 0.2451,
+      "step": 12328
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1932291148099706e-05,
+      "loss": 0.2321,
+      "step": 12329
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1931118500528266e-05,
+      "loss": 0.207,
+      "step": 12330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1929945825372674e-05,
+      "loss": 0.2038,
+      "step": 12331
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1928773122649677e-05,
+      "loss": 0.2221,
+      "step": 12332
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1927600392376028e-05,
+      "loss": 0.2357,
+      "step": 12333
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192642763456848e-05,
+      "loss": 0.2284,
+      "step": 12334
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1925254849243783e-05,
+      "loss": 0.2334,
+      "step": 12335
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1924082036418687e-05,
+      "loss": 0.2197,
+      "step": 12336
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1922909196109947e-05,
+      "loss": 0.2439,
+      "step": 12337
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1921736328334314e-05,
+      "loss": 0.2472,
+      "step": 12338
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1920563433108544e-05,
+      "loss": 0.2461,
+      "step": 12339
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919390510449392e-05,
+      "loss": 0.2272,
+      "step": 12340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1918217560373607e-05,
+      "loss": 0.2437,
+      "step": 12341
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1917044582897944e-05,
+      "loss": 0.2319,
+      "step": 12342
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191587157803916e-05,
+      "loss": 0.2236,
+      "step": 12343
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1914698545814012e-05,
+      "loss": 0.2504,
+      "step": 12344
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1913525486239252e-05,
+      "loss": 0.2318,
+      "step": 12345
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191235239933164e-05,
+      "loss": 0.2165,
+      "step": 12346
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1911179285107927e-05,
+      "loss": 0.2575,
+      "step": 12347
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1910006143584876e-05,
+      "loss": 0.2316,
+      "step": 12348
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1908832974779237e-05,
+      "loss": 0.234,
+      "step": 12349
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1907659778707772e-05,
+      "loss": 0.2488,
+      "step": 12350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1906486555387242e-05,
+      "loss": 0.215,
+      "step": 12351
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905313304834398e-05,
+      "loss": 0.2326,
+      "step": 12352
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904140027066007e-05,
+      "loss": 0.2139,
+      "step": 12353
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1902966722098821e-05,
+      "loss": 0.2207,
+      "step": 12354
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.19017933899496e-05,
+      "loss": 0.2347,
+      "step": 12355
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900620030635109e-05,
+      "loss": 0.2233,
+      "step": 12356
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1899446644172106e-05,
+      "loss": 0.2216,
+      "step": 12357
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189827323057735e-05,
+      "loss": 0.2343,
+      "step": 12358
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1897099789867604e-05,
+      "loss": 0.2182,
+      "step": 12359
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895926322059628e-05,
+      "loss": 0.2482,
+      "step": 12360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1894752827170184e-05,
+      "loss": 0.2473,
+      "step": 12361
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893579305216034e-05,
+      "loss": 0.22,
+      "step": 12362
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892405756213944e-05,
+      "loss": 0.2256,
+      "step": 12363
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1891232180180672e-05,
+      "loss": 0.2422,
+      "step": 12364
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890058577132983e-05,
+      "loss": 0.2143,
+      "step": 12365
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1888884947087647e-05,
+      "loss": 0.2178,
+      "step": 12366
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1887711290061418e-05,
+      "loss": 0.2286,
+      "step": 12367
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886537606071068e-05,
+      "loss": 0.225,
+      "step": 12368
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1885363895133357e-05,
+      "loss": 0.1996,
+      "step": 12369
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1884190157265051e-05,
+      "loss": 0.2764,
+      "step": 12370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188301639248292e-05,
+      "loss": 0.229,
+      "step": 12371
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881842600803726e-05,
+      "loss": 0.2258,
+      "step": 12372
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188066878224424e-05,
+      "loss": 0.2363,
+      "step": 12373
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879494936821224e-05,
+      "loss": 0.2437,
+      "step": 12374
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1878321064551446e-05,
+      "loss": 0.2253,
+      "step": 12375
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877147165451674e-05,
+      "loss": 0.1961,
+      "step": 12376
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875973239538677e-05,
+      "loss": 0.2384,
+      "step": 12377
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874799286829221e-05,
+      "loss": 0.2261,
+      "step": 12378
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1873625307340082e-05,
+      "loss": 0.2354,
+      "step": 12379
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1872451301088022e-05,
+      "loss": 0.2172,
+      "step": 12380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871277268089809e-05,
+      "loss": 0.2296,
+      "step": 12381
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1870103208362217e-05,
+      "loss": 0.2261,
+      "step": 12382
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868929121922019e-05,
+      "loss": 0.2284,
+      "step": 12383
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867755008785982e-05,
+      "loss": 0.2364,
+      "step": 12384
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1866580868970875e-05,
+      "loss": 0.2603,
+      "step": 12385
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865406702493475e-05,
+      "loss": 0.2342,
+      "step": 12386
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864232509370547e-05,
+      "loss": 0.2135,
+      "step": 12387
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1863058289618867e-05,
+      "loss": 0.2455,
+      "step": 12388
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186188404325521e-05,
+      "loss": 0.2132,
+      "step": 12389
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860709770296346e-05,
+      "loss": 0.242,
+      "step": 12390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1859535470759046e-05,
+      "loss": 0.2171,
+      "step": 12391
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1858361144660088e-05,
+      "loss": 0.2256,
+      "step": 12392
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857186792016249e-05,
+      "loss": 0.2597,
+      "step": 12393
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856012412844293e-05,
+      "loss": 0.2464,
+      "step": 12394
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1854838007161002e-05,
+      "loss": 0.2305,
+      "step": 12395
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1853663574983154e-05,
+      "loss": 0.2125,
+      "step": 12396
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185248911632752e-05,
+      "loss": 0.2145,
+      "step": 12397
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851314631210873e-05,
+      "loss": 0.2506,
+      "step": 12398
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1850140119649998e-05,
+      "loss": 0.2083,
+      "step": 12399
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1848965581661662e-05,
+      "loss": 0.2275,
+      "step": 12400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184779101726265e-05,
+      "loss": 0.2358,
+      "step": 12401
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1846616426469737e-05,
+      "loss": 0.2333,
+      "step": 12402
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18454418092997e-05,
+      "loss": 0.2272,
+      "step": 12403
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844267165769317e-05,
+      "loss": 0.2337,
+      "step": 12404
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1843092495895368e-05,
+      "loss": 0.2397,
+      "step": 12405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1841917799694631e-05,
+      "loss": 0.2517,
+      "step": 12406
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184074307718389e-05,
+      "loss": 0.254,
+      "step": 12407
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1839568328379915e-05,
+      "loss": 0.2522,
+      "step": 12408
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838393553299495e-05,
+      "loss": 0.2523,
+      "step": 12409
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837218751959403e-05,
+      "loss": 0.2289,
+      "step": 12410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836043924376428e-05,
+      "loss": 0.2361,
+      "step": 12411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834869070567348e-05,
+      "loss": 0.2463,
+      "step": 12412
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1833694190548944e-05,
+      "loss": 0.2319,
+      "step": 12413
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1832519284337996e-05,
+      "loss": 0.2375,
+      "step": 12414
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1831344351951288e-05,
+      "loss": 0.2215,
+      "step": 12415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830169393405607e-05,
+      "loss": 0.2422,
+      "step": 12416
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1828994408717729e-05,
+      "loss": 0.2427,
+      "step": 12417
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1827819397904443e-05,
+      "loss": 0.2429,
+      "step": 12418
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1826644360982531e-05,
+      "loss": 0.2661,
+      "step": 12419
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1825469297968775e-05,
+      "loss": 0.2237,
+      "step": 12420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1824294208879963e-05,
+      "loss": 0.2355,
+      "step": 12421
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1823119093732878e-05,
+      "loss": 0.2177,
+      "step": 12422
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821943952544308e-05,
+      "loss": 0.232,
+      "step": 12423
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1820768785331033e-05,
+      "loss": 0.2278,
+      "step": 12424
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1819593592109846e-05,
+      "loss": 0.2381,
+      "step": 12425
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1818418372897533e-05,
+      "loss": 0.2304,
+      "step": 12426
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817243127710873e-05,
+      "loss": 0.2488,
+      "step": 12427
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1816067856566662e-05,
+      "loss": 0.226,
+      "step": 12428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1814892559481685e-05,
+      "loss": 0.2309,
+      "step": 12429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813717236472725e-05,
+      "loss": 0.2236,
+      "step": 12430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812541887556577e-05,
+      "loss": 0.2313,
+      "step": 12431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1811366512750028e-05,
+      "loss": 0.2298,
+      "step": 12432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1810191112069867e-05,
+      "loss": 0.2523,
+      "step": 12433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1809015685532881e-05,
+      "loss": 0.2418,
+      "step": 12434
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807840233155863e-05,
+      "loss": 0.2168,
+      "step": 12435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806664754955599e-05,
+      "loss": 0.2233,
+      "step": 12436
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1805489250948883e-05,
+      "loss": 0.2461,
+      "step": 12437
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1804313721152506e-05,
+      "loss": 0.2407,
+      "step": 12438
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1803138165583262e-05,
+      "loss": 0.2362,
+      "step": 12439
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801962584257937e-05,
+      "loss": 0.2335,
+      "step": 12440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1800786977193325e-05,
+      "loss": 0.2412,
+      "step": 12441
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1799611344406217e-05,
+      "loss": 0.2426,
+      "step": 12442
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798435685913407e-05,
+      "loss": 0.2258,
+      "step": 12443
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1797260001731691e-05,
+      "loss": 0.2135,
+      "step": 12444
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1796084291877864e-05,
+      "loss": 0.2573,
+      "step": 12445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1794908556368712e-05,
+      "loss": 0.236,
+      "step": 12446
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179373279522103e-05,
+      "loss": 0.2163,
+      "step": 12447
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792557008451619e-05,
+      "loss": 0.2365,
+      "step": 12448
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1791381196077272e-05,
+      "loss": 0.2232,
+      "step": 12449
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179020535811478e-05,
+      "loss": 0.2508,
+      "step": 12450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789029494580947e-05,
+      "loss": 0.2325,
+      "step": 12451
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178785360549256e-05,
+      "loss": 0.2647,
+      "step": 12452
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786677690866422e-05,
+      "loss": 0.2515,
+      "step": 12453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785501750719323e-05,
+      "loss": 0.2254,
+      "step": 12454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1784325785068067e-05,
+      "loss": 0.2284,
+      "step": 12455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178314979392945e-05,
+      "loss": 0.2232,
+      "step": 12456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1781973777320267e-05,
+      "loss": 0.2392,
+      "step": 12457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780797735257318e-05,
+      "loss": 0.243,
+      "step": 12458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1779621667757403e-05,
+      "loss": 0.2407,
+      "step": 12459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778445574837318e-05,
+      "loss": 0.2489,
+      "step": 12460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777269456513865e-05,
+      "loss": 0.2405,
+      "step": 12461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1776093312803845e-05,
+      "loss": 0.2554,
+      "step": 12462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1774917143724052e-05,
+      "loss": 0.2618,
+      "step": 12463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177374094929129e-05,
+      "loss": 0.2431,
+      "step": 12464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772564729522365e-05,
+      "loss": 0.2591,
+      "step": 12465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771388484434068e-05,
+      "loss": 0.2393,
+      "step": 12466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770212214043205e-05,
+      "loss": 0.2245,
+      "step": 12467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769035918366582e-05,
+      "loss": 0.2484,
+      "step": 12468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1767859597420997e-05,
+      "loss": 0.2206,
+      "step": 12469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766683251223252e-05,
+      "loss": 0.2601,
+      "step": 12470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1765506879790152e-05,
+      "loss": 0.2583,
+      "step": 12471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1764330483138502e-05,
+      "loss": 0.2462,
+      "step": 12472
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1763154061285102e-05,
+      "loss": 0.243,
+      "step": 12473
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1761977614246757e-05,
+      "loss": 0.2406,
+      "step": 12474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176080114204027e-05,
+      "loss": 0.231,
+      "step": 12475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175962464468245e-05,
+      "loss": 0.2272,
+      "step": 12476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758448122190099e-05,
+      "loss": 0.2314,
+      "step": 12477
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1757271574580025e-05,
+      "loss": 0.252,
+      "step": 12478
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175609500186903e-05,
+      "loss": 0.2245,
+      "step": 12479
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754918404073922e-05,
+      "loss": 0.2367,
+      "step": 12480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753741781211507e-05,
+      "loss": 0.2269,
+      "step": 12481
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1752565133298594e-05,
+      "loss": 0.2264,
+      "step": 12482
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175138846035199e-05,
+      "loss": 0.2322,
+      "step": 12483
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17502117623885e-05,
+      "loss": 0.2193,
+      "step": 12484
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749035039424936e-05,
+      "loss": 0.2228,
+      "step": 12485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747858291478103e-05,
+      "loss": 0.2409,
+      "step": 12486
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746681518564809e-05,
+      "loss": 0.2264,
+      "step": 12487
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745504720701863e-05,
+      "loss": 0.2263,
+      "step": 12488
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174432789790608e-05,
+      "loss": 0.2377,
+      "step": 12489
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1743151050194263e-05,
+      "loss": 0.2096,
+      "step": 12490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741974177583228e-05,
+      "loss": 0.2455,
+      "step": 12491
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1740797280089782e-05,
+      "loss": 0.2291,
+      "step": 12492
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739620357730735e-05,
+      "loss": 0.229,
+      "step": 12493
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1738443410522899e-05,
+      "loss": 0.2612,
+      "step": 12494
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1737266438483087e-05,
+      "loss": 0.2355,
+      "step": 12495
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736089441628109e-05,
+      "loss": 0.2518,
+      "step": 12496
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173491241997478e-05,
+      "loss": 0.2165,
+      "step": 12497
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173373537353991e-05,
+      "loss": 0.2725,
+      "step": 12498
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1732558302340313e-05,
+      "loss": 0.2415,
+      "step": 12499
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1731381206392802e-05,
+      "loss": 0.2434,
+      "step": 12500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1730204085714192e-05,
+      "loss": 0.2533,
+      "step": 12501
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729026940321294e-05,
+      "loss": 0.2039,
+      "step": 12502
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1727849770230924e-05,
+      "loss": 0.2206,
+      "step": 12503
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1726672575459896e-05,
+      "loss": 0.2489,
+      "step": 12504
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172549535602503e-05,
+      "loss": 0.2558,
+      "step": 12505
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724318111943134e-05,
+      "loss": 0.2077,
+      "step": 12506
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172314084323103e-05,
+      "loss": 0.2328,
+      "step": 12507
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172196354990553e-05,
+      "loss": 0.2484,
+      "step": 12508
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172078623198345e-05,
+      "loss": 0.2405,
+      "step": 12509
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1719608889481611e-05,
+      "loss": 0.2345,
+      "step": 12510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171843152241683e-05,
+      "loss": 0.1979,
+      "step": 12511
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171725413080592e-05,
+      "loss": 0.2277,
+      "step": 12512
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1716076714665701e-05,
+      "loss": 0.2315,
+      "step": 12513
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1714899274012993e-05,
+      "loss": 0.2282,
+      "step": 12514
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171372180886461e-05,
+      "loss": 0.2354,
+      "step": 12515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712544319237376e-05,
+      "loss": 0.2204,
+      "step": 12516
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171136680514811e-05,
+      "loss": 0.2311,
+      "step": 12517
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171018926661363e-05,
+      "loss": 0.2172,
+      "step": 12518
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1709011703650755e-05,
+      "loss": 0.2557,
+      "step": 12519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707834116276303e-05,
+      "loss": 0.2221,
+      "step": 12520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1706656504507103e-05,
+      "loss": 0.2557,
+      "step": 12521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170547886835997e-05,
+      "loss": 0.233,
+      "step": 12522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1704301207851728e-05,
+      "loss": 0.2489,
+      "step": 12523
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703123522999195e-05,
+      "loss": 0.2407,
+      "step": 12524
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1701945813819195e-05,
+      "loss": 0.2259,
+      "step": 12525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700768080328553e-05,
+      "loss": 0.2347,
+      "step": 12526
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1699590322544088e-05,
+      "loss": 0.2343,
+      "step": 12527
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698412540482627e-05,
+      "loss": 0.2231,
+      "step": 12528
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697234734160988e-05,
+      "loss": 0.2547,
+      "step": 12529
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1696056903595998e-05,
+      "loss": 0.2357,
+      "step": 12530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1694879048804486e-05,
+      "loss": 0.2284,
+      "step": 12531
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1693701169803266e-05,
+      "loss": 0.209,
+      "step": 12532
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169252326660917e-05,
+      "loss": 0.2521,
+      "step": 12533
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691345339239027e-05,
+      "loss": 0.2516,
+      "step": 12534
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169016738770965e-05,
+      "loss": 0.2519,
+      "step": 12535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688989412037874e-05,
+      "loss": 0.2183,
+      "step": 12536
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1687811412240526e-05,
+      "loss": 0.2477,
+      "step": 12537
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168663338833443e-05,
+      "loss": 0.2334,
+      "step": 12538
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168545534033641e-05,
+      "loss": 0.2318,
+      "step": 12539
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16842772682633e-05,
+      "loss": 0.2447,
+      "step": 12540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168309917213192e-05,
+      "loss": 0.2196,
+      "step": 12541
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681921051959105e-05,
+      "loss": 0.2119,
+      "step": 12542
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1680742907761677e-05,
+      "loss": 0.2232,
+      "step": 12543
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1679564739556471e-05,
+      "loss": 0.2174,
+      "step": 12544
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678386547360312e-05,
+      "loss": 0.2285,
+      "step": 12545
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677208331190029e-05,
+      "loss": 0.2655,
+      "step": 12546
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676030091062454e-05,
+      "loss": 0.2373,
+      "step": 12547
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674851826994413e-05,
+      "loss": 0.2274,
+      "step": 12548
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167367353900274e-05,
+      "loss": 0.2255,
+      "step": 12549
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1672495227104268e-05,
+      "loss": 0.2515,
+      "step": 12550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671316891315823e-05,
+      "loss": 0.2516,
+      "step": 12551
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1670138531654238e-05,
+      "loss": 0.2458,
+      "step": 12552
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668960148136344e-05,
+      "loss": 0.2122,
+      "step": 12553
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1667781740778974e-05,
+      "loss": 0.2329,
+      "step": 12554
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1666603309598961e-05,
+      "loss": 0.2361,
+      "step": 12555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1665424854613139e-05,
+      "loss": 0.2435,
+      "step": 12556
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166424637583834e-05,
+      "loss": 0.2442,
+      "step": 12557
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1663067873291394e-05,
+      "loss": 0.2325,
+      "step": 12558
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661889346989136e-05,
+      "loss": 0.2214,
+      "step": 12559
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660710796948405e-05,
+      "loss": 0.2285,
+      "step": 12560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1659532223186033e-05,
+      "loss": 0.216,
+      "step": 12561
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165835362571885e-05,
+      "loss": 0.2463,
+      "step": 12562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657175004563694e-05,
+      "loss": 0.2416,
+      "step": 12563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1655996359737407e-05,
+      "loss": 0.2291,
+      "step": 12564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654817691256814e-05,
+      "loss": 0.229,
+      "step": 12565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1653638999138759e-05,
+      "loss": 0.2335,
+      "step": 12566
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1652460283400075e-05,
+      "loss": 0.25,
+      "step": 12567
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16512815440576e-05,
+      "loss": 0.2412,
+      "step": 12568
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650102781128169e-05,
+      "loss": 0.2346,
+      "step": 12569
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1648923994628625e-05,
+      "loss": 0.2316,
+      "step": 12570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1647745184575798e-05,
+      "loss": 0.261,
+      "step": 12571
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1646566350986533e-05,
+      "loss": 0.2245,
+      "step": 12572
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1645387493877664e-05,
+      "loss": 0.2504,
+      "step": 12573
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644208613266033e-05,
+      "loss": 0.2342,
+      "step": 12574
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643029709168477e-05,
+      "loss": 0.2485,
+      "step": 12575
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1641850781601835e-05,
+      "loss": 0.2349,
+      "step": 12576
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164067183058295e-05,
+      "loss": 0.2383,
+      "step": 12577
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163949285612866e-05,
+      "loss": 0.2396,
+      "step": 12578
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1638313858255806e-05,
+      "loss": 0.2319,
+      "step": 12579
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1637134836981225e-05,
+      "loss": 0.2097,
+      "step": 12580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1635955792321765e-05,
+      "loss": 0.2511,
+      "step": 12581
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634776724294263e-05,
+      "loss": 0.242,
+      "step": 12582
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633597632915563e-05,
+      "loss": 0.2445,
+      "step": 12583
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632418518202509e-05,
+      "loss": 0.2461,
+      "step": 12584
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1631239380171937e-05,
+      "loss": 0.216,
+      "step": 12585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1630060218840692e-05,
+      "loss": 0.2381,
+      "step": 12586
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628881034225621e-05,
+      "loss": 0.2231,
+      "step": 12587
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1627701826343564e-05,
+      "loss": 0.2355,
+      "step": 12588
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1626522595211371e-05,
+      "loss": 0.2436,
+      "step": 12589
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625343340845878e-05,
+      "loss": 0.2265,
+      "step": 12590
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1624164063263931e-05,
+      "loss": 0.2162,
+      "step": 12591
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162298476248238e-05,
+      "loss": 0.2529,
+      "step": 12592
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621805438518064e-05,
+      "loss": 0.2124,
+      "step": 12593
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1620626091387833e-05,
+      "loss": 0.2269,
+      "step": 12594
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161944672110853e-05,
+      "loss": 0.2288,
+      "step": 12595
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1618267327697001e-05,
+      "loss": 0.2328,
+      "step": 12596
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16170879111701e-05,
+      "loss": 0.2206,
+      "step": 12597
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615908471544662e-05,
+      "loss": 0.2401,
+      "step": 12598
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161472900883754e-05,
+      "loss": 0.2575,
+      "step": 12599
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613549523065583e-05,
+      "loss": 0.2437,
+      "step": 12600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1612370014245636e-05,
+      "loss": 0.2066,
+      "step": 12601
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1611190482394548e-05,
+      "loss": 0.2312,
+      "step": 12602
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.161001092752917e-05,
+      "loss": 0.2261,
+      "step": 12603
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1608831349666345e-05,
+      "loss": 0.2718,
+      "step": 12604
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1607651748822928e-05,
+      "loss": 0.2207,
+      "step": 12605
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606472125015766e-05,
+      "loss": 0.2491,
+      "step": 12606
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605292478261707e-05,
+      "loss": 0.2293,
+      "step": 12607
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1604112808577603e-05,
+      "loss": 0.2328,
+      "step": 12608
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602933115980304e-05,
+      "loss": 0.2363,
+      "step": 12609
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1601753400486664e-05,
+      "loss": 0.2141,
+      "step": 12610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.160057366211353e-05,
+      "loss": 0.2367,
+      "step": 12611
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1599393900877756e-05,
+      "loss": 0.233,
+      "step": 12612
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1598214116796189e-05,
+      "loss": 0.2434,
+      "step": 12613
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597034309885686e-05,
+      "loss": 0.2278,
+      "step": 12614
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1595854480163097e-05,
+      "loss": 0.2488,
+      "step": 12615
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159467462764528e-05,
+      "loss": 0.2304,
+      "step": 12616
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593494752349079e-05,
+      "loss": 0.2108,
+      "step": 12617
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1592314854291353e-05,
+      "loss": 0.2641,
+      "step": 12618
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591134933488954e-05,
+      "loss": 0.2626,
+      "step": 12619
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1589954989958736e-05,
+      "loss": 0.2471,
+      "step": 12620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588775023717556e-05,
+      "loss": 0.2443,
+      "step": 12621
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1587595034782268e-05,
+      "loss": 0.2499,
+      "step": 12622
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1586415023169726e-05,
+      "loss": 0.2329,
+      "step": 12623
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158523498889678e-05,
+      "loss": 0.2509,
+      "step": 12624
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584054931980295e-05,
+      "loss": 0.2426,
+      "step": 12625
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158287485243712e-05,
+      "loss": 0.2366,
+      "step": 12626
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1581694750284114e-05,
+      "loss": 0.2452,
+      "step": 12627
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1580514625538138e-05,
+      "loss": 0.2484,
+      "step": 12628
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157933447821604e-05,
+      "loss": 0.2093,
+      "step": 12629
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1578154308334683e-05,
+      "loss": 0.2536,
+      "step": 12630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576974115910924e-05,
+      "loss": 0.2097,
+      "step": 12631
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157579390096162e-05,
+      "loss": 0.2265,
+      "step": 12632
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574613663503632e-05,
+      "loss": 0.2243,
+      "step": 12633
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573433403553811e-05,
+      "loss": 0.2256,
+      "step": 12634
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1572253121129022e-05,
+      "loss": 0.243,
+      "step": 12635
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571072816246125e-05,
+      "loss": 0.2142,
+      "step": 12636
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1569892488921977e-05,
+      "loss": 0.2249,
+      "step": 12637
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1568712139173437e-05,
+      "loss": 0.2635,
+      "step": 12638
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156753176701737e-05,
+      "loss": 0.2287,
+      "step": 12639
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156635137247063e-05,
+      "loss": 0.2335,
+      "step": 12640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156517095555008e-05,
+      "loss": 0.2471,
+      "step": 12641
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563990516272582e-05,
+      "loss": 0.23,
+      "step": 12642
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1562810054655002e-05,
+      "loss": 0.2335,
+      "step": 12643
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561629570714192e-05,
+      "loss": 0.2332,
+      "step": 12644
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560449064467023e-05,
+      "loss": 0.2789,
+      "step": 12645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1559268535930352e-05,
+      "loss": 0.2289,
+      "step": 12646
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1558087985121042e-05,
+      "loss": 0.2388,
+      "step": 12647
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1556907412055957e-05,
+      "loss": 0.249,
+      "step": 12648
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555726816751965e-05,
+      "loss": 0.232,
+      "step": 12649
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1554546199225922e-05,
+      "loss": 0.2435,
+      "step": 12650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553365559494696e-05,
+      "loss": 0.2377,
+      "step": 12651
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155218489757515e-05,
+      "loss": 0.2394,
+      "step": 12652
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1551004213484147e-05,
+      "loss": 0.2536,
+      "step": 12653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549823507238557e-05,
+      "loss": 0.2486,
+      "step": 12654
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1548642778855243e-05,
+      "loss": 0.2344,
+      "step": 12655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154746202835107e-05,
+      "loss": 0.2455,
+      "step": 12656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546281255742902e-05,
+      "loss": 0.2317,
+      "step": 12657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1545100461047606e-05,
+      "loss": 0.2162,
+      "step": 12658
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154391964428205e-05,
+      "loss": 0.244,
+      "step": 12659
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15427388054631e-05,
+      "loss": 0.2462,
+      "step": 12660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1541557944607626e-05,
+      "loss": 0.2406,
+      "step": 12661
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154037706173249e-05,
+      "loss": 0.2604,
+      "step": 12662
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1539196156854566e-05,
+      "loss": 0.2308,
+      "step": 12663
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1538015229990717e-05,
+      "loss": 0.21,
+      "step": 12664
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536834281157812e-05,
+      "loss": 0.2486,
+      "step": 12665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1535653310372723e-05,
+      "loss": 0.2223,
+      "step": 12666
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1534472317652315e-05,
+      "loss": 0.2348,
+      "step": 12667
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1533291303013459e-05,
+      "loss": 0.2402,
+      "step": 12668
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532110266473026e-05,
+      "loss": 0.2287,
+      "step": 12669
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530929208047885e-05,
+      "loss": 0.2514,
+      "step": 12670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529748127754905e-05,
+      "loss": 0.2325,
+      "step": 12671
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1528567025610959e-05,
+      "loss": 0.2614,
+      "step": 12672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1527385901632914e-05,
+      "loss": 0.2189,
+      "step": 12673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1526204755837647e-05,
+      "loss": 0.2338,
+      "step": 12674
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1525023588242024e-05,
+      "loss": 0.2451,
+      "step": 12675
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523842398862919e-05,
+      "loss": 0.2382,
+      "step": 12676
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1522661187717206e-05,
+      "loss": 0.2098,
+      "step": 12677
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1521479954821755e-05,
+      "loss": 0.2771,
+      "step": 12678
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152029870019344e-05,
+      "loss": 0.2266,
+      "step": 12679
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519117423849132e-05,
+      "loss": 0.2349,
+      "step": 12680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1517936125805704e-05,
+      "loss": 0.2266,
+      "step": 12681
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1516754806080036e-05,
+      "loss": 0.2445,
+      "step": 12682
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1515573464688998e-05,
+      "loss": 0.266,
+      "step": 12683
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151439210164946e-05,
+      "loss": 0.2374,
+      "step": 12684
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513210716978303e-05,
+      "loss": 0.2068,
+      "step": 12685
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1512029310692398e-05,
+      "loss": 0.2533,
+      "step": 12686
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1510847882808622e-05,
+      "loss": 0.247,
+      "step": 12687
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509666433343853e-05,
+      "loss": 0.2505,
+      "step": 12688
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1508484962314963e-05,
+      "loss": 0.2271,
+      "step": 12689
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1507303469738828e-05,
+      "loss": 0.2684,
+      "step": 12690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1506121955632328e-05,
+      "loss": 0.242,
+      "step": 12691
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1504940420012334e-05,
+      "loss": 0.2367,
+      "step": 12692
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503758862895728e-05,
+      "loss": 0.2243,
+      "step": 12693
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1502577284299388e-05,
+      "loss": 0.2257,
+      "step": 12694
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1501395684240189e-05,
+      "loss": 0.2625,
+      "step": 12695
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150021406273501e-05,
+      "loss": 0.2169,
+      "step": 12696
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1499032419800729e-05,
+      "loss": 0.2144,
+      "step": 12697
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1497850755454222e-05,
+      "loss": 0.2279,
+      "step": 12698
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496669069712372e-05,
+      "loss": 0.2508,
+      "step": 12699
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149548736259206e-05,
+      "loss": 0.2415,
+      "step": 12700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1494305634110158e-05,
+      "loss": 0.2267,
+      "step": 12701
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1493123884283555e-05,
+      "loss": 0.2359,
+      "step": 12702
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491942113129122e-05,
+      "loss": 0.252,
+      "step": 12703
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490760320663745e-05,
+      "loss": 0.2441,
+      "step": 12704
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1489578506904307e-05,
+      "loss": 0.2263,
+      "step": 12705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488396671867683e-05,
+      "loss": 0.2337,
+      "step": 12706
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1487214815570754e-05,
+      "loss": 0.2298,
+      "step": 12707
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1486032938030409e-05,
+      "loss": 0.2338,
+      "step": 12708
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1484851039263525e-05,
+      "loss": 0.243,
+      "step": 12709
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1483669119286984e-05,
+      "loss": 0.2365,
+      "step": 12710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1482487178117671e-05,
+      "loss": 0.2216,
+      "step": 12711
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481305215772467e-05,
+      "loss": 0.2495,
+      "step": 12712
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1480123232268254e-05,
+      "loss": 0.2647,
+      "step": 12713
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1478941227621919e-05,
+      "loss": 0.2259,
+      "step": 12714
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1477759201850343e-05,
+      "loss": 0.2345,
+      "step": 12715
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476577154970412e-05,
+      "loss": 0.2269,
+      "step": 12716
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1475395086999008e-05,
+      "loss": 0.2519,
+      "step": 12717
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474212997953018e-05,
+      "loss": 0.2372,
+      "step": 12718
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473030887849326e-05,
+      "loss": 0.264,
+      "step": 12719
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471848756704816e-05,
+      "loss": 0.2011,
+      "step": 12720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470666604536378e-05,
+      "loss": 0.2449,
+      "step": 12721
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1469484431360894e-05,
+      "loss": 0.2447,
+      "step": 12722
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146830223719525e-05,
+      "loss": 0.2294,
+      "step": 12723
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467120022056333e-05,
+      "loss": 0.2336,
+      "step": 12724
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465937785961029e-05,
+      "loss": 0.227,
+      "step": 12725
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464755528926228e-05,
+      "loss": 0.2446,
+      "step": 12726
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1463573250968818e-05,
+      "loss": 0.2236,
+      "step": 12727
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1462390952105683e-05,
+      "loss": 0.2365,
+      "step": 12728
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1461208632353714e-05,
+      "loss": 0.2204,
+      "step": 12729
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460026291729794e-05,
+      "loss": 0.2632,
+      "step": 12730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1458843930250818e-05,
+      "loss": 0.2468,
+      "step": 12731
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457661547933673e-05,
+      "loss": 0.2212,
+      "step": 12732
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1456479144795245e-05,
+      "loss": 0.2323,
+      "step": 12733
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455296720852427e-05,
+      "loss": 0.2593,
+      "step": 12734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145411427612211e-05,
+      "loss": 0.2418,
+      "step": 12735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145293181062118e-05,
+      "loss": 0.2331,
+      "step": 12736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451749324366529e-05,
+      "loss": 0.2393,
+      "step": 12737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450566817375048e-05,
+      "loss": 0.2424,
+      "step": 12738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449384289663628e-05,
+      "loss": 0.2211,
+      "step": 12739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1448201741249158e-05,
+      "loss": 0.2121,
+      "step": 12740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1447019172148533e-05,
+      "loss": 0.2308,
+      "step": 12741
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1445836582378645e-05,
+      "loss": 0.2042,
+      "step": 12742
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444653971956384e-05,
+      "loss": 0.2366,
+      "step": 12743
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1443471340898644e-05,
+      "loss": 0.2365,
+      "step": 12744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1442288689222314e-05,
+      "loss": 0.24,
+      "step": 12745
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144110601694429e-05,
+      "loss": 0.2514,
+      "step": 12746
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1439923324081465e-05,
+      "loss": 0.2177,
+      "step": 12747
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1438740610650738e-05,
+      "loss": 0.2403,
+      "step": 12748
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1437557876668991e-05,
+      "loss": 0.2276,
+      "step": 12749
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436375122153131e-05,
+      "loss": 0.2528,
+      "step": 12750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435192347120043e-05,
+      "loss": 0.2074,
+      "step": 12751
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434009551586624e-05,
+      "loss": 0.2607,
+      "step": 12752
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143282673556977e-05,
+      "loss": 0.2132,
+      "step": 12753
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431643899086381e-05,
+      "loss": 0.2373,
+      "step": 12754
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1430461042153349e-05,
+      "loss": 0.233,
+      "step": 12755
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1429278164787565e-05,
+      "loss": 0.2238,
+      "step": 12756
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142809526700593e-05,
+      "loss": 0.2348,
+      "step": 12757
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142691234882534e-05,
+      "loss": 0.2219,
+      "step": 12758
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142572941026269e-05,
+      "loss": 0.2116,
+      "step": 12759
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1424546451334885e-05,
+      "loss": 0.2224,
+      "step": 12760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1423363472058816e-05,
+      "loss": 0.2213,
+      "step": 12761
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422180472451377e-05,
+      "loss": 0.2297,
+      "step": 12762
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420997452529473e-05,
+      "loss": 0.219,
+      "step": 12763
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419814412309996e-05,
+      "loss": 0.2266,
+      "step": 12764
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141863135180985e-05,
+      "loss": 0.2653,
+      "step": 12765
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1417448271045935e-05,
+      "loss": 0.2376,
+      "step": 12766
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416265170035144e-05,
+      "loss": 0.2444,
+      "step": 12767
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1415082048794379e-05,
+      "loss": 0.2465,
+      "step": 12768
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413898907340542e-05,
+      "loss": 0.2522,
+      "step": 12769
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141271574569053e-05,
+      "loss": 0.2337,
+      "step": 12770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1411532563861246e-05,
+      "loss": 0.2067,
+      "step": 12771
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1410349361869587e-05,
+      "loss": 0.2261,
+      "step": 12772
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1409166139732457e-05,
+      "loss": 0.2619,
+      "step": 12773
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1407982897466758e-05,
+      "loss": 0.2187,
+      "step": 12774
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140679963508939e-05,
+      "loss": 0.2373,
+      "step": 12775
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405616352617253e-05,
+      "loss": 0.2215,
+      "step": 12776
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404433050067253e-05,
+      "loss": 0.2145,
+      "step": 12777
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1403249727456286e-05,
+      "loss": 0.2352,
+      "step": 12778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140206638480126e-05,
+      "loss": 0.2526,
+      "step": 12779
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1400883022119076e-05,
+      "loss": 0.2319,
+      "step": 12780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.139969963942664e-05,
+      "loss": 0.2354,
+      "step": 12781
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1398516236740852e-05,
+      "loss": 0.2523,
+      "step": 12782
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1397332814078616e-05,
+      "loss": 0.2056,
+      "step": 12783
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396149371456838e-05,
+      "loss": 0.2117,
+      "step": 12784
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1394965908892418e-05,
+      "loss": 0.2308,
+      "step": 12785
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393782426402267e-05,
+      "loss": 0.2423,
+      "step": 12786
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392598924003286e-05,
+      "loss": 0.2286,
+      "step": 12787
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.139141540171238e-05,
+      "loss": 0.2146,
+      "step": 12788
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.139023185954646e-05,
+      "loss": 0.242,
+      "step": 12789
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389048297522422e-05,
+      "loss": 0.2419,
+      "step": 12790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1387864715657178e-05,
+      "loss": 0.2375,
+      "step": 12791
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386681113967633e-05,
+      "loss": 0.2524,
+      "step": 12792
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385497492470697e-05,
+      "loss": 0.2168,
+      "step": 12793
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384313851183272e-05,
+      "loss": 0.2488,
+      "step": 12794
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1383130190122269e-05,
+      "loss": 0.2285,
+      "step": 12795
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381946509304592e-05,
+      "loss": 0.2186,
+      "step": 12796
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380762808747148e-05,
+      "loss": 0.2381,
+      "step": 12797
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137957908846685e-05,
+      "loss": 0.2208,
+      "step": 12798
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378395348480604e-05,
+      "loss": 0.2238,
+      "step": 12799
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1377211588805318e-05,
+      "loss": 0.2146,
+      "step": 12800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1376027809457901e-05,
+      "loss": 0.2182,
+      "step": 12801
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374844010455262e-05,
+      "loss": 0.232,
+      "step": 12802
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1373660191814309e-05,
+      "loss": 0.2362,
+      "step": 12803
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1372476353551953e-05,
+      "loss": 0.2252,
+      "step": 12804
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137129249568511e-05,
+      "loss": 0.2472,
+      "step": 12805
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1370108618230678e-05,
+      "loss": 0.2401,
+      "step": 12806
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1368924721205579e-05,
+      "loss": 0.1991,
+      "step": 12807
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367740804626714e-05,
+      "loss": 0.227,
+      "step": 12808
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366556868511003e-05,
+      "loss": 0.2149,
+      "step": 12809
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365372912875352e-05,
+      "loss": 0.2401,
+      "step": 12810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1364188937736675e-05,
+      "loss": 0.2323,
+      "step": 12811
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1363004943111881e-05,
+      "loss": 0.2525,
+      "step": 12812
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1361820929017884e-05,
+      "loss": 0.2403,
+      "step": 12813
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13606368954716e-05,
+      "loss": 0.2252,
+      "step": 12814
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359452842489936e-05,
+      "loss": 0.2219,
+      "step": 12815
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135826877008981e-05,
+      "loss": 0.211,
+      "step": 12816
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135708467828813e-05,
+      "loss": 0.2537,
+      "step": 12817
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355900567101812e-05,
+      "loss": 0.2134,
+      "step": 12818
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1354716436547771e-05,
+      "loss": 0.2289,
+      "step": 12819
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135353228664292e-05,
+      "loss": 0.24,
+      "step": 12820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1352348117404177e-05,
+      "loss": 0.2257,
+      "step": 12821
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351163928848454e-05,
+      "loss": 0.1937,
+      "step": 12822
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349979720992663e-05,
+      "loss": 0.2231,
+      "step": 12823
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1348795493853722e-05,
+      "loss": 0.2196,
+      "step": 12824
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1347611247448544e-05,
+      "loss": 0.2359,
+      "step": 12825
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346426981794053e-05,
+      "loss": 0.2197,
+      "step": 12826
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1345242696907158e-05,
+      "loss": 0.2365,
+      "step": 12827
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344058392804774e-05,
+      "loss": 0.2234,
+      "step": 12828
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134287406950382e-05,
+      "loss": 0.2585,
+      "step": 12829
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1341689727021213e-05,
+      "loss": 0.2485,
+      "step": 12830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134050536537387e-05,
+      "loss": 0.2244,
+      "step": 12831
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339320984578712e-05,
+      "loss": 0.2486,
+      "step": 12832
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338136584652653e-05,
+      "loss": 0.2512,
+      "step": 12833
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133695216561261e-05,
+      "loss": 0.2503,
+      "step": 12834
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1335767727475504e-05,
+      "loss": 0.2442,
+      "step": 12835
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1334583270258252e-05,
+      "loss": 0.2234,
+      "step": 12836
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1333398793977772e-05,
+      "loss": 0.2258,
+      "step": 12837
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332214298650988e-05,
+      "loss": 0.2334,
+      "step": 12838
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133102978429481e-05,
+      "loss": 0.2327,
+      "step": 12839
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132984525092617e-05,
+      "loss": 0.2148,
+      "step": 12840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328660698561978e-05,
+      "loss": 0.2353,
+      "step": 12841
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327476127219155e-05,
+      "loss": 0.217,
+      "step": 12842
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1326291536914628e-05,
+      "loss": 0.2185,
+      "step": 12843
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325106927665311e-05,
+      "loss": 0.2275,
+      "step": 12844
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323922299488128e-05,
+      "loss": 0.2326,
+      "step": 12845
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13227376524e-05,
+      "loss": 0.2334,
+      "step": 12846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132155298641785e-05,
+      "loss": 0.2235,
+      "step": 12847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1320368301558595e-05,
+      "loss": 0.2433,
+      "step": 12848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319183597839163e-05,
+      "loss": 0.2496,
+      "step": 12849
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1317998875276472e-05,
+      "loss": 0.246,
+      "step": 12850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1316814133887447e-05,
+      "loss": 0.2405,
+      "step": 12851
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315629373689008e-05,
+      "loss": 0.233,
+      "step": 12852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314444594698082e-05,
+      "loss": 0.2418,
+      "step": 12853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313259796931592e-05,
+      "loss": 0.2163,
+      "step": 12854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131207498040646e-05,
+      "loss": 0.2328,
+      "step": 12855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131089014513961e-05,
+      "loss": 0.2437,
+      "step": 12856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1309705291147965e-05,
+      "loss": 0.2265,
+      "step": 12857
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130852041844845e-05,
+      "loss": 0.2192,
+      "step": 12858
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307335527057995e-05,
+      "loss": 0.2286,
+      "step": 12859
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130615061699352e-05,
+      "loss": 0.235,
+      "step": 12860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130496568827195e-05,
+      "loss": 0.2497,
+      "step": 12861
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1303780740910209e-05,
+      "loss": 0.2392,
+      "step": 12862
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1302595774925226e-05,
+      "loss": 0.2308,
+      "step": 12863
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130141079033393e-05,
+      "loss": 0.2403,
+      "step": 12864
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1300225787153243e-05,
+      "loss": 0.2228,
+      "step": 12865
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1299040765400092e-05,
+      "loss": 0.2346,
+      "step": 12866
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1297855725091405e-05,
+      "loss": 0.2268,
+      "step": 12867
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1296670666244108e-05,
+      "loss": 0.2104,
+      "step": 12868
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295485588875126e-05,
+      "loss": 0.2414,
+      "step": 12869
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294300493001391e-05,
+      "loss": 0.2314,
+      "step": 12870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1293115378639834e-05,
+      "loss": 0.2543,
+      "step": 12871
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1291930245807374e-05,
+      "loss": 0.2663,
+      "step": 12872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1290745094520947e-05,
+      "loss": 0.2507,
+      "step": 12873
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1289559924797477e-05,
+      "loss": 0.2205,
+      "step": 12874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1288374736653896e-05,
+      "loss": 0.2296,
+      "step": 12875
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1287189530107134e-05,
+      "loss": 0.213,
+      "step": 12876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286004305174117e-05,
+      "loss": 0.2263,
+      "step": 12877
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1284819061871777e-05,
+      "loss": 0.2672,
+      "step": 12878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283633800217042e-05,
+      "loss": 0.2403,
+      "step": 12879
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1282448520226847e-05,
+      "loss": 0.2436,
+      "step": 12880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281263221918118e-05,
+      "loss": 0.2253,
+      "step": 12881
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280077905307787e-05,
+      "loss": 0.2776,
+      "step": 12882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278892570412784e-05,
+      "loss": 0.2307,
+      "step": 12883
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277707217250044e-05,
+      "loss": 0.2412,
+      "step": 12884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1276521845836497e-05,
+      "loss": 0.2198,
+      "step": 12885
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1275336456189074e-05,
+      "loss": 0.2241,
+      "step": 12886
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274151048324706e-05,
+      "loss": 0.2512,
+      "step": 12887
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1272965622260328e-05,
+      "loss": 0.241,
+      "step": 12888
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271780178012872e-05,
+      "loss": 0.2407,
+      "step": 12889
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1270594715599267e-05,
+      "loss": 0.2319,
+      "step": 12890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269409235036453e-05,
+      "loss": 0.2104,
+      "step": 12891
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126822373634136e-05,
+      "loss": 0.2355,
+      "step": 12892
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126703821953092e-05,
+      "loss": 0.2192,
+      "step": 12893
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265852684622071e-05,
+      "loss": 0.226,
+      "step": 12894
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1264667131631746e-05,
+      "loss": 0.2291,
+      "step": 12895
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263481560576875e-05,
+      "loss": 0.2287,
+      "step": 12896
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1262295971474395e-05,
+      "loss": 0.2623,
+      "step": 12897
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261110364341246e-05,
+      "loss": 0.2329,
+      "step": 12898
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259924739194358e-05,
+      "loss": 0.2322,
+      "step": 12899
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1258739096050668e-05,
+      "loss": 0.2378,
+      "step": 12900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125755343492711e-05,
+      "loss": 0.2578,
+      "step": 12901
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1256367755840626e-05,
+      "loss": 0.2581,
+      "step": 12902
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255182058808143e-05,
+      "loss": 0.2364,
+      "step": 12903
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1253996343846604e-05,
+      "loss": 0.2241,
+      "step": 12904
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252810610972946e-05,
+      "loss": 0.2173,
+      "step": 12905
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1251624860204104e-05,
+      "loss": 0.251,
+      "step": 12906
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250439091557013e-05,
+      "loss": 0.2167,
+      "step": 12907
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1249253305048615e-05,
+      "loss": 0.2801,
+      "step": 12908
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1248067500695845e-05,
+      "loss": 0.2189,
+      "step": 12909
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1246881678515645e-05,
+      "loss": 0.1999,
+      "step": 12910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245695838524946e-05,
+      "loss": 0.2326,
+      "step": 12911
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244509980740693e-05,
+      "loss": 0.2464,
+      "step": 12912
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243324105179825e-05,
+      "loss": 0.2562,
+      "step": 12913
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1242138211859275e-05,
+      "loss": 0.2216,
+      "step": 12914
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240952300795989e-05,
+      "loss": 0.2306,
+      "step": 12915
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1239766372006902e-05,
+      "loss": 0.2261,
+      "step": 12916
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238580425508956e-05,
+      "loss": 0.2477,
+      "step": 12917
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123739446131909e-05,
+      "loss": 0.2437,
+      "step": 12918
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1236208479454248e-05,
+      "loss": 0.2383,
+      "step": 12919
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1235022479931363e-05,
+      "loss": 0.251,
+      "step": 12920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233836462767387e-05,
+      "loss": 0.2382,
+      "step": 12921
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1232650427979248e-05,
+      "loss": 0.2343,
+      "step": 12922
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1231464375583896e-05,
+      "loss": 0.225,
+      "step": 12923
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123027830559827e-05,
+      "loss": 0.2254,
+      "step": 12924
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1229092218039315e-05,
+      "loss": 0.2505,
+      "step": 12925
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122790611292397e-05,
+      "loss": 0.2482,
+      "step": 12926
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226719990269177e-05,
+      "loss": 0.2267,
+      "step": 12927
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122553385009188e-05,
+      "loss": 0.2481,
+      "step": 12928
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122434769240902e-05,
+      "loss": 0.2332,
+      "step": 12929
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223161517237542e-05,
+      "loss": 0.2495,
+      "step": 12930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122197532459439e-05,
+      "loss": 0.2268,
+      "step": 12931
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1220789114496504e-05,
+      "loss": 0.2634,
+      "step": 12932
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1219602886960834e-05,
+      "loss": 0.2237,
+      "step": 12933
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1218416642004317e-05,
+      "loss": 0.2277,
+      "step": 12934
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217230379643903e-05,
+      "loss": 0.2479,
+      "step": 12935
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216044099896533e-05,
+      "loss": 0.237,
+      "step": 12936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1214857802779153e-05,
+      "loss": 0.2385,
+      "step": 12937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121367148830871e-05,
+      "loss": 0.2127,
+      "step": 12938
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212485156502148e-05,
+      "loss": 0.2506,
+      "step": 12939
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121129880737641e-05,
+      "loss": 0.2283,
+      "step": 12940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210112440948445e-05,
+      "loss": 0.2408,
+      "step": 12941
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208926057235197e-05,
+      "loss": 0.237,
+      "step": 12942
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207739656253615e-05,
+      "loss": 0.241,
+      "step": 12943
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206553238020646e-05,
+      "loss": 0.2245,
+      "step": 12944
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.2457,
+      "step": 12945
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204180349868325e-05,
+      "loss": 0.2264,
+      "step": 12946
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1202993879982868e-05,
+      "loss": 0.2192,
+      "step": 12947
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1201807392913814e-05,
+      "loss": 0.214,
+      "step": 12948
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200620888678107e-05,
+      "loss": 0.2406,
+      "step": 12949
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1199434367292694e-05,
+      "loss": 0.2386,
+      "step": 12950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1198247828774524e-05,
+      "loss": 0.2096,
+      "step": 12951
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1197061273140551e-05,
+      "loss": 0.2397,
+      "step": 12952
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1195874700407717e-05,
+      "loss": 0.2101,
+      "step": 12953
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1194688110592976e-05,
+      "loss": 0.2284,
+      "step": 12954
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193501503713272e-05,
+      "loss": 0.2262,
+      "step": 12955
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1192314879785556e-05,
+      "loss": 0.2636,
+      "step": 12956
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1191128238826781e-05,
+      "loss": 0.2469,
+      "step": 12957
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1189941580853894e-05,
+      "loss": 0.2202,
+      "step": 12958
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188754905883849e-05,
+      "loss": 0.2147,
+      "step": 12959
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1187568213933596e-05,
+      "loss": 0.2164,
+      "step": 12960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.118638150502008e-05,
+      "loss": 0.2345,
+      "step": 12961
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1185194779160253e-05,
+      "loss": 0.2486,
+      "step": 12962
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118400803637107e-05,
+      "loss": 0.2215,
+      "step": 12963
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1182821276669485e-05,
+      "loss": 0.2418,
+      "step": 12964
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1181634500072444e-05,
+      "loss": 0.2382,
+      "step": 12965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180447706596901e-05,
+      "loss": 0.2286,
+      "step": 12966
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179260896259809e-05,
+      "loss": 0.2325,
+      "step": 12967
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178074069078118e-05,
+      "loss": 0.2611,
+      "step": 12968
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1176887225068783e-05,
+      "loss": 0.2254,
+      "step": 12969
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1175700364248755e-05,
+      "loss": 0.2444,
+      "step": 12970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117451348663499e-05,
+      "loss": 0.235,
+      "step": 12971
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117332659224444e-05,
+      "loss": 0.2573,
+      "step": 12972
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117213968109406e-05,
+      "loss": 0.2322,
+      "step": 12973
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170952753200796e-05,
+      "loss": 0.241,
+      "step": 12974
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1169765808581612e-05,
+      "loss": 0.2258,
+      "step": 12975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116857884725346e-05,
+      "loss": 0.267,
+      "step": 12976
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167391869233292e-05,
+      "loss": 0.2257,
+      "step": 12977
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1166204874538065e-05,
+      "loss": 0.224,
+      "step": 12978
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165017863184733e-05,
+      "loss": 0.2397,
+      "step": 12979
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116383083519025e-05,
+      "loss": 0.2468,
+      "step": 12980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1162643790571574e-05,
+      "loss": 0.2183,
+      "step": 12981
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116145672934566e-05,
+      "loss": 0.2314,
+      "step": 12982
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1160269651529463e-05,
+      "loss": 0.226,
+      "step": 12983
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1159082557139941e-05,
+      "loss": 0.2527,
+      "step": 12984
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157895446194051e-05,
+      "loss": 0.2345,
+      "step": 12985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1156708318708747e-05,
+      "loss": 0.223,
+      "step": 12986
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155521174700987e-05,
+      "loss": 0.2446,
+      "step": 12987
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154334014187728e-05,
+      "loss": 0.2487,
+      "step": 12988
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115314683718593e-05,
+      "loss": 0.2428,
+      "step": 12989
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1151959643712545e-05,
+      "loss": 0.2283,
+      "step": 12990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1150772433784536e-05,
+      "loss": 0.2385,
+      "step": 12991
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114958520741886e-05,
+      "loss": 0.2362,
+      "step": 12992
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1148397964632477e-05,
+      "loss": 0.2319,
+      "step": 12993
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1147210705442342e-05,
+      "loss": 0.2248,
+      "step": 12994
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1146023429865413e-05,
+      "loss": 0.2368,
+      "step": 12995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1144836137918654e-05,
+      "loss": 0.223,
+      "step": 12996
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114364882961902e-05,
+      "loss": 0.2303,
+      "step": 12997
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142461504983478e-05,
+      "loss": 0.2148,
+      "step": 12998
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141274164028979e-05,
+      "loss": 0.2477,
+      "step": 12999
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140086806772484e-05,
+      "loss": 0.247,
+      "step": 13000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138899433230956e-05,
+      "loss": 0.2247,
+      "step": 13001
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1137712043421356e-05,
+      "loss": 0.2414,
+      "step": 13002
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136524637360644e-05,
+      "loss": 0.2356,
+      "step": 13003
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113533721506578e-05,
+      "loss": 0.2299,
+      "step": 13004
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134149776553726e-05,
+      "loss": 0.2538,
+      "step": 13005
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1132962321841443e-05,
+      "loss": 0.2421,
+      "step": 13006
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131774850945892e-05,
+      "loss": 0.2427,
+      "step": 13007
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1130587363884035e-05,
+      "loss": 0.2399,
+      "step": 13008
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129399860672839e-05,
+      "loss": 0.2407,
+      "step": 13009
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128212341329257e-05,
+      "loss": 0.2464,
+      "step": 13010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1127024805870259e-05,
+      "loss": 0.2209,
+      "step": 13011
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125837254312805e-05,
+      "loss": 0.2274,
+      "step": 13012
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1124649686673862e-05,
+      "loss": 0.2566,
+      "step": 13013
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123462102970384e-05,
+      "loss": 0.2408,
+      "step": 13014
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1122274503219344e-05,
+      "loss": 0.2454,
+      "step": 13015
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11210868874377e-05,
+      "loss": 0.2322,
+      "step": 13016
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119899255642416e-05,
+      "loss": 0.2258,
+      "step": 13017
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1118711607850463e-05,
+      "loss": 0.2396,
+      "step": 13018
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117523944078799e-05,
+      "loss": 0.2666,
+      "step": 13019
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111633626434439e-05,
+      "loss": 0.2351,
+      "step": 13020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1115148568664198e-05,
+      "loss": 0.2428,
+      "step": 13021
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113960857055193e-05,
+      "loss": 0.2306,
+      "step": 13022
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112773129534337e-05,
+      "loss": 0.2442,
+      "step": 13023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1111585386118598e-05,
+      "loss": 0.2372,
+      "step": 13024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111039762682494e-05,
+      "loss": 0.231,
+      "step": 13025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1109209851670332e-05,
+      "loss": 0.2349,
+      "step": 13026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1108022060671736e-05,
+      "loss": 0.227,
+      "step": 13027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110683425384612e-05,
+      "loss": 0.2392,
+      "step": 13028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110564643121045e-05,
+      "loss": 0.2546,
+      "step": 13029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1104458592781695e-05,
+      "loss": 0.2409,
+      "step": 13030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1103270738576821e-05,
+      "loss": 0.2381,
+      "step": 13031
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102082868612796e-05,
+      "loss": 0.2325,
+      "step": 13032
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1100894982906584e-05,
+      "loss": 0.2305,
+      "step": 13033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099707081475158e-05,
+      "loss": 0.2353,
+      "step": 13034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098519164335482e-05,
+      "loss": 0.25,
+      "step": 13035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109733123150453e-05,
+      "loss": 0.2122,
+      "step": 13036
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109614328299926e-05,
+      "loss": 0.2289,
+      "step": 13037
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094955318836654e-05,
+      "loss": 0.2139,
+      "step": 13038
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109376733903367e-05,
+      "loss": 0.2249,
+      "step": 13039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1092579343607279e-05,
+      "loss": 0.2571,
+      "step": 13040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1091391332574455e-05,
+      "loss": 0.2626,
+      "step": 13041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1090203305952166e-05,
+      "loss": 0.2125,
+      "step": 13042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089015263757379e-05,
+      "loss": 0.2197,
+      "step": 13043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1087827206007064e-05,
+      "loss": 0.2298,
+      "step": 13044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086639132718197e-05,
+      "loss": 0.2384,
+      "step": 13045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1085451043907745e-05,
+      "loss": 0.2196,
+      "step": 13046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084262939592675e-05,
+      "loss": 0.2334,
+      "step": 13047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1083074819789965e-05,
+      "loss": 0.2087,
+      "step": 13048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108188668451658e-05,
+      "loss": 0.2475,
+      "step": 13049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108069853378949e-05,
+      "loss": 0.2367,
+      "step": 13050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1079510367625677e-05,
+      "loss": 0.245,
+      "step": 13051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078322186042104e-05,
+      "loss": 0.2322,
+      "step": 13052
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1077133989055744e-05,
+      "loss": 0.2177,
+      "step": 13053
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107594577668357e-05,
+      "loss": 0.2347,
+      "step": 13054
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1074757548942555e-05,
+      "loss": 0.2336,
+      "step": 13055
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1073569305849672e-05,
+      "loss": 0.2289,
+      "step": 13056
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072381047421892e-05,
+      "loss": 0.223,
+      "step": 13057
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1071192773676192e-05,
+      "loss": 0.2159,
+      "step": 13058
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070004484629543e-05,
+      "loss": 0.2278,
+      "step": 13059
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1068816180298915e-05,
+      "loss": 0.2175,
+      "step": 13060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1067627860701289e-05,
+      "loss": 0.2373,
+      "step": 13061
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1066439525853632e-05,
+      "loss": 0.2414,
+      "step": 13062
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065251175772921e-05,
+      "loss": 0.2564,
+      "step": 13063
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064062810476134e-05,
+      "loss": 0.2238,
+      "step": 13064
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1062874429980244e-05,
+      "loss": 0.2484,
+      "step": 13065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061686034302222e-05,
+      "loss": 0.2352,
+      "step": 13066
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060497623459043e-05,
+      "loss": 0.2263,
+      "step": 13067
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1059309197467685e-05,
+      "loss": 0.2293,
+      "step": 13068
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058120756345126e-05,
+      "loss": 0.2461,
+      "step": 13069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1056932300108337e-05,
+      "loss": 0.2277,
+      "step": 13070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055743828774297e-05,
+      "loss": 0.2532,
+      "step": 13071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105455534235998e-05,
+      "loss": 0.2494,
+      "step": 13072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1053366840882365e-05,
+      "loss": 0.2192,
+      "step": 13073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1052178324358425e-05,
+      "loss": 0.2327,
+      "step": 13074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105098979280514e-05,
+      "loss": 0.2295,
+      "step": 13075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049801246239484e-05,
+      "loss": 0.1908,
+      "step": 13076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104861268467844e-05,
+      "loss": 0.238,
+      "step": 13077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047424108138978e-05,
+      "loss": 0.2042,
+      "step": 13078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104623551663808e-05,
+      "loss": 0.2225,
+      "step": 13079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045046910192722e-05,
+      "loss": 0.2192,
+      "step": 13080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1043858288819886e-05,
+      "loss": 0.2384,
+      "step": 13081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1042669652536544e-05,
+      "loss": 0.2458,
+      "step": 13082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104148100135968e-05,
+      "loss": 0.2238,
+      "step": 13083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040292335306273e-05,
+      "loss": 0.2323,
+      "step": 13084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1039103654393295e-05,
+      "loss": 0.2348,
+      "step": 13085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1037914958637732e-05,
+      "loss": 0.2507,
+      "step": 13086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1036726248056562e-05,
+      "loss": 0.2279,
+      "step": 13087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1035537522666761e-05,
+      "loss": 0.2343,
+      "step": 13088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034348782485313e-05,
+      "loss": 0.252,
+      "step": 13089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033160027529198e-05,
+      "loss": 0.2129,
+      "step": 13090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031971257815392e-05,
+      "loss": 0.2207,
+      "step": 13091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030782473360882e-05,
+      "loss": 0.2333,
+      "step": 13092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029593674182641e-05,
+      "loss": 0.2138,
+      "step": 13093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1028404860297655e-05,
+      "loss": 0.2303,
+      "step": 13094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1027216031722902e-05,
+      "loss": 0.2227,
+      "step": 13095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1026027188475366e-05,
+      "loss": 0.24,
+      "step": 13096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102483833057203e-05,
+      "loss": 0.2174,
+      "step": 13097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023649458029873e-05,
+      "loss": 0.2209,
+      "step": 13098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022460570865873e-05,
+      "loss": 0.2289,
+      "step": 13099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1021271669097019e-05,
+      "loss": 0.2389,
+      "step": 13100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1020082752740286e-05,
+      "loss": 0.2448,
+      "step": 13101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018893821812665e-05,
+      "loss": 0.2248,
+      "step": 13102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1017704876331135e-05,
+      "loss": 0.2325,
+      "step": 13103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101651591631268e-05,
+      "loss": 0.216,
+      "step": 13104
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015326941774277e-05,
+      "loss": 0.1985,
+      "step": 13105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1014137952732914e-05,
+      "loss": 0.2455,
+      "step": 13106
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012948949205577e-05,
+      "loss": 0.2285,
+      "step": 13107
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101175993120925e-05,
+      "loss": 0.2432,
+      "step": 13108
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1010570898760911e-05,
+      "loss": 0.2257,
+      "step": 13109
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009381851877547e-05,
+      "loss": 0.2545,
+      "step": 13110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1008192790576144e-05,
+      "loss": 0.2254,
+      "step": 13111
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1007003714873686e-05,
+      "loss": 0.2556,
+      "step": 13112
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005814624787154e-05,
+      "loss": 0.2356,
+      "step": 13113
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.100462552033354e-05,
+      "loss": 0.2468,
+      "step": 13114
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1003436401529823e-05,
+      "loss": 0.2367,
+      "step": 13115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002247268392994e-05,
+      "loss": 0.2418,
+      "step": 13116
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001058120940034e-05,
+      "loss": 0.2547,
+      "step": 13117
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099986895918793e-05,
+      "loss": 0.2347,
+      "step": 13118
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998679783153667e-05,
+      "loss": 0.2257,
+      "step": 13119
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0997490592854233e-05,
+      "loss": 0.2333,
+      "step": 13120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996301388306616e-05,
+      "loss": 0.2223,
+      "step": 13121
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995112169527797e-05,
+      "loss": 0.2306,
+      "step": 13122
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0993922936534768e-05,
+      "loss": 0.251,
+      "step": 13123
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0992733689344517e-05,
+      "loss": 0.229,
+      "step": 13124
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0991544427974029e-05,
+      "loss": 0.2168,
+      "step": 13125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0990355152440288e-05,
+      "loss": 0.2261,
+      "step": 13126
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989165862760285e-05,
+      "loss": 0.2328,
+      "step": 13127
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0987976558951007e-05,
+      "loss": 0.2479,
+      "step": 13128
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986787241029445e-05,
+      "loss": 0.2545,
+      "step": 13129
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0985597909012585e-05,
+      "loss": 0.2504,
+      "step": 13130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0984408562917417e-05,
+      "loss": 0.2422,
+      "step": 13131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983219202760925e-05,
+      "loss": 0.2411,
+      "step": 13132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0982029828560099e-05,
+      "loss": 0.2211,
+      "step": 13133
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0980840440331932e-05,
+      "loss": 0.2393,
+      "step": 13134
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979651038093413e-05,
+      "loss": 0.228,
+      "step": 13135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0978461621861529e-05,
+      "loss": 0.1966,
+      "step": 13136
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0977272191653272e-05,
+      "loss": 0.2486,
+      "step": 13137
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0976082747485626e-05,
+      "loss": 0.2355,
+      "step": 13138
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0974893289375587e-05,
+      "loss": 0.2419,
+      "step": 13139
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0973703817340142e-05,
+      "loss": 0.2117,
+      "step": 13140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972514331396286e-05,
+      "loss": 0.2357,
+      "step": 13141
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971324831561006e-05,
+      "loss": 0.2202,
+      "step": 13142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0970135317851293e-05,
+      "loss": 0.2146,
+      "step": 13143
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096894579028414e-05,
+      "loss": 0.2431,
+      "step": 13144
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0967756248876533e-05,
+      "loss": 0.2405,
+      "step": 13145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0966566693645468e-05,
+      "loss": 0.2485,
+      "step": 13146
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096537712460794e-05,
+      "loss": 0.2448,
+      "step": 13147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964187541780933e-05,
+      "loss": 0.2278,
+      "step": 13148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962997945181443e-05,
+      "loss": 0.2393,
+      "step": 13149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0961808334826465e-05,
+      "loss": 0.2289,
+      "step": 13150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960618710732983e-05,
+      "loss": 0.2526,
+      "step": 13151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0959429072918e-05,
+      "loss": 0.2296,
+      "step": 13152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958239421398503e-05,
+      "loss": 0.2442,
+      "step": 13153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0957049756191485e-05,
+      "loss": 0.2366,
+      "step": 13154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955860077313939e-05,
+      "loss": 0.2331,
+      "step": 13155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0954670384782861e-05,
+      "loss": 0.2354,
+      "step": 13156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095348067861524e-05,
+      "loss": 0.2245,
+      "step": 13157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0952290958828079e-05,
+      "loss": 0.254,
+      "step": 13158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0951101225438361e-05,
+      "loss": 0.2317,
+      "step": 13159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0949911478463084e-05,
+      "loss": 0.2277,
+      "step": 13160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948721717919246e-05,
+      "loss": 0.2553,
+      "step": 13161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0947531943823837e-05,
+      "loss": 0.2397,
+      "step": 13162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0946342156193858e-05,
+      "loss": 0.2409,
+      "step": 13163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0945152355046298e-05,
+      "loss": 0.2131,
+      "step": 13164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943962540398152e-05,
+      "loss": 0.2471,
+      "step": 13165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942772712266419e-05,
+      "loss": 0.2292,
+      "step": 13166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094158287066809e-05,
+      "loss": 0.2403,
+      "step": 13167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0940393015620165e-05,
+      "loss": 0.2342,
+      "step": 13168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939203147139642e-05,
+      "loss": 0.241,
+      "step": 13169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0938013265243512e-05,
+      "loss": 0.2307,
+      "step": 13170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093682336994877e-05,
+      "loss": 0.2515,
+      "step": 13171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0935633461272416e-05,
+      "loss": 0.222,
+      "step": 13172
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0934443539231448e-05,
+      "loss": 0.2135,
+      "step": 13173
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093325360384286e-05,
+      "loss": 0.2186,
+      "step": 13174
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093206365512365e-05,
+      "loss": 0.2472,
+      "step": 13175
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930873693090815e-05,
+      "loss": 0.2376,
+      "step": 13176
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929683717761352e-05,
+      "loss": 0.2445,
+      "step": 13177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0928493729152261e-05,
+      "loss": 0.2293,
+      "step": 13178
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927303727280537e-05,
+      "loss": 0.234,
+      "step": 13179
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092611371216318e-05,
+      "loss": 0.2096,
+      "step": 13180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924923683817188e-05,
+      "loss": 0.2507,
+      "step": 13181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0923733642259558e-05,
+      "loss": 0.2347,
+      "step": 13182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092254358750729e-05,
+      "loss": 0.2412,
+      "step": 13183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092135351957738e-05,
+      "loss": 0.2273,
+      "step": 13184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092016343848683e-05,
+      "loss": 0.2194,
+      "step": 13185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0918973344252641e-05,
+      "loss": 0.2091,
+      "step": 13186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917783236891808e-05,
+      "loss": 0.221,
+      "step": 13187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091659311642133e-05,
+      "loss": 0.2358,
+      "step": 13188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0915402982858212e-05,
+      "loss": 0.2338,
+      "step": 13189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091421283621945e-05,
+      "loss": 0.2289,
+      "step": 13190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0913022676522043e-05,
+      "loss": 0.2322,
+      "step": 13191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0911832503782998e-05,
+      "loss": 0.2406,
+      "step": 13192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0910642318019304e-05,
+      "loss": 0.2463,
+      "step": 13193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0909452119247971e-05,
+      "loss": 0.2556,
+      "step": 13194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908261907485997e-05,
+      "loss": 0.2406,
+      "step": 13195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0907071682750387e-05,
+      "loss": 0.2297,
+      "step": 13196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905881445058137e-05,
+      "loss": 0.2232,
+      "step": 13197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0904691194426246e-05,
+      "loss": 0.2267,
+      "step": 13198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903500930871723e-05,
+      "loss": 0.2333,
+      "step": 13199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0902310654411563e-05,
+      "loss": 0.2127,
+      "step": 13200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0901120365062772e-05,
+      "loss": 0.2676,
+      "step": 13201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899930062842353e-05,
+      "loss": 0.2412,
+      "step": 13202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0898739747767306e-05,
+      "loss": 0.2501,
+      "step": 13203
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897549419854631e-05,
+      "loss": 0.2451,
+      "step": 13204
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896359079121335e-05,
+      "loss": 0.2431,
+      "step": 13205
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0895168725584419e-05,
+      "loss": 0.2589,
+      "step": 13206
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893978359260889e-05,
+      "loss": 0.2328,
+      "step": 13207
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0892787980167746e-05,
+      "loss": 0.2335,
+      "step": 13208
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0891597588321991e-05,
+      "loss": 0.2475,
+      "step": 13209
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089040718374063e-05,
+      "loss": 0.2287,
+      "step": 13210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0889216766440667e-05,
+      "loss": 0.2431,
+      "step": 13211
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0888026336439104e-05,
+      "loss": 0.2381,
+      "step": 13212
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088683589375295e-05,
+      "loss": 0.25,
+      "step": 13213
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0885645438399204e-05,
+      "loss": 0.2302,
+      "step": 13214
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0884454970394871e-05,
+      "loss": 0.2264,
+      "step": 13215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088326448975696e-05,
+      "loss": 0.2664,
+      "step": 13216
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0882073996502471e-05,
+      "loss": 0.2291,
+      "step": 13217
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088088349064841e-05,
+      "loss": 0.2273,
+      "step": 13218
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0879692972211786e-05,
+      "loss": 0.2279,
+      "step": 13219
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878502441209599e-05,
+      "loss": 0.2451,
+      "step": 13220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087731189765886e-05,
+      "loss": 0.2444,
+      "step": 13221
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876121341576571e-05,
+      "loss": 0.2444,
+      "step": 13222
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874930772979739e-05,
+      "loss": 0.2269,
+      "step": 13223
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0873740191885368e-05,
+      "loss": 0.2547,
+      "step": 13224
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0872549598310469e-05,
+      "loss": 0.2482,
+      "step": 13225
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0871358992272046e-05,
+      "loss": 0.2538,
+      "step": 13226
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870168373787102e-05,
+      "loss": 0.2319,
+      "step": 13227
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868977742872651e-05,
+      "loss": 0.2637,
+      "step": 13228
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0867787099545694e-05,
+      "loss": 0.2313,
+      "step": 13229
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866596443823243e-05,
+      "loss": 0.2184,
+      "step": 13230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0865405775722302e-05,
+      "loss": 0.2358,
+      "step": 13231
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0864215095259878e-05,
+      "loss": 0.2422,
+      "step": 13232
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.086302440245298e-05,
+      "loss": 0.2518,
+      "step": 13233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0861833697318615e-05,
+      "loss": 0.2277,
+      "step": 13234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860642979873797e-05,
+      "loss": 0.2364,
+      "step": 13235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0859452250135526e-05,
+      "loss": 0.2183,
+      "step": 13236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0858261508120814e-05,
+      "loss": 0.255,
+      "step": 13237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857070753846668e-05,
+      "loss": 0.2229,
+      "step": 13238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0855879987330097e-05,
+      "loss": 0.2325,
+      "step": 13239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854689208588114e-05,
+      "loss": 0.2456,
+      "step": 13240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0853498417637726e-05,
+      "loss": 0.2129,
+      "step": 13241
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085230761449594e-05,
+      "loss": 0.2485,
+      "step": 13242
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851116799179766e-05,
+      "loss": 0.2136,
+      "step": 13243
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849925971706217e-05,
+      "loss": 0.2406,
+      "step": 13244
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0848735132092297e-05,
+      "loss": 0.2495,
+      "step": 13245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0847544280355022e-05,
+      "loss": 0.2512,
+      "step": 13246
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846353416511399e-05,
+      "loss": 0.2393,
+      "step": 13247
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845162540578438e-05,
+      "loss": 0.1932,
+      "step": 13248
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0843971652573152e-05,
+      "loss": 0.2367,
+      "step": 13249
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842780752512547e-05,
+      "loss": 0.2411,
+      "step": 13250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841589840413638e-05,
+      "loss": 0.2423,
+      "step": 13251
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0840398916293437e-05,
+      "loss": 0.2284,
+      "step": 13252
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0839207980168951e-05,
+      "loss": 0.2097,
+      "step": 13253
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0838017032057194e-05,
+      "loss": 0.2274,
+      "step": 13254
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083682607197518e-05,
+      "loss": 0.2281,
+      "step": 13255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0835635099939912e-05,
+      "loss": 0.2467,
+      "step": 13256
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083444411596841e-05,
+      "loss": 0.2723,
+      "step": 13257
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0833253120077688e-05,
+      "loss": 0.2312,
+      "step": 13258
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832062112284748e-05,
+      "loss": 0.2445,
+      "step": 13259
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0830871092606609e-05,
+      "loss": 0.2345,
+      "step": 13260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829680061060284e-05,
+      "loss": 0.2272,
+      "step": 13261
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828489017662786e-05,
+      "loss": 0.226,
+      "step": 13262
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0827297962431128e-05,
+      "loss": 0.2358,
+      "step": 13263
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0826106895382316e-05,
+      "loss": 0.2363,
+      "step": 13264
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082491581653337e-05,
+      "loss": 0.2135,
+      "step": 13265
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823724725901303e-05,
+      "loss": 0.2388,
+      "step": 13266
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822533623503128e-05,
+      "loss": 0.2551,
+      "step": 13267
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082134250935586e-05,
+      "loss": 0.2262,
+      "step": 13268
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082015138347651e-05,
+      "loss": 0.2448,
+      "step": 13269
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0818960245882096e-05,
+      "loss": 0.2283,
+      "step": 13270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817769096589625e-05,
+      "loss": 0.2095,
+      "step": 13271
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081657793561612e-05,
+      "loss": 0.2382,
+      "step": 13272
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815386762978588e-05,
+      "loss": 0.2433,
+      "step": 13273
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0814195578694052e-05,
+      "loss": 0.2235,
+      "step": 13274
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813004382779522e-05,
+      "loss": 0.2479,
+      "step": 13275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0811813175252009e-05,
+      "loss": 0.249,
+      "step": 13276
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0810621956128535e-05,
+      "loss": 0.225,
+      "step": 13277
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809430725426112e-05,
+      "loss": 0.2352,
+      "step": 13278
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0808239483161758e-05,
+      "loss": 0.2273,
+      "step": 13279
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807048229352488e-05,
+      "loss": 0.2245,
+      "step": 13280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0805856964015315e-05,
+      "loss": 0.2039,
+      "step": 13281
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0804665687167262e-05,
+      "loss": 0.2582,
+      "step": 13282
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803474398825336e-05,
+      "loss": 0.2436,
+      "step": 13283
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0802283099006559e-05,
+      "loss": 0.2226,
+      "step": 13284
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801091787727948e-05,
+      "loss": 0.2428,
+      "step": 13285
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799900465006517e-05,
+      "loss": 0.2178,
+      "step": 13286
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0798709130859284e-05,
+      "loss": 0.2337,
+      "step": 13287
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0797517785303269e-05,
+      "loss": 0.215,
+      "step": 13288
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0796326428355481e-05,
+      "loss": 0.2231,
+      "step": 13289
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0795135060032947e-05,
+      "loss": 0.2396,
+      "step": 13290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079394368035268e-05,
+      "loss": 0.24,
+      "step": 13291
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0792752289331698e-05,
+      "loss": 0.2276,
+      "step": 13292
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0791560886987016e-05,
+      "loss": 0.2366,
+      "step": 13293
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790369473335658e-05,
+      "loss": 0.2294,
+      "step": 13294
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078917804839464e-05,
+      "loss": 0.2129,
+      "step": 13295
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787986612180976e-05,
+      "loss": 0.2327,
+      "step": 13296
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786795164711692e-05,
+      "loss": 0.2429,
+      "step": 13297
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.07856037060038e-05,
+      "loss": 0.2013,
+      "step": 13298
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784412236074322e-05,
+      "loss": 0.2341,
+      "step": 13299
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0783220754940276e-05,
+      "loss": 0.2396,
+      "step": 13300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782029262618683e-05,
+      "loss": 0.2387,
+      "step": 13301
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780837759126562e-05,
+      "loss": 0.23,
+      "step": 13302
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0779646244480927e-05,
+      "loss": 0.2457,
+      "step": 13303
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0778454718698805e-05,
+      "loss": 0.2396,
+      "step": 13304
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777263181797212e-05,
+      "loss": 0.2406,
+      "step": 13305
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0776071633793171e-05,
+      "loss": 0.24,
+      "step": 13306
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774880074703697e-05,
+      "loss": 0.2098,
+      "step": 13307
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077368850454582e-05,
+      "loss": 0.2226,
+      "step": 13308
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0772496923336545e-05,
+      "loss": 0.2558,
+      "step": 13309
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771305331092902e-05,
+      "loss": 0.2325,
+      "step": 13310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770113727831914e-05,
+      "loss": 0.2666,
+      "step": 13311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0768922113570597e-05,
+      "loss": 0.2126,
+      "step": 13312
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0767730488325975e-05,
+      "loss": 0.2225,
+      "step": 13313
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0766538852115072e-05,
+      "loss": 0.2467,
+      "step": 13314
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0765347204954899e-05,
+      "loss": 0.2187,
+      "step": 13315
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0764155546862487e-05,
+      "loss": 0.2564,
+      "step": 13316
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0762963877854852e-05,
+      "loss": 0.2464,
+      "step": 13317
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761772197949022e-05,
+      "loss": 0.2231,
+      "step": 13318
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0760580507162012e-05,
+      "loss": 0.233,
+      "step": 13319
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0759388805510851e-05,
+      "loss": 0.2491,
+      "step": 13320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0758197093012557e-05,
+      "loss": 0.2215,
+      "step": 13321
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0757005369684153e-05,
+      "loss": 0.2408,
+      "step": 13322
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075581363554266e-05,
+      "loss": 0.2214,
+      "step": 13323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0754621890605106e-05,
+      "loss": 0.2374,
+      "step": 13324
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.075343013488851e-05,
+      "loss": 0.225,
+      "step": 13325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0752238368409893e-05,
+      "loss": 0.24,
+      "step": 13326
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0751046591186283e-05,
+      "loss": 0.2175,
+      "step": 13327
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0749854803234702e-05,
+      "loss": 0.2226,
+      "step": 13328
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074866300457217e-05,
+      "loss": 0.2259,
+      "step": 13329
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074747119521572e-05,
+      "loss": 0.2363,
+      "step": 13330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746279375182365e-05,
+      "loss": 0.2387,
+      "step": 13331
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0745087544489132e-05,
+      "loss": 0.2427,
+      "step": 13332
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0743895703153046e-05,
+      "loss": 0.2394,
+      "step": 13333
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0742703851191137e-05,
+      "loss": 0.2156,
+      "step": 13334
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074151198862042e-05,
+      "loss": 0.1992,
+      "step": 13335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0740320115457927e-05,
+      "loss": 0.2359,
+      "step": 13336
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0739128231720677e-05,
+      "loss": 0.2116,
+      "step": 13337
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0737936337425698e-05,
+      "loss": 0.2243,
+      "step": 13338
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0736744432590015e-05,
+      "loss": 0.2403,
+      "step": 13339
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735552517230654e-05,
+      "loss": 0.2375,
+      "step": 13340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734360591364638e-05,
+      "loss": 0.2137,
+      "step": 13341
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0733168655008994e-05,
+      "loss": 0.2254,
+      "step": 13342
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0731976708180745e-05,
+      "loss": 0.2235,
+      "step": 13343
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073078475089692e-05,
+      "loss": 0.223,
+      "step": 13344
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0729592783174545e-05,
+      "loss": 0.2133,
+      "step": 13345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0728400805030647e-05,
+      "loss": 0.2343,
+      "step": 13346
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072720881648225e-05,
+      "loss": 0.2298,
+      "step": 13347
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072601681754638e-05,
+      "loss": 0.2207,
+      "step": 13348
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0724824808240061e-05,
+      "loss": 0.2353,
+      "step": 13349
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723632788580328e-05,
+      "loss": 0.2229,
+      "step": 13350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.07224407585842e-05,
+      "loss": 0.2354,
+      "step": 13351
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072124871826871e-05,
+      "loss": 0.2314,
+      "step": 13352
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720056667650878e-05,
+      "loss": 0.232,
+      "step": 13353
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0718864606747738e-05,
+      "loss": 0.2645,
+      "step": 13354
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717672535576313e-05,
+      "loss": 0.263,
+      "step": 13355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0716480454153633e-05,
+      "loss": 0.2337,
+      "step": 13356
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0715288362496728e-05,
+      "loss": 0.2559,
+      "step": 13357
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071409626062262e-05,
+      "loss": 0.2223,
+      "step": 13358
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0712904148548338e-05,
+      "loss": 0.2387,
+      "step": 13359
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0711712026290914e-05,
+      "loss": 0.227,
+      "step": 13360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0710519893867376e-05,
+      "loss": 0.2267,
+      "step": 13361
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070932775129475e-05,
+      "loss": 0.2396,
+      "step": 13362
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708135598590067e-05,
+      "loss": 0.2444,
+      "step": 13363
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706943435770352e-05,
+      "loss": 0.2403,
+      "step": 13364
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705751262852636e-05,
+      "loss": 0.2233,
+      "step": 13365
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704559079853947e-05,
+      "loss": 0.2306,
+      "step": 13366
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0703366886791317e-05,
+      "loss": 0.2335,
+      "step": 13367
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702174683681777e-05,
+      "loss": 0.2255,
+      "step": 13368
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700982470542349e-05,
+      "loss": 0.2241,
+      "step": 13369
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699790247390067e-05,
+      "loss": 0.2467,
+      "step": 13370
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069859801424196e-05,
+      "loss": 0.2423,
+      "step": 13371
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0697405771115059e-05,
+      "loss": 0.2274,
+      "step": 13372
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0696213518026396e-05,
+      "loss": 0.2347,
+      "step": 13373
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0695021254992998e-05,
+      "loss": 0.2349,
+      "step": 13374
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0693828982031894e-05,
+      "loss": 0.2235,
+      "step": 13375
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692636699160116e-05,
+      "loss": 0.2406,
+      "step": 13376
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0691444406394695e-05,
+      "loss": 0.2327,
+      "step": 13377
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690252103752661e-05,
+      "loss": 0.2337,
+      "step": 13378
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0689059791251049e-05,
+      "loss": 0.2396,
+      "step": 13379
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687867468906885e-05,
+      "loss": 0.2412,
+      "step": 13380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0686675136737202e-05,
+      "loss": 0.2449,
+      "step": 13381
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.068548279475903e-05,
+      "loss": 0.2308,
+      "step": 13382
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.06842904429894e-05,
+      "loss": 0.2114,
+      "step": 13383
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683098081445347e-05,
+      "loss": 0.2384,
+      "step": 13384
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681905710143902e-05,
+      "loss": 0.2355,
+      "step": 13385
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0680713329102095e-05,
+      "loss": 0.2401,
+      "step": 13386
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0679520938336961e-05,
+      "loss": 0.2356,
+      "step": 13387
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678328537865526e-05,
+      "loss": 0.2231,
+      "step": 13388
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677136127704828e-05,
+      "loss": 0.2658,
+      "step": 13389
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0675943707871898e-05,
+      "loss": 0.2543,
+      "step": 13390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674751278383766e-05,
+      "loss": 0.2294,
+      "step": 13391
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673558839257467e-05,
+      "loss": 0.2382,
+      "step": 13392
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0672366390510036e-05,
+      "loss": 0.2411,
+      "step": 13393
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671173932158503e-05,
+      "loss": 0.2275,
+      "step": 13394
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.06699814642199e-05,
+      "loss": 0.2332,
+      "step": 13395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0668788986711268e-05,
+      "loss": 0.2313,
+      "step": 13396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0667596499649628e-05,
+      "loss": 0.2448,
+      "step": 13397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066640400305202e-05,
+      "loss": 0.2217,
+      "step": 13398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066521149693548e-05,
+      "loss": 0.2108,
+      "step": 13399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0664018981317039e-05,
+      "loss": 0.2402,
+      "step": 13400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662826456213731e-05,
+      "loss": 0.2369,
+      "step": 13401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066163392164259e-05,
+      "loss": 0.2191,
+      "step": 13402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0660441377620652e-05,
+      "loss": 0.2631,
+      "step": 13403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0659248824164947e-05,
+      "loss": 0.2236,
+      "step": 13404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658056261292514e-05,
+      "loss": 0.2257,
+      "step": 13405
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0656863689020388e-05,
+      "loss": 0.2619,
+      "step": 13406
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655671107365601e-05,
+      "loss": 0.2467,
+      "step": 13407
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654478516345185e-05,
+      "loss": 0.2272,
+      "step": 13408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0653285915976182e-05,
+      "loss": 0.2114,
+      "step": 13409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652093306275621e-05,
+      "loss": 0.2387,
+      "step": 13410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650900687260541e-05,
+      "loss": 0.2201,
+      "step": 13411
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0649708058947978e-05,
+      "loss": 0.226,
+      "step": 13412
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0648515421354968e-05,
+      "loss": 0.2304,
+      "step": 13413
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.064732277449854e-05,
+      "loss": 0.2215,
+      "step": 13414
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646130118395736e-05,
+      "loss": 0.2743,
+      "step": 13415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064493745306359e-05,
+      "loss": 0.249,
+      "step": 13416
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643744778519139e-05,
+      "loss": 0.2051,
+      "step": 13417
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064255209477942e-05,
+      "loss": 0.2677,
+      "step": 13418
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0641359401861464e-05,
+      "loss": 0.2303,
+      "step": 13419
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0640166699782314e-05,
+      "loss": 0.2267,
+      "step": 13420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638973988559006e-05,
+      "loss": 0.2307,
+      "step": 13421
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637781268208571e-05,
+      "loss": 0.257,
+      "step": 13422
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0636588538748052e-05,
+      "loss": 0.2441,
+      "step": 13423
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0635395800194481e-05,
+      "loss": 0.2354,
+      "step": 13424
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.06342030525649e-05,
+      "loss": 0.2188,
+      "step": 13425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633010295876342e-05,
+      "loss": 0.2139,
+      "step": 13426
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063181753014585e-05,
+      "loss": 0.2262,
+      "step": 13427
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0630624755390452e-05,
+      "loss": 0.2412,
+      "step": 13428
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0629431971627198e-05,
+      "loss": 0.2384,
+      "step": 13429
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628239178873114e-05,
+      "loss": 0.2357,
+      "step": 13430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627046377145244e-05,
+      "loss": 0.2502,
+      "step": 13431
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0625853566460625e-05,
+      "loss": 0.2035,
+      "step": 13432
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624660746836297e-05,
+      "loss": 0.2087,
+      "step": 13433
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623467918289296e-05,
+      "loss": 0.2336,
+      "step": 13434
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0622275080836662e-05,
+      "loss": 0.2185,
+      "step": 13435
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062108223449543e-05,
+      "loss": 0.2619,
+      "step": 13436
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619889379282639e-05,
+      "loss": 0.2269,
+      "step": 13437
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0618696515215335e-05,
+      "loss": 0.2321,
+      "step": 13438
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061750364231055e-05,
+      "loss": 0.2346,
+      "step": 13439
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0616310760585325e-05,
+      "loss": 0.2485,
+      "step": 13440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0615117870056696e-05,
+      "loss": 0.2245,
+      "step": 13441
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613924970741709e-05,
+      "loss": 0.2147,
+      "step": 13442
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0612732062657395e-05,
+      "loss": 0.2413,
+      "step": 13443
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0611539145820801e-05,
+      "loss": 0.2076,
+      "step": 13444
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0610346220248966e-05,
+      "loss": 0.2124,
+      "step": 13445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0609153285958926e-05,
+      "loss": 0.2493,
+      "step": 13446
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607960342967723e-05,
+      "loss": 0.2341,
+      "step": 13447
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0606767391292394e-05,
+      "loss": 0.2517,
+      "step": 13448
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0605574430949983e-05,
+      "loss": 0.2334,
+      "step": 13449
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0604381461957528e-05,
+      "loss": 0.2583,
+      "step": 13450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0603188484332075e-05,
+      "loss": 0.2328,
+      "step": 13451
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601995498090654e-05,
+      "loss": 0.235,
+      "step": 13452
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0600802503250317e-05,
+      "loss": 0.2558,
+      "step": 13453
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0599609499828093e-05,
+      "loss": 0.2309,
+      "step": 13454
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598416487841035e-05,
+      "loss": 0.2203,
+      "step": 13455
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0597223467306177e-05,
+      "loss": 0.2296,
+      "step": 13456
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059603043824056e-05,
+      "loss": 0.2483,
+      "step": 13457
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594837400661224e-05,
+      "loss": 0.2285,
+      "step": 13458
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059364435458522e-05,
+      "loss": 0.2412,
+      "step": 13459
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592451300029576e-05,
+      "loss": 0.2302,
+      "step": 13460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591258237011342e-05,
+      "loss": 0.2196,
+      "step": 13461
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0590065165547561e-05,
+      "loss": 0.2073,
+      "step": 13462
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0588872085655268e-05,
+      "loss": 0.2381,
+      "step": 13463
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0587678997351509e-05,
+      "loss": 0.2487,
+      "step": 13464
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0586485900653328e-05,
+      "loss": 0.2275,
+      "step": 13465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0585292795577764e-05,
+      "loss": 0.2307,
+      "step": 13466
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058409968214186e-05,
+      "loss": 0.2251,
+      "step": 13467
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0582906560362662e-05,
+      "loss": 0.2322,
+      "step": 13468
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581713430257205e-05,
+      "loss": 0.2338,
+      "step": 13469
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0580520291842538e-05,
+      "loss": 0.2388,
+      "step": 13470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579327145135703e-05,
+      "loss": 0.2241,
+      "step": 13471
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0578133990153741e-05,
+      "loss": 0.2452,
+      "step": 13472
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0576940826913697e-05,
+      "loss": 0.2567,
+      "step": 13473
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0575747655432614e-05,
+      "loss": 0.2344,
+      "step": 13474
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574554475727531e-05,
+      "loss": 0.2011,
+      "step": 13475
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0573361287815498e-05,
+      "loss": 0.2577,
+      "step": 13476
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572168091713552e-05,
+      "loss": 0.2407,
+      "step": 13477
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0570974887438745e-05,
+      "loss": 0.2134,
+      "step": 13478
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569781675008114e-05,
+      "loss": 0.2576,
+      "step": 13479
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0568588454438704e-05,
+      "loss": 0.2568,
+      "step": 13480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0567395225747558e-05,
+      "loss": 0.2453,
+      "step": 13481
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566201988951725e-05,
+      "loss": 0.2271,
+      "step": 13482
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0565008744068244e-05,
+      "loss": 0.2348,
+      "step": 13483
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0563815491114165e-05,
+      "loss": 0.2399,
+      "step": 13484
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562622230106525e-05,
+      "loss": 0.2189,
+      "step": 13485
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0561428961062374e-05,
+      "loss": 0.2169,
+      "step": 13486
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560235683998754e-05,
+      "loss": 0.1914,
+      "step": 13487
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0559042398932713e-05,
+      "loss": 0.2394,
+      "step": 13488
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055784910588129e-05,
+      "loss": 0.2287,
+      "step": 13489
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055665580486154e-05,
+      "loss": 0.2242,
+      "step": 13490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555462495890498e-05,
+      "loss": 0.2263,
+      "step": 13491
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0554269178985214e-05,
+      "loss": 0.2478,
+      "step": 13492
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553075854162732e-05,
+      "loss": 0.2132,
+      "step": 13493
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05518825214401e-05,
+      "loss": 0.2379,
+      "step": 13494
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0550689180834362e-05,
+      "loss": 0.2168,
+      "step": 13495
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0549495832362561e-05,
+      "loss": 0.2125,
+      "step": 13496
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0548302476041748e-05,
+      "loss": 0.2709,
+      "step": 13497
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547109111888962e-05,
+      "loss": 0.2197,
+      "step": 13498
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545915739921258e-05,
+      "loss": 0.2478,
+      "step": 13499
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0544722360155674e-05,
+      "loss": 0.2097,
+      "step": 13500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0543528972609264e-05,
+      "loss": 0.2237,
+      "step": 13501
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0542335577299067e-05,
+      "loss": 0.2418,
+      "step": 13502
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541142174242134e-05,
+      "loss": 0.2282,
+      "step": 13503
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0539948763455507e-05,
+      "loss": 0.2158,
+      "step": 13504
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0538755344956241e-05,
+      "loss": 0.2441,
+      "step": 13505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0537561918761375e-05,
+      "loss": 0.2349,
+      "step": 13506
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053636848488796e-05,
+      "loss": 0.235,
+      "step": 13507
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0535175043353039e-05,
+      "loss": 0.2441,
+      "step": 13508
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0533981594173664e-05,
+      "loss": 0.2315,
+      "step": 13509
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532788137366878e-05,
+      "loss": 0.2448,
+      "step": 13510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531594672949734e-05,
+      "loss": 0.2306,
+      "step": 13511
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0530401200939276e-05,
+      "loss": 0.2438,
+      "step": 13512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052920772135255e-05,
+      "loss": 0.2172,
+      "step": 13513
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0528014234206605e-05,
+      "loss": 0.2589,
+      "step": 13514
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526820739518488e-05,
+      "loss": 0.2584,
+      "step": 13515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0525627237305249e-05,
+      "loss": 0.2424,
+      "step": 13516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524433727583937e-05,
+      "loss": 0.2308,
+      "step": 13517
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0523240210371595e-05,
+      "loss": 0.2195,
+      "step": 13518
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0522046685685276e-05,
+      "loss": 0.2444,
+      "step": 13519
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520853153542028e-05,
+      "loss": 0.2153,
+      "step": 13520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0519659613958897e-05,
+      "loss": 0.2498,
+      "step": 13521
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051846606695293e-05,
+      "loss": 0.2513,
+      "step": 13522
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0517272512541183e-05,
+      "loss": 0.2351,
+      "step": 13523
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516078950740697e-05,
+      "loss": 0.2249,
+      "step": 13524
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0514885381568528e-05,
+      "loss": 0.2259,
+      "step": 13525
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0513691805041717e-05,
+      "loss": 0.2138,
+      "step": 13526
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0512498221177319e-05,
+      "loss": 0.2501,
+      "step": 13527
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511304629992382e-05,
+      "loss": 0.2469,
+      "step": 13528
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0510111031503954e-05,
+      "loss": 0.2282,
+      "step": 13529
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0508917425729081e-05,
+      "loss": 0.2394,
+      "step": 13530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050772381268482e-05,
+      "loss": 0.2306,
+      "step": 13531
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506530192388218e-05,
+      "loss": 0.2152,
+      "step": 13532
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0505336564856322e-05,
+      "loss": 0.2664,
+      "step": 13533
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0504142930106186e-05,
+      "loss": 0.2271,
+      "step": 13534
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0502949288154855e-05,
+      "loss": 0.234,
+      "step": 13535
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501755639019382e-05,
+      "loss": 0.2376,
+      "step": 13536
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0500561982716816e-05,
+      "loss": 0.233,
+      "step": 13537
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0499368319264211e-05,
+      "loss": 0.2312,
+      "step": 13538
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498174648678612e-05,
+      "loss": 0.2426,
+      "step": 13539
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0496980970977073e-05,
+      "loss": 0.2279,
+      "step": 13540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0495787286176642e-05,
+      "loss": 0.2427,
+      "step": 13541
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0494593594294369e-05,
+      "loss": 0.238,
+      "step": 13542
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049339989534731e-05,
+      "loss": 0.2165,
+      "step": 13543
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0492206189352512e-05,
+      "loss": 0.2175,
+      "step": 13544
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491012476327024e-05,
+      "loss": 0.2061,
+      "step": 13545
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489818756287902e-05,
+      "loss": 0.2363,
+      "step": 13546
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488625029252191e-05,
+      "loss": 0.2419,
+      "step": 13547
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487431295236948e-05,
+      "loss": 0.2203,
+      "step": 13548
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048623755425922e-05,
+      "loss": 0.2617,
+      "step": 13549
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0485043806336061e-05,
+      "loss": 0.205,
+      "step": 13550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0483850051484524e-05,
+      "loss": 0.2287,
+      "step": 13551
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482656289721658e-05,
+      "loss": 0.2093,
+      "step": 13552
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0481462521064514e-05,
+      "loss": 0.2311,
+      "step": 13553
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0480268745530146e-05,
+      "loss": 0.2492,
+      "step": 13554
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479074963135604e-05,
+      "loss": 0.2223,
+      "step": 13555
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0477881173897944e-05,
+      "loss": 0.2289,
+      "step": 13556
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476687377834212e-05,
+      "loss": 0.2376,
+      "step": 13557
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475493574961466e-05,
+      "loss": 0.2399,
+      "step": 13558
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0474299765296754e-05,
+      "loss": 0.2585,
+      "step": 13559
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047310594885713e-05,
+      "loss": 0.2436,
+      "step": 13560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0471912125659648e-05,
+      "loss": 0.201,
+      "step": 13561
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0470718295721359e-05,
+      "loss": 0.2398,
+      "step": 13562
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469524459059315e-05,
+      "loss": 0.2077,
+      "step": 13563
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046833061569057e-05,
+      "loss": 0.2551,
+      "step": 13564
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0467136765632178e-05,
+      "loss": 0.2588,
+      "step": 13565
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046594290890119e-05,
+      "loss": 0.2311,
+      "step": 13566
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0464749045514662e-05,
+      "loss": 0.2467,
+      "step": 13567
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046355517548964e-05,
+      "loss": 0.24,
+      "step": 13568
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0462361298843185e-05,
+      "loss": 0.2345,
+      "step": 13569
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461167415592345e-05,
+      "loss": 0.2312,
+      "step": 13570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0459973525754181e-05,
+      "loss": 0.2109,
+      "step": 13571
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0458779629345739e-05,
+      "loss": 0.2545,
+      "step": 13572
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0457585726384077e-05,
+      "loss": 0.2169,
+      "step": 13573
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0456391816886242e-05,
+      "loss": 0.2509,
+      "step": 13574
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0455197900869295e-05,
+      "loss": 0.2179,
+      "step": 13575
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045400397835029e-05,
+      "loss": 0.2325,
+      "step": 13576
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452810049346276e-05,
+      "loss": 0.2138,
+      "step": 13577
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451616113874312e-05,
+      "loss": 0.239,
+      "step": 13578
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0450422171951453e-05,
+      "loss": 0.2421,
+      "step": 13579
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449228223594744e-05,
+      "loss": 0.2356,
+      "step": 13580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044803426882125e-05,
+      "loss": 0.2113,
+      "step": 13581
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0446840307648019e-05,
+      "loss": 0.2316,
+      "step": 13582
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044564634009211e-05,
+      "loss": 0.2359,
+      "step": 13583
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444452366170576e-05,
+      "loss": 0.226,
+      "step": 13584
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044325838590047e-05,
+      "loss": 0.2228,
+      "step": 13585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0442064399298846e-05,
+      "loss": 0.2344,
+      "step": 13586
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0440870406382764e-05,
+      "loss": 0.2281,
+      "step": 13587
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0439676407169275e-05,
+      "loss": 0.221,
+      "step": 13588
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0438482401675435e-05,
+      "loss": 0.232,
+      "step": 13589
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437288389918298e-05,
+      "loss": 0.2527,
+      "step": 13590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436094371914927e-05,
+      "loss": 0.2381,
+      "step": 13591
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434900347682363e-05,
+      "loss": 0.2392,
+      "step": 13592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0433706317237675e-05,
+      "loss": 0.2305,
+      "step": 13593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043251228059791e-05,
+      "loss": 0.2291,
+      "step": 13594
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.043131823778013e-05,
+      "loss": 0.2368,
+      "step": 13595
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0430124188801385e-05,
+      "loss": 0.2577,
+      "step": 13596
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0428930133678734e-05,
+      "loss": 0.254,
+      "step": 13597
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0427736072429235e-05,
+      "loss": 0.211,
+      "step": 13598
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0426542005069937e-05,
+      "loss": 0.215,
+      "step": 13599
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0425347931617903e-05,
+      "loss": 0.2349,
+      "step": 13600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0424153852090187e-05,
+      "loss": 0.2365,
+      "step": 13601
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422959766503844e-05,
+      "loss": 0.2086,
+      "step": 13602
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042176567487593e-05,
+      "loss": 0.2291,
+      "step": 13603
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0420571577223504e-05,
+      "loss": 0.2318,
+      "step": 13604
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0419377473563621e-05,
+      "loss": 0.2531,
+      "step": 13605
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0418183363913338e-05,
+      "loss": 0.2194,
+      "step": 13606
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0416989248289713e-05,
+      "loss": 0.2316,
+      "step": 13607
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0415795126709797e-05,
+      "loss": 0.2515,
+      "step": 13608
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0414600999190655e-05,
+      "loss": 0.227,
+      "step": 13609
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413406865749339e-05,
+      "loss": 0.2193,
+      "step": 13610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412212726402905e-05,
+      "loss": 0.2489,
+      "step": 13611
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0411018581168418e-05,
+      "loss": 0.2367,
+      "step": 13612
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0409824430062924e-05,
+      "loss": 0.2338,
+      "step": 13613
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408630273103488e-05,
+      "loss": 0.2461,
+      "step": 13614
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0407436110307163e-05,
+      "loss": 0.2261,
+      "step": 13615
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406241941691011e-05,
+      "loss": 0.2412,
+      "step": 13616
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040504776727209e-05,
+      "loss": 0.216,
+      "step": 13617
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040385358706745e-05,
+      "loss": 0.2263,
+      "step": 13618
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 0.2378,
+      "step": 13619
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040146520936926e-05,
+      "loss": 0.223,
+      "step": 13620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0400271011909822e-05,
+      "loss": 0.2446,
+      "step": 13621
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0399076808732907e-05,
+      "loss": 0.2602,
+      "step": 13622
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397882599855562e-05,
+      "loss": 0.2459,
+      "step": 13623
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0396688385294856e-05,
+      "loss": 0.228,
+      "step": 13624
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395494165067834e-05,
+      "loss": 0.2398,
+      "step": 13625
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394299939191565e-05,
+      "loss": 0.2199,
+      "step": 13626
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0393105707683104e-05,
+      "loss": 0.2392,
+      "step": 13627
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0391911470559513e-05,
+      "loss": 0.2545,
+      "step": 13628
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0390717227837843e-05,
+      "loss": 0.2432,
+      "step": 13629
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0389522979535156e-05,
+      "loss": 0.2108,
+      "step": 13630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388328725668515e-05,
+      "loss": 0.2213,
+      "step": 13631
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0387134466254972e-05,
+      "loss": 0.2396,
+      "step": 13632
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038594020131159e-05,
+      "loss": 0.2241,
+      "step": 13633
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0384745930855426e-05,
+      "loss": 0.2633,
+      "step": 13634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0383551654903541e-05,
+      "loss": 0.2079,
+      "step": 13635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382357373472991e-05,
+      "loss": 0.2453,
+      "step": 13636
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381163086580838e-05,
+      "loss": 0.2184,
+      "step": 13637
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0379968794244143e-05,
+      "loss": 0.2249,
+      "step": 13638
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037877449647996e-05,
+      "loss": 0.2242,
+      "step": 13639
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037758019330535e-05,
+      "loss": 0.2408,
+      "step": 13640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0376385884737374e-05,
+      "loss": 0.2303,
+      "step": 13641
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375191570793091e-05,
+      "loss": 0.2368,
+      "step": 13642
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373997251489558e-05,
+      "loss": 0.2435,
+      "step": 13643
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0372802926843843e-05,
+      "loss": 0.241,
+      "step": 13644
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371608596872997e-05,
+      "loss": 0.2359,
+      "step": 13645
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0370414261594084e-05,
+      "loss": 0.2286,
+      "step": 13646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0369219921024161e-05,
+      "loss": 0.2521,
+      "step": 13647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368025575180289e-05,
+      "loss": 0.2462,
+      "step": 13648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366831224079532e-05,
+      "loss": 0.2074,
+      "step": 13649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0365636867738947e-05,
+      "loss": 0.2532,
+      "step": 13650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0364442506175592e-05,
+      "loss": 0.2041,
+      "step": 13651
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363248139406528e-05,
+      "loss": 0.2251,
+      "step": 13652
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362053767448818e-05,
+      "loss": 0.2255,
+      "step": 13653
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0360859390319522e-05,
+      "loss": 0.2296,
+      "step": 13654
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.03596650080357e-05,
+      "loss": 0.1915,
+      "step": 13655
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0358470620614412e-05,
+      "loss": 0.2221,
+      "step": 13656
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035727622807272e-05,
+      "loss": 0.2207,
+      "step": 13657
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0356081830427685e-05,
+      "loss": 0.2299,
+      "step": 13658
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354887427696364e-05,
+      "loss": 0.2407,
+      "step": 13659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035369301989582e-05,
+      "loss": 0.2186,
+      "step": 13660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352498607043118e-05,
+      "loss": 0.2152,
+      "step": 13661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0351304189155311e-05,
+      "loss": 0.227,
+      "step": 13662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035010976624947e-05,
+      "loss": 0.2133,
+      "step": 13663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0348915338342647e-05,
+      "loss": 0.2375,
+      "step": 13664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347720905451906e-05,
+      "loss": 0.2381,
+      "step": 13665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0346526467594311e-05,
+      "loss": 0.2366,
+      "step": 13666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345332024786923e-05,
+      "loss": 0.2472,
+      "step": 13667
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.03441375770468e-05,
+      "loss": 0.2374,
+      "step": 13668
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0342943124391008e-05,
+      "loss": 0.2424,
+      "step": 13669
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341748666836604e-05,
+      "loss": 0.2111,
+      "step": 13670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0340554204400653e-05,
+      "loss": 0.2388,
+      "step": 13671
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339359737100216e-05,
+      "loss": 0.2219,
+      "step": 13672
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338165264952356e-05,
+      "loss": 0.2315,
+      "step": 13673
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033697078797413e-05,
+      "loss": 0.2363,
+      "step": 13674
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335776306182603e-05,
+      "loss": 0.23,
+      "step": 13675
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033458181959484e-05,
+      "loss": 0.2182,
+      "step": 13676
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0333387328227902e-05,
+      "loss": 0.229,
+      "step": 13677
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332192832098849e-05,
+      "loss": 0.2595,
+      "step": 13678
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033099833122474e-05,
+      "loss": 0.2343,
+      "step": 13679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329803825622642e-05,
+      "loss": 0.2303,
+      "step": 13680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0328609315309616e-05,
+      "loss": 0.2219,
+      "step": 13681
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0327414800302726e-05,
+      "loss": 0.2412,
+      "step": 13682
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326220280619036e-05,
+      "loss": 0.1917,
+      "step": 13683
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0325025756275607e-05,
+      "loss": 0.2331,
+      "step": 13684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0323831227289496e-05,
+      "loss": 0.2149,
+      "step": 13685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.032263669367777e-05,
+      "loss": 0.2253,
+      "step": 13686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0321442155457492e-05,
+      "loss": 0.2158,
+      "step": 13687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320247612645729e-05,
+      "loss": 0.2263,
+      "step": 13688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0319053065259534e-05,
+      "loss": 0.2272,
+      "step": 13689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317858513315981e-05,
+      "loss": 0.2166,
+      "step": 13690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0316663956832125e-05,
+      "loss": 0.2389,
+      "step": 13691
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031546939582503e-05,
+      "loss": 0.2385,
+      "step": 13692
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031427483031176e-05,
+      "loss": 0.2251,
+      "step": 13693
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0313080260309384e-05,
+      "loss": 0.2392,
+      "step": 13694
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0311885685834958e-05,
+      "loss": 0.2404,
+      "step": 13695
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310691106905551e-05,
+      "loss": 0.2444,
+      "step": 13696
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0309496523538218e-05,
+      "loss": 0.2307,
+      "step": 13697
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030830193575003e-05,
+      "loss": 0.2326,
+      "step": 13698
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0307107343558045e-05,
+      "loss": 0.2488,
+      "step": 13699
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305912746979334e-05,
+      "loss": 0.2198,
+      "step": 13700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0304718146030952e-05,
+      "loss": 0.221,
+      "step": 13701
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0303523540729969e-05,
+      "loss": 0.2174,
+      "step": 13702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302328931093449e-05,
+      "loss": 0.212,
+      "step": 13703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0301134317138451e-05,
+      "loss": 0.2181,
+      "step": 13704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299939698882044e-05,
+      "loss": 0.2368,
+      "step": 13705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0298745076341287e-05,
+      "loss": 0.2149,
+      "step": 13706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0297550449533248e-05,
+      "loss": 0.2345,
+      "step": 13707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296355818474988e-05,
+      "loss": 0.2176,
+      "step": 13708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295161183183571e-05,
+      "loss": 0.2339,
+      "step": 13709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293966543676068e-05,
+      "loss": 0.2248,
+      "step": 13710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0292771899969538e-05,
+      "loss": 0.2243,
+      "step": 13711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0291577252081042e-05,
+      "loss": 0.2609,
+      "step": 13712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0290382600027648e-05,
+      "loss": 0.2234,
+      "step": 13713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0289187943826422e-05,
+      "loss": 0.2231,
+      "step": 13714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287993283494426e-05,
+      "loss": 0.2309,
+      "step": 13715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286798619048728e-05,
+      "loss": 0.2405,
+      "step": 13716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285603950506389e-05,
+      "loss": 0.2159,
+      "step": 13717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284409277884473e-05,
+      "loss": 0.2261,
+      "step": 13718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0283214601200047e-05,
+      "loss": 0.219,
+      "step": 13719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282019920470175e-05,
+      "loss": 0.2125,
+      "step": 13720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0280825235711924e-05,
+      "loss": 0.2222,
+      "step": 13721
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279630546942353e-05,
+      "loss": 0.2017,
+      "step": 13722
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0278435854178536e-05,
+      "loss": 0.2356,
+      "step": 13723
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0277241157437532e-05,
+      "loss": 0.25,
+      "step": 13724
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276046456736403e-05,
+      "loss": 0.2255,
+      "step": 13725
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0274851752092222e-05,
+      "loss": 0.2041,
+      "step": 13726
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0273657043522052e-05,
+      "loss": 0.247,
+      "step": 13727
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272462331042954e-05,
+      "loss": 0.2323,
+      "step": 13728
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0271267614671999e-05,
+      "loss": 0.2259,
+      "step": 13729
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270072894426243e-05,
+      "loss": 0.2334,
+      "step": 13730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0268878170322763e-05,
+      "loss": 0.2151,
+      "step": 13731
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0267683442378617e-05,
+      "loss": 0.2116,
+      "step": 13732
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266488710610876e-05,
+      "loss": 0.2268,
+      "step": 13733
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0265293975036599e-05,
+      "loss": 0.2412,
+      "step": 13734
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0264099235672855e-05,
+      "loss": 0.2357,
+      "step": 13735
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262904492536713e-05,
+      "loss": 0.2289,
+      "step": 13736
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0261709745645231e-05,
+      "loss": 0.2326,
+      "step": 13737
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0260514995015484e-05,
+      "loss": 0.2074,
+      "step": 13738
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259320240664532e-05,
+      "loss": 0.2407,
+      "step": 13739
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.025812548260944e-05,
+      "loss": 0.2353,
+      "step": 13740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0256930720867275e-05,
+      "loss": 0.2502,
+      "step": 13741
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255735955455109e-05,
+      "loss": 0.2252,
+      "step": 13742
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254541186389998e-05,
+      "loss": 0.2527,
+      "step": 13743
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0253346413689018e-05,
+      "loss": 0.2355,
+      "step": 13744
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252151637369227e-05,
+      "loss": 0.238,
+      "step": 13745
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250956857447694e-05,
+      "loss": 0.2284,
+      "step": 13746
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0249762073941488e-05,
+      "loss": 0.2598,
+      "step": 13747
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024856728686767e-05,
+      "loss": 0.2209,
+      "step": 13748
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0247372496243313e-05,
+      "loss": 0.219,
+      "step": 13749
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024617770208548e-05,
+      "loss": 0.2436,
+      "step": 13750
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244982904411235e-05,
+      "loss": 0.2306,
+      "step": 13751
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0243788103237648e-05,
+      "loss": 0.2364,
+      "step": 13752
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242593298581784e-05,
+      "loss": 0.2327,
+      "step": 13753
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241398490460712e-05,
+      "loss": 0.2373,
+      "step": 13754
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0240203678891495e-05,
+      "loss": 0.2205,
+      "step": 13755
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239008863891205e-05,
+      "loss": 0.2315,
+      "step": 13756
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237814045476901e-05,
+      "loss": 0.2206,
+      "step": 13757
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0236619223665656e-05,
+      "loss": 0.2265,
+      "step": 13758
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0235424398474533e-05,
+      "loss": 0.225,
+      "step": 13759
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234229569920605e-05,
+      "loss": 0.2286,
+      "step": 13760
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0233034738020933e-05,
+      "loss": 0.227,
+      "step": 13761
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0231839902792587e-05,
+      "loss": 0.2425,
+      "step": 13762
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023064506425263e-05,
+      "loss": 0.2369,
+      "step": 13763
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0229450222418134e-05,
+      "loss": 0.2325,
+      "step": 13764
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0228255377306163e-05,
+      "loss": 0.2273,
+      "step": 13765
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227060528933788e-05,
+      "loss": 0.2435,
+      "step": 13766
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225865677318069e-05,
+      "loss": 0.2126,
+      "step": 13767
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0224670822476082e-05,
+      "loss": 0.2449,
+      "step": 13768
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223475964424891e-05,
+      "loss": 0.2337,
+      "step": 13769
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222281103181562e-05,
+      "loss": 0.2409,
+      "step": 13770
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0221086238763161e-05,
+      "loss": 0.2313,
+      "step": 13771
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.021989137118676e-05,
+      "loss": 0.2516,
+      "step": 13772
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218696500469422e-05,
+      "loss": 0.2423,
+      "step": 13773
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0217501626628216e-05,
+      "loss": 0.2351,
+      "step": 13774
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0216306749680215e-05,
+      "loss": 0.2295,
+      "step": 13775
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.021511186964248e-05,
+      "loss": 0.2172,
+      "step": 13776
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.021391698653208e-05,
+      "loss": 0.25,
+      "step": 13777
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0212722100366084e-05,
+      "loss": 0.2097,
+      "step": 13778
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211527211161559e-05,
+      "loss": 0.231,
+      "step": 13779
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0210332318935572e-05,
+      "loss": 0.2121,
+      "step": 13780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020913742370519e-05,
+      "loss": 0.2194,
+      "step": 13781
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207942525487488e-05,
+      "loss": 0.2376,
+      "step": 13782
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206747624299529e-05,
+      "loss": 0.2357,
+      "step": 13783
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0205552720158375e-05,
+      "loss": 0.2187,
+      "step": 13784
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0204357813081105e-05,
+      "loss": 0.232,
+      "step": 13785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0203162903084778e-05,
+      "loss": 0.2196,
+      "step": 13786
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201967990186468e-05,
+      "loss": 0.2016,
+      "step": 13787
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0200773074403244e-05,
+      "loss": 0.22,
+      "step": 13788
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019957815575217e-05,
+      "loss": 0.2043,
+      "step": 13789
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0198383234250315e-05,
+      "loss": 0.2518,
+      "step": 13790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197188309914748e-05,
+      "loss": 0.2182,
+      "step": 13791
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0195993382762535e-05,
+      "loss": 0.2235,
+      "step": 13792
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019479845281075e-05,
+      "loss": 0.2475,
+      "step": 13793
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0193603520076456e-05,
+      "loss": 0.2269,
+      "step": 13794
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192408584576727e-05,
+      "loss": 0.2446,
+      "step": 13795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191213646328626e-05,
+      "loss": 0.2652,
+      "step": 13796
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190018705349223e-05,
+      "loss": 0.2191,
+      "step": 13797
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188823761655586e-05,
+      "loss": 0.2225,
+      "step": 13798
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187628815264787e-05,
+      "loss": 0.2372,
+      "step": 13799
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186433866193893e-05,
+      "loss": 0.2256,
+      "step": 13800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018523891445997e-05,
+      "loss": 0.2238,
+      "step": 13801
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0184043960080093e-05,
+      "loss": 0.2176,
+      "step": 13802
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182849003071322e-05,
+      "loss": 0.2379,
+      "step": 13803
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018165404345073e-05,
+      "loss": 0.2471,
+      "step": 13804
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0180459081235389e-05,
+      "loss": 0.2288,
+      "step": 13805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0179264116442367e-05,
+      "loss": 0.2193,
+      "step": 13806
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0178069149088727e-05,
+      "loss": 0.2264,
+      "step": 13807
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176874179191544e-05,
+      "loss": 0.2238,
+      "step": 13808
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0175679206767885e-05,
+      "loss": 0.2391,
+      "step": 13809
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017448423183482e-05,
+      "loss": 0.2244,
+      "step": 13810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173289254409415e-05,
+      "loss": 0.2197,
+      "step": 13811
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172094274508742e-05,
+      "loss": 0.2269,
+      "step": 13812
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170899292149868e-05,
+      "loss": 0.2207,
+      "step": 13813
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169704307349863e-05,
+      "loss": 0.23,
+      "step": 13814
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0168509320125801e-05,
+      "loss": 0.245,
+      "step": 13815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167314330494746e-05,
+      "loss": 0.247,
+      "step": 13816
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166119338473765e-05,
+      "loss": 0.2278,
+      "step": 13817
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0164924344079932e-05,
+      "loss": 0.228,
+      "step": 13818
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0163729347330314e-05,
+      "loss": 0.224,
+      "step": 13819
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162534348241981e-05,
+      "loss": 0.222,
+      "step": 13820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161339346832006e-05,
+      "loss": 0.2655,
+      "step": 13821
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160144343117453e-05,
+      "loss": 0.2323,
+      "step": 13822
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158949337115394e-05,
+      "loss": 0.235,
+      "step": 13823
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0157754328842895e-05,
+      "loss": 0.2053,
+      "step": 13824
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015655931831703e-05,
+      "loss": 0.2322,
+      "step": 13825
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155364305554868e-05,
+      "loss": 0.2472,
+      "step": 13826
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0154169290573479e-05,
+      "loss": 0.2348,
+      "step": 13827
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152974273389931e-05,
+      "loss": 0.226,
+      "step": 13828
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015177925402129e-05,
+      "loss": 0.2495,
+      "step": 13829
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150584232484632e-05,
+      "loss": 0.2323,
+      "step": 13830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149389208797025e-05,
+      "loss": 0.2233,
+      "step": 13831
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014819418297554e-05,
+      "loss": 0.2346,
+      "step": 13832
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014699915503724e-05,
+      "loss": 0.2502,
+      "step": 13833
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0145804124999203e-05,
+      "loss": 0.2246,
+      "step": 13834
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144609092878497e-05,
+      "loss": 0.2206,
+      "step": 13835
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143414058692187e-05,
+      "loss": 0.2138,
+      "step": 13836
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0142219022457347e-05,
+      "loss": 0.2512,
+      "step": 13837
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0141023984191049e-05,
+      "loss": 0.2211,
+      "step": 13838
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0139828943910358e-05,
+      "loss": 0.2346,
+      "step": 13839
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0138633901632344e-05,
+      "loss": 0.2396,
+      "step": 13840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137438857374084e-05,
+      "loss": 0.2293,
+      "step": 13841
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136243811152642e-05,
+      "loss": 0.2288,
+      "step": 13842
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135048762985088e-05,
+      "loss": 0.2213,
+      "step": 13843
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133853712888495e-05,
+      "loss": 0.232,
+      "step": 13844
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013265866087993e-05,
+      "loss": 0.2261,
+      "step": 13845
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0131463606976464e-05,
+      "loss": 0.2334,
+      "step": 13846
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0130268551195168e-05,
+      "loss": 0.2535,
+      "step": 13847
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129073493553117e-05,
+      "loss": 0.2149,
+      "step": 13848
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127878434067373e-05,
+      "loss": 0.2239,
+      "step": 13849
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0126683372755011e-05,
+      "loss": 0.2345,
+      "step": 13850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125488309633096e-05,
+      "loss": 0.2231,
+      "step": 13851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0124293244718706e-05,
+      "loss": 0.2339,
+      "step": 13852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123098178028905e-05,
+      "loss": 0.2119,
+      "step": 13853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0121903109580768e-05,
+      "loss": 0.232,
+      "step": 13854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0120708039391365e-05,
+      "loss": 0.2474,
+      "step": 13855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119512967477763e-05,
+      "loss": 0.2173,
+      "step": 13856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118317893857033e-05,
+      "loss": 0.2249,
+      "step": 13857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0117122818546247e-05,
+      "loss": 0.2277,
+      "step": 13858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0115927741562474e-05,
+      "loss": 0.2161,
+      "step": 13859
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011473266292279e-05,
+      "loss": 0.2133,
+      "step": 13860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0113537582644261e-05,
+      "loss": 0.25,
+      "step": 13861
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112342500743955e-05,
+      "loss": 0.2217,
+      "step": 13862
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0111147417238944e-05,
+      "loss": 0.2472,
+      "step": 13863
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109952332146303e-05,
+      "loss": 0.2319,
+      "step": 13864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01087572454831e-05,
+      "loss": 0.2275,
+      "step": 13865
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0107562157266404e-05,
+      "loss": 0.2165,
+      "step": 13866
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0106367067513285e-05,
+      "loss": 0.223,
+      "step": 13867
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010517197624082e-05,
+      "loss": 0.2283,
+      "step": 13868
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0103976883466073e-05,
+      "loss": 0.2488,
+      "step": 13869
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0102781789206115e-05,
+      "loss": 0.2356,
+      "step": 13870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101586693478025e-05,
+      "loss": 0.2095,
+      "step": 13871
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0100391596298861e-05,
+      "loss": 0.2611,
+      "step": 13872
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0099196497685703e-05,
+      "loss": 0.2316,
+      "step": 13873
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098001397655621e-05,
+      "loss": 0.2297,
+      "step": 13874
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0096806296225683e-05,
+      "loss": 0.2452,
+      "step": 13875
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009561119341296e-05,
+      "loss": 0.2457,
+      "step": 13876
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094416089234526e-05,
+      "loss": 0.2406,
+      "step": 13877
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093220983707448e-05,
+      "loss": 0.2646,
+      "step": 13878
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00920258768488e-05,
+      "loss": 0.2212,
+      "step": 13879
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090830768675648e-05,
+      "loss": 0.2434,
+      "step": 13880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0089635659205073e-05,
+      "loss": 0.2474,
+      "step": 13881
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0088440548454138e-05,
+      "loss": 0.2246,
+      "step": 13882
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0087245436439911e-05,
+      "loss": 0.2111,
+      "step": 13883
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0086050323179471e-05,
+      "loss": 0.2155,
+      "step": 13884
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084855208689883e-05,
+      "loss": 0.2321,
+      "step": 13885
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083660092988223e-05,
+      "loss": 0.2254,
+      "step": 13886
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008246497609156e-05,
+      "loss": 0.2202,
+      "step": 13887
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0081269858016967e-05,
+      "loss": 0.233,
+      "step": 13888
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080074738781509e-05,
+      "loss": 0.2146,
+      "step": 13889
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0078879618402262e-05,
+      "loss": 0.2455,
+      "step": 13890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0077684496896296e-05,
+      "loss": 0.2321,
+      "step": 13891
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007648937428068e-05,
+      "loss": 0.2187,
+      "step": 13892
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0075294250572493e-05,
+      "loss": 0.23,
+      "step": 13893
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0074099125788799e-05,
+      "loss": 0.248,
+      "step": 13894
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007290399994667e-05,
+      "loss": 0.2389,
+      "step": 13895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071708873063177e-05,
+      "loss": 0.2198,
+      "step": 13896
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070513745155392e-05,
+      "loss": 0.2201,
+      "step": 13897
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006931861624039e-05,
+      "loss": 0.2467,
+      "step": 13898
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0068123486335238e-05,
+      "loss": 0.2133,
+      "step": 13899
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066928355457007e-05,
+      "loss": 0.2335,
+      "step": 13900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065733223622769e-05,
+      "loss": 0.2424,
+      "step": 13901
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064538090849597e-05,
+      "loss": 0.2598,
+      "step": 13902
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063342957154558e-05,
+      "loss": 0.2193,
+      "step": 13903
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006214782255473e-05,
+      "loss": 0.212,
+      "step": 13904
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0060952687067178e-05,
+      "loss": 0.2247,
+      "step": 13905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059757550708978e-05,
+      "loss": 0.2173,
+      "step": 13906
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058562413497199e-05,
+      "loss": 0.217,
+      "step": 13907
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005736727544891e-05,
+      "loss": 0.2138,
+      "step": 13908
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056172136581188e-05,
+      "loss": 0.216,
+      "step": 13909
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054976996911103e-05,
+      "loss": 0.2546,
+      "step": 13910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053781856455723e-05,
+      "loss": 0.2299,
+      "step": 13911
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005258671523212e-05,
+      "loss": 0.2116,
+      "step": 13912
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0051391573257367e-05,
+      "loss": 0.2164,
+      "step": 13913
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050196430548538e-05,
+      "loss": 0.225,
+      "step": 13914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0049001287122701e-05,
+      "loss": 0.2497,
+      "step": 13915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047806142996928e-05,
+      "loss": 0.2287,
+      "step": 13916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004661099818829e-05,
+      "loss": 0.2444,
+      "step": 13917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045415852713856e-05,
+      "loss": 0.2373,
+      "step": 13918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044220706590705e-05,
+      "loss": 0.2177,
+      "step": 13919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043025559835902e-05,
+      "loss": 0.2545,
+      "step": 13920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041830412466522e-05,
+      "loss": 0.2571,
+      "step": 13921
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0040635264499633e-05,
+      "loss": 0.2301,
+      "step": 13922
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039440115952312e-05,
+      "loss": 0.2149,
+      "step": 13923
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038244966841624e-05,
+      "loss": 0.2302,
+      "step": 13924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0037049817184646e-05,
+      "loss": 0.2361,
+      "step": 13925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0035854666998448e-05,
+      "loss": 0.2365,
+      "step": 13926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00346595163001e-05,
+      "loss": 0.2265,
+      "step": 13927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0033464365106674e-05,
+      "loss": 0.2462,
+      "step": 13928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0032269213435239e-05,
+      "loss": 0.2659,
+      "step": 13929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0031074061302873e-05,
+      "loss": 0.2194,
+      "step": 13930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029878908726645e-05,
+      "loss": 0.2554,
+      "step": 13931
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0028683755723625e-05,
+      "loss": 0.2092,
+      "step": 13932
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027488602310886e-05,
+      "loss": 0.265,
+      "step": 13933
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026293448505497e-05,
+      "loss": 0.2362,
+      "step": 13934
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025098294324531e-05,
+      "loss": 0.2327,
+      "step": 13935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002390313978506e-05,
+      "loss": 0.2203,
+      "step": 13936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002270798490416e-05,
+      "loss": 0.2343,
+      "step": 13937
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0021512829698896e-05,
+      "loss": 0.216,
+      "step": 13938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020317674186341e-05,
+      "loss": 0.2384,
+      "step": 13939
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019122518383571e-05,
+      "loss": 0.2087,
+      "step": 13940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017927362307652e-05,
+      "loss": 0.2369,
+      "step": 13941
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0016732205975657e-05,
+      "loss": 0.2401,
+      "step": 13942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001553704940466e-05,
+      "loss": 0.2349,
+      "step": 13943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014341892611731e-05,
+      "loss": 0.2476,
+      "step": 13944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013146735613942e-05,
+      "loss": 0.234,
+      "step": 13945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0011951578428365e-05,
+      "loss": 0.2369,
+      "step": 13946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0010756421072072e-05,
+      "loss": 0.2364,
+      "step": 13947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009561263562135e-05,
+      "loss": 0.2302,
+      "step": 13948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0008366105915621e-05,
+      "loss": 0.2201,
+      "step": 13949
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007170948149607e-05,
+      "loss": 0.243,
+      "step": 13950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000597579028116e-05,
+      "loss": 0.2225,
+      "step": 13951
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0004780632327359e-05,
+      "loss": 0.2273,
+      "step": 13952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003585474305271e-05,
+      "loss": 0.2255,
+      "step": 13953
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0002390316231969e-05,
+      "loss": 0.2239,
+      "step": 13954
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0001195158124522e-05,
+      "loss": 0.228,
+      "step": 13955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.2532,
+      "step": 13956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.998804841875481e-06,
+      "loss": 0.2409,
+      "step": 13957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997609683768033e-06,
+      "loss": 0.2275,
+      "step": 13958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99641452569473e-06,
+      "loss": 0.2319,
+      "step": 13959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.995219367672643e-06,
+      "loss": 0.2043,
+      "step": 13960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994024209718837e-06,
+      "loss": 0.2317,
+      "step": 13961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992829051850393e-06,
+      "loss": 0.2702,
+      "step": 13962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.991633894084384e-06,
+      "loss": 0.229,
+      "step": 13963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99043873643787e-06,
+      "loss": 0.2385,
+      "step": 13964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.989243578927931e-06,
+      "loss": 0.2172,
+      "step": 13965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.988048421571638e-06,
+      "loss": 0.2039,
+      "step": 13966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98685326438606e-06,
+      "loss": 0.2089,
+      "step": 13967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985658107388272e-06,
+      "loss": 0.23,
+      "step": 13968
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984462950595341e-06,
+      "loss": 0.2104,
+      "step": 13969
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.983267794024346e-06,
+      "loss": 0.2445,
+      "step": 13970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.982072637692351e-06,
+      "loss": 0.1986,
+      "step": 13971
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980877481616432e-06,
+      "loss": 0.2418,
+      "step": 13972
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97968232581366e-06,
+      "loss": 0.2337,
+      "step": 13973
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.978487170301106e-06,
+      "loss": 0.2298,
+      "step": 13974
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977292015095843e-06,
+      "loss": 0.2172,
+      "step": 13975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976096860214942e-06,
+      "loss": 0.2313,
+      "step": 13976
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.974901705675474e-06,
+      "loss": 0.2163,
+      "step": 13977
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97370655149451e-06,
+      "loss": 0.2265,
+      "step": 13978
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97251139768912e-06,
+      "loss": 0.1839,
+      "step": 13979
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971316244276378e-06,
+      "loss": 0.2436,
+      "step": 13980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970121091273359e-06,
+      "loss": 0.2199,
+      "step": 13981
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968925938697129e-06,
+      "loss": 0.205,
+      "step": 13982
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.967730786564763e-06,
+      "loss": 0.2323,
+      "step": 13983
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96653563489333e-06,
+      "loss": 0.2265,
+      "step": 13984
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965340483699903e-06,
+      "loss": 0.2161,
+      "step": 13985
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.964145333001555e-06,
+      "loss": 0.2247,
+      "step": 13986
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.962950182815356e-06,
+      "loss": 0.2354,
+      "step": 13987
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961755033158376e-06,
+      "loss": 0.2364,
+      "step": 13988
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960559884047693e-06,
+      "loss": 0.2214,
+      "step": 13989
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95936473550037e-06,
+      "loss": 0.2429,
+      "step": 13990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.958169587533481e-06,
+      "loss": 0.2286,
+      "step": 13991
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9569744401641e-06,
+      "loss": 0.2348,
+      "step": 13992
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9557792934093e-06,
+      "loss": 0.2044,
+      "step": 13993
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954584147286147e-06,
+      "loss": 0.2244,
+      "step": 13994
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.953389001811716e-06,
+      "loss": 0.2238,
+      "step": 13995
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.952193857003076e-06,
+      "loss": 0.2221,
+      "step": 13996
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9509987128773e-06,
+      "loss": 0.2184,
+      "step": 13997
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949803569451466e-06,
+      "loss": 0.2447,
+      "step": 13998
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.948608426742634e-06,
+      "loss": 0.2347,
+      "step": 13999
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.947413284767882e-06,
+      "loss": 0.2316,
+      "step": 14000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946218143544279e-06,
+      "loss": 0.2287,
+      "step": 14001
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9450230030889e-06,
+      "loss": 0.255,
+      "step": 14002
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943827863418815e-06,
+      "loss": 0.2462,
+      "step": 14003
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.942632724551093e-06,
+      "loss": 0.2083,
+      "step": 14004
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941437586502806e-06,
+      "loss": 0.2139,
+      "step": 14005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940242449291027e-06,
+      "loss": 0.2241,
+      "step": 14006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.939047312932825e-06,
+      "loss": 0.2283,
+      "step": 14007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937852177445273e-06,
+      "loss": 0.2569,
+      "step": 14008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936657042845444e-06,
+      "loss": 0.2225,
+      "step": 14009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.935461909150407e-06,
+      "loss": 0.2285,
+      "step": 14010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934266776377234e-06,
+      "loss": 0.2378,
+      "step": 14011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933071644542995e-06,
+      "loss": 0.2184,
+      "step": 14012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.931876513664764e-06,
+      "loss": 0.227,
+      "step": 14013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930681383759611e-06,
+      "loss": 0.2009,
+      "step": 14014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929486254844608e-06,
+      "loss": 0.2011,
+      "step": 14015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928291126936828e-06,
+      "loss": 0.2158,
+      "step": 14016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.927096000053335e-06,
+      "loss": 0.2277,
+      "step": 14017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.925900874211205e-06,
+      "loss": 0.2234,
+      "step": 14018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92470574942751e-06,
+      "loss": 0.2388,
+      "step": 14019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923510625719321e-06,
+      "loss": 0.2468,
+      "step": 14020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.922315503103707e-06,
+      "loss": 0.1969,
+      "step": 14021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.921120381597742e-06,
+      "loss": 0.2136,
+      "step": 14022
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919925261218494e-06,
+      "loss": 0.2347,
+      "step": 14023
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918730141983036e-06,
+      "loss": 0.2481,
+      "step": 14024
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917535023908442e-06,
+      "loss": 0.2353,
+      "step": 14025
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916339907011779e-06,
+      "loss": 0.2495,
+      "step": 14026
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915144791310117e-06,
+      "loss": 0.2537,
+      "step": 14027
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91394967682053e-06,
+      "loss": 0.2434,
+      "step": 14028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.912754563560094e-06,
+      "loss": 0.2158,
+      "step": 14029
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.911559451545867e-06,
+      "loss": 0.2176,
+      "step": 14030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910364340794932e-06,
+      "loss": 0.2261,
+      "step": 14031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909169231324354e-06,
+      "loss": 0.2114,
+      "step": 14032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.907974123151204e-06,
+      "loss": 0.2212,
+      "step": 14033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906779016292554e-06,
+      "loss": 0.2212,
+      "step": 14034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.905583910765475e-06,
+      "loss": 0.2539,
+      "step": 14035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904388806587043e-06,
+      "loss": 0.2582,
+      "step": 14036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.90319370377432e-06,
+      "loss": 0.2292,
+      "step": 14037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.90199860234438e-06,
+      "loss": 0.2117,
+      "step": 14038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.900803502314297e-06,
+      "loss": 0.2135,
+      "step": 14039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899608403701138e-06,
+      "loss": 0.2366,
+      "step": 14040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.89841330652198e-06,
+      "loss": 0.2139,
+      "step": 14041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.897218210793886e-06,
+      "loss": 0.2293,
+      "step": 14042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.896023116533933e-06,
+      "loss": 0.2275,
+      "step": 14043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.894828023759185e-06,
+      "loss": 0.242,
+      "step": 14044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893632932486717e-06,
+      "loss": 0.2092,
+      "step": 14045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.8924378427336e-06,
+      "loss": 0.2343,
+      "step": 14046
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891242754516903e-06,
+      "loss": 0.2179,
+      "step": 14047
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8900476678537e-06,
+      "loss": 0.2246,
+      "step": 14048
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.888852582761058e-06,
+      "loss": 0.233,
+      "step": 14049
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.887657499256048e-06,
+      "loss": 0.2325,
+      "step": 14050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886462417355742e-06,
+      "loss": 0.2474,
+      "step": 14051
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.885267337077212e-06,
+      "loss": 0.2391,
+      "step": 14052
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884072258437525e-06,
+      "loss": 0.2272,
+      "step": 14053
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.882877181453755e-06,
+      "loss": 0.2079,
+      "step": 14054
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881682106142972e-06,
+      "loss": 0.226,
+      "step": 14055
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880487032522242e-06,
+      "loss": 0.2338,
+      "step": 14056
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87929196060864e-06,
+      "loss": 0.2155,
+      "step": 14057
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.878096890419234e-06,
+      "loss": 0.2456,
+      "step": 14058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876901821971097e-06,
+      "loss": 0.2294,
+      "step": 14059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.875706755281297e-06,
+      "loss": 0.2302,
+      "step": 14060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874511690366906e-06,
+      "loss": 0.2209,
+      "step": 14061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873316627244994e-06,
+      "loss": 0.2342,
+      "step": 14062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872121565932628e-06,
+      "loss": 0.2047,
+      "step": 14063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870926506446886e-06,
+      "loss": 0.2311,
+      "step": 14064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.869731448804832e-06,
+      "loss": 0.2313,
+      "step": 14065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868536393023537e-06,
+      "loss": 0.2009,
+      "step": 14066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86734133912007e-06,
+      "loss": 0.2452,
+      "step": 14067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86614628711151e-06,
+      "loss": 0.2401,
+      "step": 14068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864951237014917e-06,
+      "loss": 0.2408,
+      "step": 14069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863756188847363e-06,
+      "loss": 0.2356,
+      "step": 14070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862561142625921e-06,
+      "loss": 0.2316,
+      "step": 14071
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.861366098367658e-06,
+      "loss": 0.226,
+      "step": 14072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.860171056089646e-06,
+      "loss": 0.2464,
+      "step": 14073
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.858976015808954e-06,
+      "loss": 0.2015,
+      "step": 14074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857780977542657e-06,
+      "loss": 0.2215,
+      "step": 14075
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856585941307816e-06,
+      "loss": 0.2399,
+      "step": 14076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855390907121507e-06,
+      "loss": 0.2058,
+      "step": 14077
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.854195875000799e-06,
+      "loss": 0.2003,
+      "step": 14078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.85300084496276e-06,
+      "loss": 0.2245,
+      "step": 14079
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.851805817024464e-06,
+      "loss": 0.2279,
+      "step": 14080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850610791202975e-06,
+      "loss": 0.2314,
+      "step": 14081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.849415767515371e-06,
+      "loss": 0.2389,
+      "step": 14082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848220745978715e-06,
+      "loss": 0.224,
+      "step": 14083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847025726610076e-06,
+      "loss": 0.2193,
+      "step": 14084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.845830709426524e-06,
+      "loss": 0.2111,
+      "step": 14085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844635694445133e-06,
+      "loss": 0.2237,
+      "step": 14086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.843440681682974e-06,
+      "loss": 0.2281,
+      "step": 14087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842245671157108e-06,
+      "loss": 0.2654,
+      "step": 14088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84105066288461e-06,
+      "loss": 0.2228,
+      "step": 14089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839855656882549e-06,
+      "loss": 0.2379,
+      "step": 14090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.838660653167997e-06,
+      "loss": 0.2428,
+      "step": 14091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837465651758019e-06,
+      "loss": 0.239,
+      "step": 14092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836270652669686e-06,
+      "loss": 0.2158,
+      "step": 14093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83507565592007e-06,
+      "loss": 0.2289,
+      "step": 14094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83388066152624e-06,
+      "loss": 0.2304,
+      "step": 14095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83268566950526e-06,
+      "loss": 0.2118,
+      "step": 14096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.831490679874202e-06,
+      "loss": 0.2377,
+      "step": 14097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83029569265014e-06,
+      "loss": 0.2367,
+      "step": 14098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.829100707850134e-06,
+      "loss": 0.2439,
+      "step": 14099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827905725491261e-06,
+      "loss": 0.2374,
+      "step": 14100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826710745590587e-06,
+      "loss": 0.2426,
+      "step": 14101
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.825515768165183e-06,
+      "loss": 0.2087,
+      "step": 14102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.824320793232118e-06,
+      "loss": 0.2385,
+      "step": 14103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823125820808457e-06,
+      "loss": 0.2346,
+      "step": 14104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.821930850911274e-06,
+      "loss": 0.2381,
+      "step": 14105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.820735883557634e-06,
+      "loss": 0.213,
+      "step": 14106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819540918764613e-06,
+      "loss": 0.2497,
+      "step": 14107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818345956549273e-06,
+      "loss": 0.2583,
+      "step": 14108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817150996928683e-06,
+      "loss": 0.2306,
+      "step": 14109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.815956039919914e-06,
+      "loss": 0.2569,
+      "step": 14110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.814761085540032e-06,
+      "loss": 0.2361,
+      "step": 14111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81356613380611e-06,
+      "loss": 0.1988,
+      "step": 14112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812371184735214e-06,
+      "loss": 0.2175,
+      "step": 14113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811176238344417e-06,
+      "loss": 0.2288,
+      "step": 14114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80998129465078e-06,
+      "loss": 0.2229,
+      "step": 14115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.808786353671379e-06,
+      "loss": 0.239,
+      "step": 14116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807591415423275e-06,
+      "loss": 0.2389,
+      "step": 14117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.806396479923544e-06,
+      "loss": 0.265,
+      "step": 14118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.805201547189252e-06,
+      "loss": 0.226,
+      "step": 14119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.804006617237467e-06,
+      "loss": 0.2181,
+      "step": 14120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80281169008526e-06,
+      "loss": 0.2102,
+      "step": 14121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80161676574969e-06,
+      "loss": 0.2356,
+      "step": 14122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800421844247834e-06,
+      "loss": 0.2618,
+      "step": 14123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79922692559676e-06,
+      "loss": 0.219,
+      "step": 14124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798032009813533e-06,
+      "loss": 0.2533,
+      "step": 14125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.796837096915223e-06,
+      "loss": 0.2306,
+      "step": 14126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.795642186918899e-06,
+      "loss": 0.2302,
+      "step": 14127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.794447279841626e-06,
+      "loss": 0.2475,
+      "step": 14128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.793252375700475e-06,
+      "loss": 0.2398,
+      "step": 14129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.792057474512515e-06,
+      "loss": 0.2424,
+      "step": 14130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79086257629481e-06,
+      "loss": 0.2262,
+      "step": 14131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78966768106443e-06,
+      "loss": 0.2367,
+      "step": 14132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788472788838443e-06,
+      "loss": 0.1987,
+      "step": 14133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.787277899633921e-06,
+      "loss": 0.2452,
+      "step": 14134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786083013467923e-06,
+      "loss": 0.21,
+      "step": 14135
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.784888130357524e-06,
+      "loss": 0.2483,
+      "step": 14136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.78369325031979e-06,
+      "loss": 0.2497,
+      "step": 14137
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.782498373371785e-06,
+      "loss": 0.2054,
+      "step": 14138
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.78130349953058e-06,
+      "loss": 0.2608,
+      "step": 14139
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.780108628813243e-06,
+      "loss": 0.2042,
+      "step": 14140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.778913761236842e-06,
+      "loss": 0.2369,
+      "step": 14141
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777718896818443e-06,
+      "loss": 0.2224,
+      "step": 14142
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776524035575112e-06,
+      "loss": 0.2235,
+      "step": 14143
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.775329177523918e-06,
+      "loss": 0.2381,
+      "step": 14144
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77413432268193e-06,
+      "loss": 0.254,
+      "step": 14145
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.772939471066216e-06,
+      "loss": 0.2258,
+      "step": 14146
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77174462269384e-06,
+      "loss": 0.2405,
+      "step": 14147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770549777581873e-06,
+      "loss": 0.2364,
+      "step": 14148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769354935747374e-06,
+      "loss": 0.2368,
+      "step": 14149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.768160097207418e-06,
+      "loss": 0.2333,
+      "step": 14150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.766965261979072e-06,
+      "loss": 0.2001,
+      "step": 14151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.765770430079397e-06,
+      "loss": 0.2363,
+      "step": 14152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.764575601525468e-06,
+      "loss": 0.2227,
+      "step": 14153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763380776334347e-06,
+      "loss": 0.214,
+      "step": 14154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7621859545231e-06,
+      "loss": 0.2336,
+      "step": 14155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760991136108798e-06,
+      "loss": 0.229,
+      "step": 14156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759796321108505e-06,
+      "loss": 0.2281,
+      "step": 14157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758601509539291e-06,
+      "loss": 0.237,
+      "step": 14158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757406701418216e-06,
+      "loss": 0.2687,
+      "step": 14159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.756211896762351e-06,
+      "loss": 0.2335,
+      "step": 14160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75501709558877e-06,
+      "loss": 0.2455,
+      "step": 14161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.753822297914525e-06,
+      "loss": 0.2319,
+      "step": 14162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75262750375669e-06,
+      "loss": 0.2454,
+      "step": 14163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.751432713132333e-06,
+      "loss": 0.2036,
+      "step": 14164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.750237926058517e-06,
+      "loss": 0.2311,
+      "step": 14165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74904314255231e-06,
+      "loss": 0.2441,
+      "step": 14166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747848362630776e-06,
+      "loss": 0.2324,
+      "step": 14167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.746653586310986e-06,
+      "loss": 0.2502,
+      "step": 14168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.745458813610003e-06,
+      "loss": 0.2334,
+      "step": 14169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744264044544894e-06,
+      "loss": 0.2416,
+      "step": 14170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.743069279132725e-06,
+      "loss": 0.245,
+      "step": 14171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74187451739056e-06,
+      "loss": 0.2313,
+      "step": 14172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74067975933547e-06,
+      "loss": 0.2209,
+      "step": 14173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73948500498452e-06,
+      "loss": 0.2103,
+      "step": 14174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73829025435477e-06,
+      "loss": 0.2461,
+      "step": 14175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737095507463292e-06,
+      "loss": 0.2117,
+      "step": 14176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.735900764327147e-06,
+      "loss": 0.228,
+      "step": 14177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734706024963404e-06,
+      "loss": 0.2205,
+      "step": 14178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733511289389128e-06,
+      "loss": 0.2389,
+      "step": 14179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.732316557621387e-06,
+      "loss": 0.2211,
+      "step": 14180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73112182967724e-06,
+      "loss": 0.2227,
+      "step": 14181
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729927105573758e-06,
+      "loss": 0.2249,
+      "step": 14182
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728732385328006e-06,
+      "loss": 0.2512,
+      "step": 14183
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727537668957048e-06,
+      "loss": 0.252,
+      "step": 14184
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.726342956477951e-06,
+      "loss": 0.2276,
+      "step": 14185
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72514824790778e-06,
+      "loss": 0.2327,
+      "step": 14186
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7239535432636e-06,
+      "loss": 0.2512,
+      "step": 14187
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.722758842562475e-06,
+      "loss": 0.2262,
+      "step": 14188
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.721564145821469e-06,
+      "loss": 0.251,
+      "step": 14189
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720369453057648e-06,
+      "loss": 0.2255,
+      "step": 14190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71917476428808e-06,
+      "loss": 0.2314,
+      "step": 14191
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717980079529829e-06,
+      "loss": 0.2417,
+      "step": 14192
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.716785398799956e-06,
+      "loss": 0.2545,
+      "step": 14193
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71559072211553e-06,
+      "loss": 0.2577,
+      "step": 14194
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.714396049493614e-06,
+      "loss": 0.2366,
+      "step": 14195
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.713201380951276e-06,
+      "loss": 0.2284,
+      "step": 14196
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712006716505576e-06,
+      "loss": 0.2297,
+      "step": 14197
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71081205617358e-06,
+      "loss": 0.2274,
+      "step": 14198
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709617399972352e-06,
+      "loss": 0.2212,
+      "step": 14199
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.708422747918963e-06,
+      "loss": 0.2291,
+      "step": 14200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707228100030468e-06,
+      "loss": 0.2082,
+      "step": 14201
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.706033456323935e-06,
+      "loss": 0.2236,
+      "step": 14202
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70483881681643e-06,
+      "loss": 0.2434,
+      "step": 14203
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703644181525015e-06,
+      "loss": 0.2266,
+      "step": 14204
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.702449550466756e-06,
+      "loss": 0.2124,
+      "step": 14205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701254923658716e-06,
+      "loss": 0.2257,
+      "step": 14206
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700060301117959e-06,
+      "loss": 0.2334,
+      "step": 14207
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69886568286155e-06,
+      "loss": 0.186,
+      "step": 14208
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697671068906553e-06,
+      "loss": 0.1976,
+      "step": 14209
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.696476459270031e-06,
+      "loss": 0.2428,
+      "step": 14210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.695281853969047e-06,
+      "loss": 0.2459,
+      "step": 14211
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69408725302067e-06,
+      "loss": 0.2129,
+      "step": 14212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.692892656441959e-06,
+      "loss": 0.2469,
+      "step": 14213
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691698064249977e-06,
+      "loss": 0.2035,
+      "step": 14214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690503476461787e-06,
+      "loss": 0.2399,
+      "step": 14215
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.689308893094456e-06,
+      "loss": 0.2592,
+      "step": 14216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.688114314165045e-06,
+      "loss": 0.2535,
+      "step": 14217
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686919739690617e-06,
+      "loss": 0.2197,
+      "step": 14218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685725169688241e-06,
+      "loss": 0.2064,
+      "step": 14219
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.684530604174971e-06,
+      "loss": 0.2443,
+      "step": 14220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68333604316788e-06,
+      "loss": 0.2097,
+      "step": 14221
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.682141486684022e-06,
+      "loss": 0.2214,
+      "step": 14222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680946934740466e-06,
+      "loss": 0.2153,
+      "step": 14223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.679752387354276e-06,
+      "loss": 0.2313,
+      "step": 14224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.67855784454251e-06,
+      "loss": 0.2459,
+      "step": 14225
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.677363306322235e-06,
+      "loss": 0.242,
+      "step": 14226
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67616877271051e-06,
+      "loss": 0.2566,
+      "step": 14227
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.6749742437244e-06,
+      "loss": 0.2225,
+      "step": 14228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.673779719380967e-06,
+      "loss": 0.2247,
+      "step": 14229
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.672585199697276e-06,
+      "loss": 0.2756,
+      "step": 14230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.671390684690387e-06,
+      "loss": 0.2486,
+      "step": 14231
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670196174377362e-06,
+      "loss": 0.2439,
+      "step": 14232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.669001668775263e-06,
+      "loss": 0.2359,
+      "step": 14233
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667807167901156e-06,
+      "loss": 0.2316,
+      "step": 14234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666612671772103e-06,
+      "loss": 0.2136,
+      "step": 14235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.665418180405161e-06,
+      "loss": 0.2146,
+      "step": 14236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664223693817395e-06,
+      "loss": 0.2215,
+      "step": 14237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66302921202587e-06,
+      "loss": 0.2307,
+      "step": 14238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.661834735047645e-06,
+      "loss": 0.2539,
+      "step": 14239
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660640262899787e-06,
+      "loss": 0.2345,
+      "step": 14240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.659445795599349e-06,
+      "loss": 0.2501,
+      "step": 14241
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.6582513331634e-06,
+      "loss": 0.2862,
+      "step": 14242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.657056875608997e-06,
+      "loss": 0.2359,
+      "step": 14243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.655862422953203e-06,
+      "loss": 0.2447,
+      "step": 14244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65466797521308e-06,
+      "loss": 0.2698,
+      "step": 14245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.653473532405692e-06,
+      "loss": 0.2312,
+      "step": 14246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652279094548098e-06,
+      "loss": 0.2141,
+      "step": 14247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651084661657357e-06,
+      "loss": 0.2241,
+      "step": 14248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649890233750534e-06,
+      "loss": 0.2221,
+      "step": 14249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.648695810844689e-06,
+      "loss": 0.2569,
+      "step": 14250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647501392956887e-06,
+      "loss": 0.2214,
+      "step": 14251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64630698010418e-06,
+      "loss": 0.2449,
+      "step": 14252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.645112572303641e-06,
+      "loss": 0.2195,
+      "step": 14253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.643918169572322e-06,
+      "loss": 0.254,
+      "step": 14254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.642723771927284e-06,
+      "loss": 0.2229,
+      "step": 14255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64152937938559e-06,
+      "loss": 0.21,
+      "step": 14256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640334991964302e-06,
+      "loss": 0.2356,
+      "step": 14257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.639140609680482e-06,
+      "loss": 0.2375,
+      "step": 14258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.637946232551185e-06,
+      "loss": 0.2313,
+      "step": 14259
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636751860593475e-06,
+      "loss": 0.2104,
+      "step": 14260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.635557493824412e-06,
+      "loss": 0.2349,
+      "step": 14261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.634363132261057e-06,
+      "loss": 0.2357,
+      "step": 14262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.633168775920471e-06,
+      "loss": 0.2312,
+      "step": 14263
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631974424819713e-06,
+      "loss": 0.2481,
+      "step": 14264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63078007897584e-06,
+      "loss": 0.2265,
+      "step": 14265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.629585738405923e-06,
+      "loss": 0.2132,
+      "step": 14266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628391403127006e-06,
+      "loss": 0.2237,
+      "step": 14267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62719707315616e-06,
+      "loss": 0.2422,
+      "step": 14268
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626002748510443e-06,
+      "loss": 0.2706,
+      "step": 14269
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624808429206914e-06,
+      "loss": 0.2387,
+      "step": 14270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62361411526263e-06,
+      "loss": 0.24,
+      "step": 14271
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.622419806694653e-06,
+      "loss": 0.2392,
+      "step": 14272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621225503520044e-06,
+      "loss": 0.2403,
+      "step": 14273
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.620031205755862e-06,
+      "loss": 0.2311,
+      "step": 14274
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.618836913419163e-06,
+      "loss": 0.2231,
+      "step": 14275
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617642626527009e-06,
+      "loss": 0.2135,
+      "step": 14276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61644834509646e-06,
+      "loss": 0.2278,
+      "step": 14277
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615254069144574e-06,
+      "loss": 0.2233,
+      "step": 14278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614059798688414e-06,
+      "loss": 0.2689,
+      "step": 14279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61286553374503e-06,
+      "loss": 0.2084,
+      "step": 14280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61167127433149e-06,
+      "loss": 0.2052,
+      "step": 14281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.610477020464847e-06,
+      "loss": 0.2271,
+      "step": 14282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60928277216216e-06,
+      "loss": 0.2333,
+      "step": 14283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60808852944049e-06,
+      "loss": 0.2497,
+      "step": 14284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.606894292316899e-06,
+      "loss": 0.2377,
+      "step": 14285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605700060808438e-06,
+      "loss": 0.2478,
+      "step": 14286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604505834932168e-06,
+      "loss": 0.2266,
+      "step": 14287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.603311614705149e-06,
+      "loss": 0.2434,
+      "step": 14288
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.602117400144438e-06,
+      "loss": 0.2333,
+      "step": 14289
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.600923191267096e-06,
+      "loss": 0.2279,
+      "step": 14290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599728988090178e-06,
+      "loss": 0.229,
+      "step": 14291
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.598534790630747e-06,
+      "loss": 0.2379,
+      "step": 14292
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 0.2135,
+      "step": 14293
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.596146412932555e-06,
+      "loss": 0.1948,
+      "step": 14294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.594952232727917e-06,
+      "loss": 0.2336,
+      "step": 14295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59375805830899e-06,
+      "loss": 0.2226,
+      "step": 14296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.592563889692839e-06,
+      "loss": 0.2378,
+      "step": 14297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591369726896516e-06,
+      "loss": 0.2265,
+      "step": 14298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590175569937078e-06,
+      "loss": 0.214,
+      "step": 14299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588981418831586e-06,
+      "loss": 0.2197,
+      "step": 14300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587787273597096e-06,
+      "loss": 0.2301,
+      "step": 14301
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586593134250663e-06,
+      "loss": 0.2563,
+      "step": 14302
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.585399000809347e-06,
+      "loss": 0.2347,
+      "step": 14303
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.584204873290203e-06,
+      "loss": 0.2343,
+      "step": 14304
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.583010751710294e-06,
+      "loss": 0.2152,
+      "step": 14305
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.581816636086667e-06,
+      "loss": 0.2359,
+      "step": 14306
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.580622526436382e-06,
+      "loss": 0.2174,
+      "step": 14307
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.5794284227765e-06,
+      "loss": 0.239,
+      "step": 14308
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578234325124073e-06,
+      "loss": 0.201,
+      "step": 14309
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57704023349616e-06,
+      "loss": 0.2255,
+      "step": 14310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.575846147909816e-06,
+      "loss": 0.2117,
+      "step": 14311
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574652068382098e-06,
+      "loss": 0.2319,
+      "step": 14312
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.573457994930065e-06,
+      "loss": 0.2014,
+      "step": 14313
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572263927570768e-06,
+      "loss": 0.2436,
+      "step": 14314
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.571069866321267e-06,
+      "loss": 0.2343,
+      "step": 14315
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.569875811198615e-06,
+      "loss": 0.2333,
+      "step": 14316
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.568681762219872e-06,
+      "loss": 0.2401,
+      "step": 14317
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56748771940209e-06,
+      "loss": 0.2164,
+      "step": 14318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566293682762331e-06,
+      "loss": 0.2426,
+      "step": 14319
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56509965231764e-06,
+      "loss": 0.254,
+      "step": 14320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56390562808508e-06,
+      "loss": 0.2231,
+      "step": 14321
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.562711610081703e-06,
+      "loss": 0.2271,
+      "step": 14322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.561517598324568e-06,
+      "loss": 0.2026,
+      "step": 14323
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56032359283073e-06,
+      "loss": 0.2208,
+      "step": 14324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55912959361724e-06,
+      "loss": 0.2193,
+      "step": 14325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557935600701156e-06,
+      "loss": 0.2029,
+      "step": 14326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.556741614099532e-06,
+      "loss": 0.2486,
+      "step": 14327
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555547633829425e-06,
+      "loss": 0.2316,
+      "step": 14328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554353659907892e-06,
+      "loss": 0.2315,
+      "step": 14329
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.553159692351981e-06,
+      "loss": 0.2555,
+      "step": 14330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55196573117875e-06,
+      "loss": 0.2503,
+      "step": 14331
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55077177640526e-06,
+      "loss": 0.2124,
+      "step": 14332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549577828048553e-06,
+      "loss": 0.214,
+      "step": 14333
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54838388612569e-06,
+      "loss": 0.2613,
+      "step": 14334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547189950653727e-06,
+      "loss": 0.1933,
+      "step": 14335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545996021649714e-06,
+      "loss": 0.2294,
+      "step": 14336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544802099130706e-06,
+      "loss": 0.2041,
+      "step": 14337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54360818311376e-06,
+      "loss": 0.2412,
+      "step": 14338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.542414273615928e-06,
+      "loss": 0.2428,
+      "step": 14339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.541220370654265e-06,
+      "loss": 0.242,
+      "step": 14340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.540026474245822e-06,
+      "loss": 0.244,
+      "step": 14341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538832584407655e-06,
+      "loss": 0.2456,
+      "step": 14342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.537638701156817e-06,
+      "loss": 0.2199,
+      "step": 14343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536444824510361e-06,
+      "loss": 0.2337,
+      "step": 14344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.535250954485345e-06,
+      "loss": 0.2419,
+      "step": 14345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534057091098813e-06,
+      "loss": 0.2138,
+      "step": 14346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532863234367827e-06,
+      "loss": 0.2558,
+      "step": 14347
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.531669384309433e-06,
+      "loss": 0.2271,
+      "step": 14348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530475540940688e-06,
+      "loss": 0.2437,
+      "step": 14349
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.529281704278643e-06,
+      "loss": 0.2393,
+      "step": 14350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.528087874340353e-06,
+      "loss": 0.2351,
+      "step": 14351
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.526894051142872e-06,
+      "loss": 0.242,
+      "step": 14352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.525700234703249e-06,
+      "loss": 0.2228,
+      "step": 14353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524506425038537e-06,
+      "loss": 0.2346,
+      "step": 14354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523312622165788e-06,
+      "loss": 0.225,
+      "step": 14355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52211882610206e-06,
+      "loss": 0.2205,
+      "step": 14356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.520925036864397e-06,
+      "loss": 0.2309,
+      "step": 14357
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.519731254469859e-06,
+      "loss": 0.2301,
+      "step": 14358
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51853747893549e-06,
+      "loss": 0.2232,
+      "step": 14359
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517343710278347e-06,
+      "loss": 0.2318,
+      "step": 14360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51614994851548e-06,
+      "loss": 0.2106,
+      "step": 14361
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51495619366394e-06,
+      "loss": 0.2088,
+      "step": 14362
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513762445740785e-06,
+      "loss": 0.2514,
+      "step": 14363
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.512568704763057e-06,
+      "loss": 0.2289,
+      "step": 14364
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511374970747812e-06,
+      "loss": 0.1991,
+      "step": 14365
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510181243712103e-06,
+      "loss": 0.2411,
+      "step": 14366
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.508987523672978e-06,
+      "loss": 0.2065,
+      "step": 14367
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507793810647492e-06,
+      "loss": 0.2384,
+      "step": 14368
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506600104652692e-06,
+      "loss": 0.2327,
+      "step": 14369
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.505406405705631e-06,
+      "loss": 0.2508,
+      "step": 14370
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.504212713823363e-06,
+      "loss": 0.2459,
+      "step": 14371
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503019029022932e-06,
+      "loss": 0.2353,
+      "step": 14372
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50182535132139e-06,
+      "loss": 0.2056,
+      "step": 14373
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.500631680735794e-06,
+      "loss": 0.2288,
+      "step": 14374
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499438017283185e-06,
+      "loss": 0.2356,
+      "step": 14375
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49824436098062e-06,
+      "loss": 0.2127,
+      "step": 14376
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.497050711845146e-06,
+      "loss": 0.196,
+      "step": 14377
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.495857069893815e-06,
+      "loss": 0.2443,
+      "step": 14378
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494663435143681e-06,
+      "loss": 0.1959,
+      "step": 14379
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.493469807611783e-06,
+      "loss": 0.21,
+      "step": 14380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49227618731518e-06,
+      "loss": 0.2257,
+      "step": 14381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491082574270919e-06,
+      "loss": 0.225,
+      "step": 14382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.489888968496048e-06,
+      "loss": 0.2142,
+      "step": 14383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488695370007621e-06,
+      "loss": 0.2284,
+      "step": 14384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.487501778822685e-06,
+      "loss": 0.2254,
+      "step": 14385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.486308194958286e-06,
+      "loss": 0.2266,
+      "step": 14386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.485114618431477e-06,
+      "loss": 0.1981,
+      "step": 14387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.483921049259304e-06,
+      "loss": 0.2254,
+      "step": 14388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48272748745882e-06,
+      "loss": 0.2487,
+      "step": 14389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481533933047072e-06,
+      "loss": 0.2404,
+      "step": 14390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.480340386041108e-06,
+      "loss": 0.2221,
+      "step": 14391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479146846457976e-06,
+      "loss": 0.2416,
+      "step": 14392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.477953314314726e-06,
+      "loss": 0.2403,
+      "step": 14393
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.476759789628406e-06,
+      "loss": 0.2329,
+      "step": 14394
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475566272416067e-06,
+      "loss": 0.2384,
+      "step": 14395
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.474372762694753e-06,
+      "loss": 0.2114,
+      "step": 14396
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.473179260481514e-06,
+      "loss": 0.2456,
+      "step": 14397
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.471985765793401e-06,
+      "loss": 0.2502,
+      "step": 14398
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470792278647455e-06,
+      "loss": 0.2081,
+      "step": 14399
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46959879906073e-06,
+      "loss": 0.2062,
+      "step": 14400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.468405327050269e-06,
+      "loss": 0.2404,
+      "step": 14401
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467211862633124e-06,
+      "loss": 0.2409,
+      "step": 14402
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46601840582634e-06,
+      "loss": 0.2167,
+      "step": 14403
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.464824956646963e-06,
+      "loss": 0.2334,
+      "step": 14404
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.463631515112043e-06,
+      "loss": 0.216,
+      "step": 14405
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.462438081238628e-06,
+      "loss": 0.2305,
+      "step": 14406
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.461244655043762e-06,
+      "loss": 0.2167,
+      "step": 14407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.460051236544493e-06,
+      "loss": 0.2162,
+      "step": 14408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458857825757867e-06,
+      "loss": 0.2272,
+      "step": 14409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.457664422700932e-06,
+      "loss": 0.2304,
+      "step": 14410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45647102739074e-06,
+      "loss": 0.2425,
+      "step": 14411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.455277639844327e-06,
+      "loss": 0.2359,
+      "step": 14412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454084260078747e-06,
+      "loss": 0.2436,
+      "step": 14413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45289088811104e-06,
+      "loss": 0.2062,
+      "step": 14414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.451697523958257e-06,
+      "loss": 0.226,
+      "step": 14415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45050416763744e-06,
+      "loss": 0.2363,
+      "step": 14416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.449310819165641e-06,
+      "loss": 0.2299,
+      "step": 14417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.448117478559903e-06,
+      "loss": 0.2457,
+      "step": 14418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.446924145837269e-06,
+      "loss": 0.2294,
+      "step": 14419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.445730821014787e-06,
+      "loss": 0.2493,
+      "step": 14420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444537504109504e-06,
+      "loss": 0.2557,
+      "step": 14421
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.443344195138463e-06,
+      "loss": 0.2338,
+      "step": 14422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.442150894118708e-06,
+      "loss": 0.2134,
+      "step": 14423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440957601067294e-06,
+      "loss": 0.2209,
+      "step": 14424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439764316001249e-06,
+      "loss": 0.2171,
+      "step": 14425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43857103893763e-06,
+      "loss": 0.2274,
+      "step": 14426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437377769893478e-06,
+      "loss": 0.2153,
+      "step": 14427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43618450888584e-06,
+      "loss": 0.213,
+      "step": 14428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43499125593176e-06,
+      "loss": 0.2266,
+      "step": 14429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433798011048278e-06,
+      "loss": 0.242,
+      "step": 14430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.432604774252444e-06,
+      "loss": 0.2244,
+      "step": 14431
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.431411545561298e-06,
+      "loss": 0.2325,
+      "step": 14432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430218324991887e-06,
+      "loss": 0.2625,
+      "step": 14433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.429025112561257e-06,
+      "loss": 0.2141,
+      "step": 14434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427831908286448e-06,
+      "loss": 0.2165,
+      "step": 14435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.426638712184504e-06,
+      "loss": 0.2225,
+      "step": 14436
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425445524272474e-06,
+      "loss": 0.2228,
+      "step": 14437
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.424252344567391e-06,
+      "loss": 0.2558,
+      "step": 14438
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423059173086306e-06,
+      "loss": 0.2266,
+      "step": 14439
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.421866009846264e-06,
+      "loss": 0.2383,
+      "step": 14440
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420672854864302e-06,
+      "loss": 0.2311,
+      "step": 14441
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.419479708157463e-06,
+      "loss": 0.2096,
+      "step": 14442
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418286569742798e-06,
+      "loss": 0.2312,
+      "step": 14443
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.417093439637341e-06,
+      "loss": 0.2242,
+      "step": 14444
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415900317858142e-06,
+      "loss": 0.2344,
+      "step": 14445
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.414707204422238e-06,
+      "loss": 0.2504,
+      "step": 14446
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413514099346674e-06,
+      "loss": 0.2095,
+      "step": 14447
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412321002648491e-06,
+      "loss": 0.2561,
+      "step": 14448
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.411127914344732e-06,
+      "loss": 0.2159,
+      "step": 14449
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.409934834452444e-06,
+      "loss": 0.2261,
+      "step": 14450
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40874176298866e-06,
+      "loss": 0.2005,
+      "step": 14451
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407548699970429e-06,
+      "loss": 0.2136,
+      "step": 14452
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.406355645414786e-06,
+      "loss": 0.2368,
+      "step": 14453
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405162599338777e-06,
+      "loss": 0.2232,
+      "step": 14454
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.403969561759444e-06,
+      "loss": 0.2116,
+      "step": 14455
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.402776532693827e-06,
+      "loss": 0.2162,
+      "step": 14456
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40158351215897e-06,
+      "loss": 0.2337,
+      "step": 14457
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.400390500171908e-06,
+      "loss": 0.2222,
+      "step": 14458
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.399197496749687e-06,
+      "loss": 0.2389,
+      "step": 14459
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.398004501909346e-06,
+      "loss": 0.225,
+      "step": 14460
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39681151566793e-06,
+      "loss": 0.2418,
+      "step": 14461
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395618538042473e-06,
+      "loss": 0.2236,
+      "step": 14462
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.394425569050018e-06,
+      "loss": 0.2431,
+      "step": 14463
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.393232608707611e-06,
+      "loss": 0.2523,
+      "step": 14464
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392039657032282e-06,
+      "loss": 0.2115,
+      "step": 14465
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390846714041077e-06,
+      "loss": 0.2003,
+      "step": 14466
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389653779751037e-06,
+      "loss": 0.2316,
+      "step": 14467
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3884608541792e-06,
+      "loss": 0.21,
+      "step": 14468
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387267937342606e-06,
+      "loss": 0.2196,
+      "step": 14469
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386075029258295e-06,
+      "loss": 0.2296,
+      "step": 14470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.384882129943305e-06,
+      "loss": 0.2376,
+      "step": 14471
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.383689239414677e-06,
+      "loss": 0.2254,
+      "step": 14472
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.382496357689453e-06,
+      "loss": 0.2363,
+      "step": 14473
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381303484784667e-06,
+      "loss": 0.2266,
+      "step": 14474
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38011062071736e-06,
+      "loss": 0.2245,
+      "step": 14475
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.378917765504572e-06,
+      "loss": 0.2192,
+      "step": 14476
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.377724919163343e-06,
+      "loss": 0.2225,
+      "step": 14477
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376532081710707e-06,
+      "loss": 0.2347,
+      "step": 14478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375339253163708e-06,
+      "loss": 0.2341,
+      "step": 14479
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374146433539377e-06,
+      "loss": 0.2197,
+      "step": 14480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372953622854759e-06,
+      "loss": 0.2223,
+      "step": 14481
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37176082112689e-06,
+      "loss": 0.2583,
+      "step": 14482
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370568028372806e-06,
+      "loss": 0.2116,
+      "step": 14483
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36937524460955e-06,
+      "loss": 0.2355,
+      "step": 14484
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.368182469854155e-06,
+      "loss": 0.2016,
+      "step": 14485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36698970412366e-06,
+      "loss": 0.247,
+      "step": 14486
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365796947435102e-06,
+      "loss": 0.2169,
+      "step": 14487
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364604199805519e-06,
+      "loss": 0.2117,
+      "step": 14488
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.363411461251952e-06,
+      "loss": 0.2279,
+      "step": 14489
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362218731791432e-06,
+      "loss": 0.2274,
+      "step": 14490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361026011441e-06,
+      "loss": 0.2215,
+      "step": 14491
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.35983330021769e-06,
+      "loss": 0.2393,
+      "step": 14492
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.358640598138538e-06,
+      "loss": 0.2358,
+      "step": 14493
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357447905220585e-06,
+      "loss": 0.233,
+      "step": 14494
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356255221480866e-06,
+      "loss": 0.2551,
+      "step": 14495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355062546936413e-06,
+      "loss": 0.2271,
+      "step": 14496
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353869881604267e-06,
+      "loss": 0.2082,
+      "step": 14497
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.352677225501461e-06,
+      "loss": 0.2337,
+      "step": 14498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.351484578645035e-06,
+      "loss": 0.2259,
+      "step": 14499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.350291941052023e-06,
+      "loss": 0.2378,
+      "step": 14500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.349099312739459e-06,
+      "loss": 0.2254,
+      "step": 14501
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347906693724379e-06,
+      "loss": 0.2085,
+      "step": 14502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346714084023823e-06,
+      "loss": 0.2354,
+      "step": 14503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.345521483654818e-06,
+      "loss": 0.2111,
+      "step": 14504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.344328892634404e-06,
+      "loss": 0.2184,
+      "step": 14505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343136310979614e-06,
+      "loss": 0.2408,
+      "step": 14506
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341943738707488e-06,
+      "loss": 0.2064,
+      "step": 14507
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340751175835056e-06,
+      "loss": 0.213,
+      "step": 14508
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.339558622379352e-06,
+      "loss": 0.2423,
+      "step": 14509
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338366078357411e-06,
+      "loss": 0.2105,
+      "step": 14510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.337173543786272e-06,
+      "loss": 0.2374,
+      "step": 14511
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.335981018682963e-06,
+      "loss": 0.2315,
+      "step": 14512
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33478850306452e-06,
+      "loss": 0.2085,
+      "step": 14513
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33359599694798e-06,
+      "loss": 0.2162,
+      "step": 14514
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332403500350374e-06,
+      "loss": 0.2428,
+      "step": 14515
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.331211013288739e-06,
+      "loss": 0.2215,
+      "step": 14516
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.330018535780101e-06,
+      "loss": 0.2422,
+      "step": 14517
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3288260678415e-06,
+      "loss": 0.2474,
+      "step": 14518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.327633609489968e-06,
+      "loss": 0.2428,
+      "step": 14519
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.326441160742535e-06,
+      "loss": 0.2468,
+      "step": 14520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325248721616236e-06,
+      "loss": 0.2312,
+      "step": 14521
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324056292128105e-06,
+      "loss": 0.2263,
+      "step": 14522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322863872295175e-06,
+      "loss": 0.2355,
+      "step": 14523
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321671462134476e-06,
+      "loss": 0.2362,
+      "step": 14524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.320479061663042e-06,
+      "loss": 0.2253,
+      "step": 14525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.319286670897905e-06,
+      "loss": 0.2256,
+      "step": 14526
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3180942898561e-06,
+      "loss": 0.2408,
+      "step": 14527
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316901918554652e-06,
+      "loss": 0.2089,
+      "step": 14528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.315709557010603e-06,
+      "loss": 0.2197,
+      "step": 14529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314517205240976e-06,
+      "loss": 0.2067,
+      "step": 14530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.313324863262803e-06,
+      "loss": 0.2334,
+      "step": 14531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312132531093118e-06,
+      "loss": 0.2285,
+      "step": 14532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.310940208748954e-06,
+      "loss": 0.2428,
+      "step": 14533
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309747896247342e-06,
+      "loss": 0.2431,
+      "step": 14534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.308555593605309e-06,
+      "loss": 0.221,
+      "step": 14535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307363300839888e-06,
+      "loss": 0.2037,
+      "step": 14536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30617101796811e-06,
+      "loss": 0.2131,
+      "step": 14537
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.304978745007005e-06,
+      "loss": 0.2359,
+      "step": 14538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.303786481973606e-06,
+      "loss": 0.23,
+      "step": 14539
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302594228884941e-06,
+      "loss": 0.2083,
+      "step": 14540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30140198575804e-06,
+      "loss": 0.2454,
+      "step": 14541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300209752609933e-06,
+      "loss": 0.2111,
+      "step": 14542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299017529457656e-06,
+      "loss": 0.2374,
+      "step": 14543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297825316318228e-06,
+      "loss": 0.2199,
+      "step": 14544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296633113208685e-06,
+      "loss": 0.2449,
+      "step": 14545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295440920146057e-06,
+      "loss": 0.2137,
+      "step": 14546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.294248737147368e-06,
+      "loss": 0.2491,
+      "step": 14547
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293056564229651e-06,
+      "loss": 0.2094,
+      "step": 14548
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291864401409937e-06,
+      "loss": 0.2436,
+      "step": 14549
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290672248705254e-06,
+      "loss": 0.2296,
+      "step": 14550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.289480106132627e-06,
+      "loss": 0.2311,
+      "step": 14551
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288287973709088e-06,
+      "loss": 0.2239,
+      "step": 14552
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287095851451662e-06,
+      "loss": 0.236,
+      "step": 14553
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285903739377382e-06,
+      "loss": 0.2298,
+      "step": 14554
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.284711637503276e-06,
+      "loss": 0.2187,
+      "step": 14555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28351954584637e-06,
+      "loss": 0.2404,
+      "step": 14556
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282327464423692e-06,
+      "loss": 0.2225,
+      "step": 14557
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.281135393252267e-06,
+      "loss": 0.2474,
+      "step": 14558
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279943332349125e-06,
+      "loss": 0.2364,
+      "step": 14559
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278751281731294e-06,
+      "loss": 0.2197,
+      "step": 14560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.277559241415801e-06,
+      "loss": 0.2237,
+      "step": 14561
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276367211419676e-06,
+      "loss": 0.2386,
+      "step": 14562
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27517519175994e-06,
+      "loss": 0.2023,
+      "step": 14563
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273983182453622e-06,
+      "loss": 0.2359,
+      "step": 14564
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.272791183517753e-06,
+      "loss": 0.2227,
+      "step": 14565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271599194969356e-06,
+      "loss": 0.225,
+      "step": 14566
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270407216825455e-06,
+      "loss": 0.2336,
+      "step": 14567
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26921524910308e-06,
+      "loss": 0.2257,
+      "step": 14568
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268023291819259e-06,
+      "loss": 0.2496,
+      "step": 14569
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266831344991013e-06,
+      "loss": 0.2152,
+      "step": 14570
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.265639408635367e-06,
+      "loss": 0.2153,
+      "step": 14571
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26444748276935e-06,
+      "loss": 0.2408,
+      "step": 14572
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263255567409989e-06,
+      "loss": 0.2357,
+      "step": 14573
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262063662574306e-06,
+      "loss": 0.2352,
+      "step": 14574
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.260871768279326e-06,
+      "loss": 0.2277,
+      "step": 14575
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259679884542076e-06,
+      "loss": 0.21,
+      "step": 14576
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258488011379581e-06,
+      "loss": 0.2606,
+      "step": 14577
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257296148808867e-06,
+      "loss": 0.2363,
+      "step": 14578
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.256104296846954e-06,
+      "loss": 0.2519,
+      "step": 14579
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25491245551087e-06,
+      "loss": 0.2379,
+      "step": 14580
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253720624817637e-06,
+      "loss": 0.2303,
+      "step": 14581
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252528804784287e-06,
+      "loss": 0.2608,
+      "step": 14582
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.251336995427832e-06,
+      "loss": 0.2302,
+      "step": 14583
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.250145196765303e-06,
+      "loss": 0.2363,
+      "step": 14584
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24895340881372e-06,
+      "loss": 0.2326,
+      "step": 14585
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24776163159011e-06,
+      "loss": 0.2071,
+      "step": 14586
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246569865111494e-06,
+      "loss": 0.2309,
+      "step": 14587
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.245378109394896e-06,
+      "loss": 0.2299,
+      "step": 14588
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.244186364457342e-06,
+      "loss": 0.2531,
+      "step": 14589
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.24299463031585e-06,
+      "loss": 0.2143,
+      "step": 14590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.241802906987446e-06,
+      "loss": 0.1977,
+      "step": 14591
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.24061119448915e-06,
+      "loss": 0.2092,
+      "step": 14592
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.239419492837986e-06,
+      "loss": 0.2437,
+      "step": 14593
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238227802050981e-06,
+      "loss": 0.2316,
+      "step": 14594
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.237036122145151e-06,
+      "loss": 0.2469,
+      "step": 14595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235844453137518e-06,
+      "loss": 0.2308,
+      "step": 14596
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.234652795045106e-06,
+      "loss": 0.2145,
+      "step": 14597
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233461147884935e-06,
+      "loss": 0.2334,
+      "step": 14598
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.232269511674026e-06,
+      "loss": 0.2255,
+      "step": 14599
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231077886429406e-06,
+      "loss": 0.2234,
+      "step": 14600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.229886272168089e-06,
+      "loss": 0.2055,
+      "step": 14601
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.2286946689071e-06,
+      "loss": 0.2313,
+      "step": 14602
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227503076663457e-06,
+      "loss": 0.2153,
+      "step": 14603
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226311495454186e-06,
+      "loss": 0.2257,
+      "step": 14604
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225119925296304e-06,
+      "loss": 0.1983,
+      "step": 14605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22392836620683e-06,
+      "loss": 0.2125,
+      "step": 14606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.222736818202787e-06,
+      "loss": 0.2209,
+      "step": 14607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221545281301198e-06,
+      "loss": 0.2364,
+      "step": 14608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220353755519076e-06,
+      "loss": 0.2244,
+      "step": 14609
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219162240873443e-06,
+      "loss": 0.2265,
+      "step": 14610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21797073738132e-06,
+      "loss": 0.2475,
+      "step": 14611
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216779245059727e-06,
+      "loss": 0.2267,
+      "step": 14612
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215587763925683e-06,
+      "loss": 0.2428,
+      "step": 14613
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214396293996203e-06,
+      "loss": 0.2139,
+      "step": 14614
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213204835288311e-06,
+      "loss": 0.2443,
+      "step": 14615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.212013387819027e-06,
+      "loss": 0.2119,
+      "step": 14616
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210821951605364e-06,
+      "loss": 0.2185,
+      "step": 14617
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209630526664344e-06,
+      "loss": 0.2343,
+      "step": 14618
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208439113012984e-06,
+      "loss": 0.2174,
+      "step": 14619
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207247710668304e-06,
+      "loss": 0.2411,
+      "step": 14620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206056319647323e-06,
+      "loss": 0.2395,
+      "step": 14621
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204864939967056e-06,
+      "loss": 0.2279,
+      "step": 14622
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203673571644522e-06,
+      "loss": 0.2366,
+      "step": 14623
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202482214696738e-06,
+      "loss": 0.2332,
+      "step": 14624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20129086914072e-06,
+      "loss": 0.2388,
+      "step": 14625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200099534993486e-06,
+      "loss": 0.2406,
+      "step": 14626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198908212272055e-06,
+      "loss": 0.238,
+      "step": 14627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.197716900993444e-06,
+      "loss": 0.2734,
+      "step": 14628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196525601174667e-06,
+      "loss": 0.2575,
+      "step": 14629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195334312832742e-06,
+      "loss": 0.2265,
+      "step": 14630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194143035984685e-06,
+      "loss": 0.2498,
+      "step": 14631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192951770647515e-06,
+      "loss": 0.229,
+      "step": 14632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191760516838243e-06,
+      "loss": 0.2334,
+      "step": 14633
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190569274573888e-06,
+      "loss": 0.224,
+      "step": 14634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18937804387147e-06,
+      "loss": 0.2023,
+      "step": 14635
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188186824747996e-06,
+      "loss": 0.2264,
+      "step": 14636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.186995617220484e-06,
+      "loss": 0.2191,
+      "step": 14637
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185804421305952e-06,
+      "loss": 0.2318,
+      "step": 14638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184613237021415e-06,
+      "loss": 0.2104,
+      "step": 14639
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183422064383885e-06,
+      "loss": 0.2319,
+      "step": 14640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182230903410376e-06,
+      "loss": 0.2143,
+      "step": 14641
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.181039754117908e-06,
+      "loss": 0.2279,
+      "step": 14642
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179848616523491e-06,
+      "loss": 0.2187,
+      "step": 14643
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178657490644141e-06,
+      "loss": 0.2191,
+      "step": 14644
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177466376496872e-06,
+      "loss": 0.2282,
+      "step": 14645
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176275274098697e-06,
+      "loss": 0.2223,
+      "step": 14646
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175084183466629e-06,
+      "loss": 0.221,
+      "step": 14647
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.173893104617688e-06,
+      "loss": 0.2143,
+      "step": 14648
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172702037568879e-06,
+      "loss": 0.2224,
+      "step": 14649
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.171510982337217e-06,
+      "loss": 0.2475,
+      "step": 14650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17031993893972e-06,
+      "loss": 0.2138,
+      "step": 14651
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.169128907393394e-06,
+      "loss": 0.2355,
+      "step": 14652
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167937887715254e-06,
+      "loss": 0.231,
+      "step": 14653
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.166746879922317e-06,
+      "loss": 0.2041,
+      "step": 14654
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165555884031591e-06,
+      "loss": 0.2164,
+      "step": 14655
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16436490006009e-06,
+      "loss": 0.2112,
+      "step": 14656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163173928024825e-06,
+      "loss": 0.2302,
+      "step": 14657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.161982967942806e-06,
+      "loss": 0.2254,
+      "step": 14658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160792019831049e-06,
+      "loss": 0.2234,
+      "step": 14659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.159601083706565e-06,
+      "loss": 0.2613,
+      "step": 14660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158410159586365e-06,
+      "loss": 0.2233,
+      "step": 14661
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.157219247487458e-06,
+      "loss": 0.215,
+      "step": 14662
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.156028347426853e-06,
+      "loss": 0.2364,
+      "step": 14663
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154837459421565e-06,
+      "loss": 0.2392,
+      "step": 14664
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153646583488604e-06,
+      "loss": 0.24,
+      "step": 14665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15245571964498e-06,
+      "loss": 0.2274,
+      "step": 14666
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.151264867907705e-06,
+      "loss": 0.2264,
+      "step": 14667
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150074028293787e-06,
+      "loss": 0.241,
+      "step": 14668
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148883200820236e-06,
+      "loss": 0.2415,
+      "step": 14669
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.147692385504061e-06,
+      "loss": 0.2383,
+      "step": 14670
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.146501582362277e-06,
+      "loss": 0.2247,
+      "step": 14671
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145310791411886e-06,
+      "loss": 0.2243,
+      "step": 14672
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.144120012669903e-06,
+      "loss": 0.2142,
+      "step": 14673
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.142929246153337e-06,
+      "loss": 0.2387,
+      "step": 14674
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.141738491879191e-06,
+      "loss": 0.2298,
+      "step": 14675
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.140547749864479e-06,
+      "loss": 0.2375,
+      "step": 14676
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.139357020126208e-06,
+      "loss": 0.2412,
+      "step": 14677
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.138166302681388e-06,
+      "loss": 0.2158,
+      "step": 14678
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.136975597547023e-06,
+      "loss": 0.2281,
+      "step": 14679
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135784904740126e-06,
+      "loss": 0.2147,
+      "step": 14680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.134594224277701e-06,
+      "loss": 0.2165,
+      "step": 14681
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133403556176759e-06,
+      "loss": 0.2315,
+      "step": 14682
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.132212900454309e-06,
+      "loss": 0.2263,
+      "step": 14683
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13102225712735e-06,
+      "loss": 0.2371,
+      "step": 14684
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129831626212898e-06,
+      "loss": 0.228,
+      "step": 14685
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128641007727956e-06,
+      "loss": 0.2189,
+      "step": 14686
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.127450401689534e-06,
+      "loss": 0.2451,
+      "step": 14687
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126259808114635e-06,
+      "loss": 0.2283,
+      "step": 14688
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125069227020266e-06,
+      "loss": 0.2192,
+      "step": 14689
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123878658423432e-06,
+      "loss": 0.2481,
+      "step": 14690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122688102341144e-06,
+      "loss": 0.2273,
+      "step": 14691
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121497558790403e-06,
+      "loss": 0.2328,
+      "step": 14692
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120307027788217e-06,
+      "loss": 0.233,
+      "step": 14693
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.119116509351592e-06,
+      "loss": 0.2224,
+      "step": 14694
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.117926003497532e-06,
+      "loss": 0.2294,
+      "step": 14695
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116735510243043e-06,
+      "loss": 0.198,
+      "step": 14696
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115545029605129e-06,
+      "loss": 0.2053,
+      "step": 14697
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.114354561600797e-06,
+      "loss": 0.2082,
+      "step": 14698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113164106247054e-06,
+      "loss": 0.2336,
+      "step": 14699
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111973663560898e-06,
+      "loss": 0.2422,
+      "step": 14700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110783233559338e-06,
+      "loss": 0.2192,
+      "step": 14701
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109592816259374e-06,
+      "loss": 0.2062,
+      "step": 14702
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.108402411678014e-06,
+      "loss": 0.2101,
+      "step": 14703
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107212019832258e-06,
+      "loss": 0.2013,
+      "step": 14704
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106021640739116e-06,
+      "loss": 0.2454,
+      "step": 14705
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104831274415583e-06,
+      "loss": 0.2346,
+      "step": 14706
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103640920878667e-06,
+      "loss": 0.2454,
+      "step": 14707
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10245058014537e-06,
+      "loss": 0.2274,
+      "step": 14708
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.101260252232698e-06,
+      "loss": 0.2163,
+      "step": 14709
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10006993715765e-06,
+      "loss": 0.2239,
+      "step": 14710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09887963493723e-06,
+      "loss": 0.2466,
+      "step": 14711
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.097689345588439e-06,
+      "loss": 0.2423,
+      "step": 14712
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096499069128279e-06,
+      "loss": 0.2098,
+      "step": 14713
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095308805573759e-06,
+      "loss": 0.2621,
+      "step": 14714
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09411855494187e-06,
+      "loss": 0.2478,
+      "step": 14715
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092928317249616e-06,
+      "loss": 0.2206,
+      "step": 14716
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091738092514004e-06,
+      "loss": 0.2292,
+      "step": 14717
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09054788075203e-06,
+      "loss": 0.2286,
+      "step": 14718
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.089357681980697e-06,
+      "loss": 0.2359,
+      "step": 14719
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088167496217007e-06,
+      "loss": 0.2368,
+      "step": 14720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.086977323477959e-06,
+      "loss": 0.2487,
+      "step": 14721
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.085787163780552e-06,
+      "loss": 0.2305,
+      "step": 14722
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08459701714179e-06,
+      "loss": 0.2408,
+      "step": 14723
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08340688357867e-06,
+      "loss": 0.23,
+      "step": 14724
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.082216763108194e-06,
+      "loss": 0.2335,
+      "step": 14725
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08102665574736e-06,
+      "loss": 0.2172,
+      "step": 14726
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079836561513173e-06,
+      "loss": 0.2192,
+      "step": 14727
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.078646480422624e-06,
+      "loss": 0.2085,
+      "step": 14728
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.077456412492716e-06,
+      "loss": 0.2207,
+      "step": 14729
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.076266357740447e-06,
+      "loss": 0.2387,
+      "step": 14730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075076316182817e-06,
+      "loss": 0.2448,
+      "step": 14731
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.073886287836822e-06,
+      "loss": 0.2373,
+      "step": 14732
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072696272719465e-06,
+      "loss": 0.2147,
+      "step": 14733
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.071506270847742e-06,
+      "loss": 0.2263,
+      "step": 14734
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07031628223865e-06,
+      "loss": 0.25,
+      "step": 14735
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069126306909187e-06,
+      "loss": 0.2118,
+      "step": 14736
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067936344876351e-06,
+      "loss": 0.2104,
+      "step": 14737
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.066746396157142e-06,
+      "loss": 0.2324,
+      "step": 14738
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065556460768554e-06,
+      "loss": 0.2123,
+      "step": 14739
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.064366538727587e-06,
+      "loss": 0.2248,
+      "step": 14740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.063176630051235e-06,
+      "loss": 0.2283,
+      "step": 14741
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.061986734756495e-06,
+      "loss": 0.2204,
+      "step": 14742
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.060796852860362e-06,
+      "loss": 0.2257,
+      "step": 14743
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.059606984379838e-06,
+      "loss": 0.2251,
+      "step": 14744
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.058417129331912e-06,
+      "loss": 0.2591,
+      "step": 14745
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.057227287733585e-06,
+      "loss": 0.2154,
+      "step": 14746
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05603745960185e-06,
+      "loss": 0.2446,
+      "step": 14747
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.054847644953704e-06,
+      "loss": 0.2379,
+      "step": 14748
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.053657843806145e-06,
+      "loss": 0.2344,
+      "step": 14749
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.052468056176163e-06,
+      "loss": 0.2561,
+      "step": 14750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.051278282080754e-06,
+      "loss": 0.2483,
+      "step": 14751
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.050088521536916e-06,
+      "loss": 0.2355,
+      "step": 14752
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.048898774561644e-06,
+      "loss": 0.2314,
+      "step": 14753
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.047709041171926e-06,
+      "loss": 0.2085,
+      "step": 14754
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.046519321384762e-06,
+      "loss": 0.2115,
+      "step": 14755
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.045329615217144e-06,
+      "loss": 0.2526,
+      "step": 14756
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044139922686065e-06,
+      "loss": 0.2486,
+      "step": 14757
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.04295024380852e-06,
+      "loss": 0.2156,
+      "step": 14758
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.0417605786015e-06,
+      "loss": 0.2397,
+      "step": 14759
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.040570927082003e-06,
+      "loss": 0.2427,
+      "step": 14760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039381289267018e-06,
+      "loss": 0.2043,
+      "step": 14761
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.038191665173538e-06,
+      "loss": 0.2298,
+      "step": 14762
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037002054818558e-06,
+      "loss": 0.23,
+      "step": 14763
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035812458219068e-06,
+      "loss": 0.2504,
+      "step": 14764
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.034622875392063e-06,
+      "loss": 0.2064,
+      "step": 14765
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.033433306354533e-06,
+      "loss": 0.2113,
+      "step": 14766
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.032243751123472e-06,
+      "loss": 0.2407,
+      "step": 14767
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.031054209715866e-06,
+      "loss": 0.236,
+      "step": 14768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02986468214871e-06,
+      "loss": 0.2347,
+      "step": 14769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028675168438995e-06,
+      "loss": 0.2241,
+      "step": 14770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.027485668603716e-06,
+      "loss": 0.2592,
+      "step": 14771
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02629618265986e-06,
+      "loss": 0.2333,
+      "step": 14772
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.025106710624416e-06,
+      "loss": 0.2086,
+      "step": 14773
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.023917252514376e-06,
+      "loss": 0.2262,
+      "step": 14774
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022727808346731e-06,
+      "loss": 0.2666,
+      "step": 14775
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.021538378138473e-06,
+      "loss": 0.2338,
+      "step": 14776
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.020348961906589e-06,
+      "loss": 0.2184,
+      "step": 14777
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019159559668068e-06,
+      "loss": 0.2252,
+      "step": 14778
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.0179701714399e-06,
+      "loss": 0.2422,
+      "step": 14779
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01678079723908e-06,
+      "loss": 0.2178,
+      "step": 14780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01559143708259e-06,
+      "loss": 0.239,
+      "step": 14781
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.014402090987417e-06,
+      "loss": 0.2297,
+      "step": 14782
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013212758970559e-06,
+      "loss": 0.2502,
+      "step": 14783
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.012023441048994e-06,
+      "loss": 0.2572,
+      "step": 14784
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010834137239719e-06,
+      "loss": 0.2152,
+      "step": 14785
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.009644847559715e-06,
+      "loss": 0.2376,
+      "step": 14786
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008455572025973e-06,
+      "loss": 0.2135,
+      "step": 14787
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007266310655486e-06,
+      "loss": 0.2342,
+      "step": 14788
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.006077063465232e-06,
+      "loss": 0.2631,
+      "step": 14789
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004887830472203e-06,
+      "loss": 0.2476,
+      "step": 14790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003698611693385e-06,
+      "loss": 0.1998,
+      "step": 14791
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.002509407145768e-06,
+      "loss": 0.2518,
+      "step": 14792
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001320216846336e-06,
+      "loss": 0.2306,
+      "step": 14793
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.000131040812077e-06,
+      "loss": 0.217,
+      "step": 14794
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998941879059971e-06,
+      "loss": 0.2202,
+      "step": 14795
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99775273160701e-06,
+      "loss": 0.2092,
+      "step": 14796
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.996563598470179e-06,
+      "loss": 0.2339,
+      "step": 14797
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995374479666461e-06,
+      "loss": 0.2389,
+      "step": 14798
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994185375212848e-06,
+      "loss": 0.2396,
+      "step": 14799
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.992996285126317e-06,
+      "loss": 0.2501,
+      "step": 14800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.991807209423858e-06,
+      "loss": 0.2412,
+      "step": 14801
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990618148122455e-06,
+      "loss": 0.2295,
+      "step": 14802
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98942910123909e-06,
+      "loss": 0.2274,
+      "step": 14803
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.988240068790753e-06,
+      "loss": 0.2266,
+      "step": 14804
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987051050794423e-06,
+      "loss": 0.2248,
+      "step": 14805
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.985862047267089e-06,
+      "loss": 0.2308,
+      "step": 14806
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984673058225728e-06,
+      "loss": 0.233,
+      "step": 14807
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983484083687325e-06,
+      "loss": 0.2014,
+      "step": 14808
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.982295123668867e-06,
+      "loss": 0.2192,
+      "step": 14809
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981106178187337e-06,
+      "loss": 0.2213,
+      "step": 14810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979917247259715e-06,
+      "loss": 0.2314,
+      "step": 14811
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.978728330902985e-06,
+      "loss": 0.21,
+      "step": 14812
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97753942913413e-06,
+      "loss": 0.251,
+      "step": 14813
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97635054197013e-06,
+      "loss": 0.222,
+      "step": 14814
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.975161669427973e-06,
+      "loss": 0.2093,
+      "step": 14815
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.973972811524634e-06,
+      "loss": 0.2257,
+      "step": 14816
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.972783968277098e-06,
+      "loss": 0.2408,
+      "step": 14817
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.971595139702345e-06,
+      "loss": 0.2518,
+      "step": 14818
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970406325817364e-06,
+      "loss": 0.2302,
+      "step": 14819
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969217526639122e-06,
+      "loss": 0.2311,
+      "step": 14820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96802874218461e-06,
+      "loss": 0.2351,
+      "step": 14821
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.966839972470807e-06,
+      "loss": 0.22,
+      "step": 14822
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965651217514689e-06,
+      "loss": 0.2198,
+      "step": 14823
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.964462477333242e-06,
+      "loss": 0.2361,
+      "step": 14824
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.963273751943442e-06,
+      "loss": 0.2441,
+      "step": 14825
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962085041362271e-06,
+      "loss": 0.2288,
+      "step": 14826
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.960896345606708e-06,
+      "loss": 0.2108,
+      "step": 14827
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95970766469373e-06,
+      "loss": 0.2217,
+      "step": 14828
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958518998640322e-06,
+      "loss": 0.2083,
+      "step": 14829
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957330347463456e-06,
+      "loss": 0.222,
+      "step": 14830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.956141711180117e-06,
+      "loss": 0.2002,
+      "step": 14831
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954953089807281e-06,
+      "loss": 0.2194,
+      "step": 14832
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.953764483361925e-06,
+      "loss": 0.2307,
+      "step": 14833
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.952575891861027e-06,
+      "loss": 0.2323,
+      "step": 14834
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951387315321565e-06,
+      "loss": 0.2191,
+      "step": 14835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950198753760519e-06,
+      "loss": 0.2402,
+      "step": 14836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.949010207194863e-06,
+      "loss": 0.2116,
+      "step": 14837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.947821675641578e-06,
+      "loss": 0.237,
+      "step": 14838
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94663315911764e-06,
+      "loss": 0.2649,
+      "step": 14839
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945444657640022e-06,
+      "loss": 0.2289,
+      "step": 14840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944256171225704e-06,
+      "loss": 0.2362,
+      "step": 14841
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943067699891663e-06,
+      "loss": 0.271,
+      "step": 14842
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941879243654876e-06,
+      "loss": 0.2171,
+      "step": 14843
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.940690802532316e-06,
+      "loss": 0.2301,
+      "step": 14844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.939502376540963e-06,
+      "loss": 0.2456,
+      "step": 14845
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.938313965697785e-06,
+      "loss": 0.24,
+      "step": 14846
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.937125570019761e-06,
+      "loss": 0.2301,
+      "step": 14847
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935937189523867e-06,
+      "loss": 0.2326,
+      "step": 14848
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93474882422708e-06,
+      "loss": 0.2196,
+      "step": 14849
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.933560474146371e-06,
+      "loss": 0.1898,
+      "step": 14850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.932372139298715e-06,
+      "loss": 0.2183,
+      "step": 14851
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.931183819701087e-06,
+      "loss": 0.208,
+      "step": 14852
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92999551537046e-06,
+      "loss": 0.2456,
+      "step": 14853
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92880722632381e-06,
+      "loss": 0.2056,
+      "step": 14854
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92761895257811e-06,
+      "loss": 0.2023,
+      "step": 14855
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92643069415033e-06,
+      "loss": 0.2266,
+      "step": 14856
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925242451057445e-06,
+      "loss": 0.2344,
+      "step": 14857
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92405422331643e-06,
+      "loss": 0.2134,
+      "step": 14858
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.922866010944261e-06,
+      "loss": 0.2166,
+      "step": 14859
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.921677813957901e-06,
+      "loss": 0.2666,
+      "step": 14860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.920489632374328e-06,
+      "loss": 0.2205,
+      "step": 14861
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.919301466210511e-06,
+      "loss": 0.2317,
+      "step": 14862
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918113315483424e-06,
+      "loss": 0.2319,
+      "step": 14863
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.916925180210038e-06,
+      "loss": 0.2188,
+      "step": 14864
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.915737060407328e-06,
+      "loss": 0.2063,
+      "step": 14865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.914548956092258e-06,
+      "loss": 0.2209,
+      "step": 14866
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913360867281804e-06,
+      "loss": 0.1942,
+      "step": 14867
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912172793992934e-06,
+      "loss": 0.2295,
+      "step": 14868
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910984736242621e-06,
+      "loss": 0.2246,
+      "step": 14869
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.909796694047838e-06,
+      "loss": 0.2216,
+      "step": 14870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.908608667425547e-06,
+      "loss": 0.2173,
+      "step": 14871
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907420656392726e-06,
+      "loss": 0.2061,
+      "step": 14872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.906232660966337e-06,
+      "loss": 0.2387,
+      "step": 14873
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905044681163351e-06,
+      "loss": 0.2265,
+      "step": 14874
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903856717000741e-06,
+      "loss": 0.2126,
+      "step": 14875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.902668768495474e-06,
+      "loss": 0.2533,
+      "step": 14876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90148083566452e-06,
+      "loss": 0.2417,
+      "step": 14877
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.900292918524844e-06,
+      "loss": 0.2332,
+      "step": 14878
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899105017093418e-06,
+      "loss": 0.2459,
+      "step": 14879
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897917131387205e-06,
+      "loss": 0.2373,
+      "step": 14880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896729261423182e-06,
+      "loss": 0.2226,
+      "step": 14881
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.895541407218306e-06,
+      "loss": 0.2236,
+      "step": 14882
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.894353568789552e-06,
+      "loss": 0.2449,
+      "step": 14883
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89316574615388e-06,
+      "loss": 0.2415,
+      "step": 14884
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.891977939328269e-06,
+      "loss": 0.2264,
+      "step": 14885
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.890790148329673e-06,
+      "loss": 0.2321,
+      "step": 14886
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.889602373175062e-06,
+      "loss": 0.2058,
+      "step": 14887
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.888414613881405e-06,
+      "loss": 0.2286,
+      "step": 14888
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887226870465666e-06,
+      "loss": 0.2472,
+      "step": 14889
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88603914294481e-06,
+      "loss": 0.2121,
+      "step": 14890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.884851431335805e-06,
+      "loss": 0.2174,
+      "step": 14891
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.883663735655612e-06,
+      "loss": 0.2257,
+      "step": 14892
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882476055921206e-06,
+      "loss": 0.2282,
+      "step": 14893
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.881288392149539e-06,
+      "loss": 0.2425,
+      "step": 14894
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880100744357582e-06,
+      "loss": 0.2316,
+      "step": 14895
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878913112562301e-06,
+      "loss": 0.2114,
+      "step": 14896
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87772549678066e-06,
+      "loss": 0.2204,
+      "step": 14897
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876537897029621e-06,
+      "loss": 0.2517,
+      "step": 14898
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875350313326145e-06,
+      "loss": 0.2183,
+      "step": 14899
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.874162745687198e-06,
+      "loss": 0.2219,
+      "step": 14900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872975194129744e-06,
+      "loss": 0.2436,
+      "step": 14901
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871787658670744e-06,
+      "loss": 0.2134,
+      "step": 14902
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870600139327165e-06,
+      "loss": 0.1997,
+      "step": 14903
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.869412636115968e-06,
+      "loss": 0.2507,
+      "step": 14904
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868225149054112e-06,
+      "loss": 0.238,
+      "step": 14905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86703767815856e-06,
+      "loss": 0.2197,
+      "step": 14906
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865850223446276e-06,
+      "loss": 0.241,
+      "step": 14907
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.864662784934222e-06,
+      "loss": 0.2411,
+      "step": 14908
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86347536263936e-06,
+      "loss": 0.2394,
+      "step": 14909
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.862287956578645e-06,
+      "loss": 0.2366,
+      "step": 14910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86110056676905e-06,
+      "loss": 0.2236,
+      "step": 14911
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859913193227521e-06,
+      "loss": 0.2422,
+      "step": 14912
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858725835971028e-06,
+      "loss": 0.2649,
+      "step": 14913
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.857538495016527e-06,
+      "loss": 0.2484,
+      "step": 14914
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856351170380982e-06,
+      "loss": 0.2195,
+      "step": 14915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855163862081348e-06,
+      "loss": 0.2245,
+      "step": 14916
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85397657013459e-06,
+      "loss": 0.2128,
+      "step": 14917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.852789294557662e-06,
+      "loss": 0.206,
+      "step": 14918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851602035367526e-06,
+      "loss": 0.2152,
+      "step": 14919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850414792581142e-06,
+      "loss": 0.221,
+      "step": 14920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.849227566215466e-06,
+      "loss": 0.2064,
+      "step": 14921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848040356287457e-06,
+      "loss": 0.2216,
+      "step": 14922
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.846853162814072e-06,
+      "loss": 0.2273,
+      "step": 14923
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845665985812277e-06,
+      "loss": 0.217,
+      "step": 14924
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844478825299018e-06,
+      "loss": 0.2165,
+      "step": 14925
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.843291681291257e-06,
+      "loss": 0.2331,
+      "step": 14926
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842104553805954e-06,
+      "loss": 0.2237,
+      "step": 14927
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.840917442860062e-06,
+      "loss": 0.2247,
+      "step": 14928
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.839730348470538e-06,
+      "loss": 0.2302,
+      "step": 14929
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838543270654341e-06,
+      "loss": 0.23,
+      "step": 14930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.837356209428428e-06,
+      "loss": 0.1989,
+      "step": 14931
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836169164809752e-06,
+      "loss": 0.2294,
+      "step": 14932
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83498213681527e-06,
+      "loss": 0.2227,
+      "step": 14933
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.833795125461937e-06,
+      "loss": 0.2277,
+      "step": 14934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83260813076671e-06,
+      "loss": 0.2343,
+      "step": 14935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831421152746541e-06,
+      "loss": 0.2126,
+      "step": 14936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83023419141839e-06,
+      "loss": 0.2029,
+      "step": 14937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.829047246799209e-06,
+      "loss": 0.2186,
+      "step": 14938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827860318905948e-06,
+      "loss": 0.1935,
+      "step": 14939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.826673407755563e-06,
+      "loss": 0.2227,
+      "step": 14940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825486513365013e-06,
+      "loss": 0.2253,
+      "step": 14941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.824299635751246e-06,
+      "loss": 0.2283,
+      "step": 14942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.823112774931222e-06,
+      "loss": 0.2462,
+      "step": 14943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.821925930921885e-06,
+      "loss": 0.22,
+      "step": 14944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.820739103740194e-06,
+      "loss": 0.2273,
+      "step": 14945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8195522934031e-06,
+      "loss": 0.224,
+      "step": 14946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818365499927557e-06,
+      "loss": 0.2232,
+      "step": 14947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817178723330517e-06,
+      "loss": 0.2397,
+      "step": 14948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.815991963628928e-06,
+      "loss": 0.2285,
+      "step": 14949
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.814805220839747e-06,
+      "loss": 0.2326,
+      "step": 14950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813618494979927e-06,
+      "loss": 0.2163,
+      "step": 14951
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.812431786066411e-06,
+      "loss": 0.218,
+      "step": 14952
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811245094116153e-06,
+      "loss": 0.2097,
+      "step": 14953
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.810058419146108e-06,
+      "loss": 0.242,
+      "step": 14954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.808871761173222e-06,
+      "loss": 0.2418,
+      "step": 14955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807685120214446e-06,
+      "loss": 0.2018,
+      "step": 14956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806498496286732e-06,
+      "loss": 0.2457,
+      "step": 14957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.805311889407028e-06,
+      "loss": 0.2205,
+      "step": 14958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.804125299592287e-06,
+      "loss": 0.2105,
+      "step": 14959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80293872685945e-06,
+      "loss": 0.2363,
+      "step": 14960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.801752171225475e-06,
+      "loss": 0.255,
+      "step": 14961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.800565632707307e-06,
+      "loss": 0.2298,
+      "step": 14962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.799379111321895e-06,
+      "loss": 0.2316,
+      "step": 14963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79819260708619e-06,
+      "loss": 0.2257,
+      "step": 14964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.797006120017134e-06,
+      "loss": 0.2037,
+      "step": 14965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79581965013168e-06,
+      "loss": 0.235,
+      "step": 14966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.2169,
+      "step": 14967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.793446761979359e-06,
+      "loss": 0.2302,
+      "step": 14968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792260343746386e-06,
+      "loss": 0.2309,
+      "step": 14969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791073942764806e-06,
+      "loss": 0.2353,
+      "step": 14970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789887559051559e-06,
+      "loss": 0.2295,
+      "step": 14971
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.788701192623594e-06,
+      "loss": 0.2091,
+      "step": 14972
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787514843497855e-06,
+      "loss": 0.2214,
+      "step": 14973
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.786328511691291e-06,
+      "loss": 0.2098,
+      "step": 14974
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.785142197220849e-06,
+      "loss": 0.2458,
+      "step": 14975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.783955900103469e-06,
+      "loss": 0.2072,
+      "step": 14976
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782769620356102e-06,
+      "loss": 0.2203,
+      "step": 14977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.781583357995686e-06,
+      "loss": 0.2532,
+      "step": 14978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780397113039171e-06,
+      "loss": 0.2123,
+      "step": 14979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.779210885503498e-06,
+      "loss": 0.2228,
+      "step": 14980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778024675405614e-06,
+      "loss": 0.2211,
+      "step": 14981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776838482762462e-06,
+      "loss": 0.2337,
+      "step": 14982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775652307590982e-06,
+      "loss": 0.2391,
+      "step": 14983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.774466149908122e-06,
+      "loss": 0.2325,
+      "step": 14984
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.773280009730824e-06,
+      "loss": 0.198,
+      "step": 14985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772093887076033e-06,
+      "loss": 0.2286,
+      "step": 14986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.770907781960686e-06,
+      "loss": 0.2354,
+      "step": 14987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76972169440173e-06,
+      "loss": 0.2269,
+      "step": 14988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.768535624416104e-06,
+      "loss": 0.2201,
+      "step": 14989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.767349572020755e-06,
+      "loss": 0.2201,
+      "step": 14990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76616353723262e-06,
+      "loss": 0.2196,
+      "step": 14991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764977520068638e-06,
+      "loss": 0.2218,
+      "step": 14992
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.763791520545757e-06,
+      "loss": 0.2293,
+      "step": 14993
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.762605538680913e-06,
+      "loss": 0.2306,
+      "step": 14994
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761419574491048e-06,
+      "loss": 0.2227,
+      "step": 14995
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7602336279931e-06,
+      "loss": 0.2409,
+      "step": 14996
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759047699204013e-06,
+      "loss": 0.2519,
+      "step": 14997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.757861788140728e-06,
+      "loss": 0.2164,
+      "step": 14998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75667589482018e-06,
+      "loss": 0.2228,
+      "step": 14999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755490019259307e-06,
+      "loss": 0.2483,
+      "step": 15000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754304161475054e-06,
+      "loss": 0.2464,
+      "step": 15001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.753118321484359e-06,
+      "loss": 0.2414,
+      "step": 15002
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751932499304158e-06,
+      "loss": 0.2243,
+      "step": 15003
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75074669495139e-06,
+      "loss": 0.2175,
+      "step": 15004
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74956090844299e-06,
+      "loss": 0.2154,
+      "step": 15005
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.748375139795902e-06,
+      "loss": 0.2293,
+      "step": 15006
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747189389027057e-06,
+      "loss": 0.2183,
+      "step": 15007
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.746003656153397e-06,
+      "loss": 0.2383,
+      "step": 15008
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744817941191862e-06,
+      "loss": 0.2362,
+      "step": 15009
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74363224415938e-06,
+      "loss": 0.2327,
+      "step": 15010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.742446565072891e-06,
+      "loss": 0.2149,
+      "step": 15011
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.741260903949334e-06,
+      "loss": 0.2224,
+      "step": 15012
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740075260805644e-06,
+      "loss": 0.2064,
+      "step": 15013
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738889635658755e-06,
+      "loss": 0.2422,
+      "step": 15014
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.737704028525605e-06,
+      "loss": 0.2217,
+      "step": 15015
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736518439423127e-06,
+      "loss": 0.2312,
+      "step": 15016
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73533286836826e-06,
+      "loss": 0.2288,
+      "step": 15017
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734147315377932e-06,
+      "loss": 0.2514,
+      "step": 15018
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73296178046908e-06,
+      "loss": 0.21,
+      "step": 15019
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.731776263658645e-06,
+      "loss": 0.2249,
+      "step": 15020
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73059076496355e-06,
+      "loss": 0.2152,
+      "step": 15021
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.729405284400735e-06,
+      "loss": 0.2316,
+      "step": 15022
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728219821987131e-06,
+      "loss": 0.2233,
+      "step": 15023
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.727034377739673e-06,
+      "loss": 0.2618,
+      "step": 15024
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725848951675298e-06,
+      "loss": 0.2135,
+      "step": 15025
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.724663543810929e-06,
+      "loss": 0.2139,
+      "step": 15026
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723478154163505e-06,
+      "loss": 0.2464,
+      "step": 15027
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.722292782749956e-06,
+      "loss": 0.2275,
+      "step": 15028
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721107429587214e-06,
+      "loss": 0.2127,
+      "step": 15029
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719922094692217e-06,
+      "loss": 0.2309,
+      "step": 15030
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.718736778081887e-06,
+      "loss": 0.2169,
+      "step": 15031
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.717551479773158e-06,
+      "loss": 0.2384,
+      "step": 15032
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716366199782962e-06,
+      "loss": 0.2231,
+      "step": 15033
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.715180938128226e-06,
+      "loss": 0.2042,
+      "step": 15034
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713995694825887e-06,
+      "loss": 0.2348,
+      "step": 15035
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.712810469892869e-06,
+      "loss": 0.2278,
+      "step": 15036
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.711625263346106e-06,
+      "loss": 0.2404,
+      "step": 15037
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710440075202524e-06,
+      "loss": 0.2173,
+      "step": 15038
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.709254905479055e-06,
+      "loss": 0.2349,
+      "step": 15039
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.708069754192626e-06,
+      "loss": 0.21,
+      "step": 15040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.70688462136017e-06,
+      "loss": 0.2325,
+      "step": 15041
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.705699506998609e-06,
+      "loss": 0.1955,
+      "step": 15042
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.704514411124879e-06,
+      "loss": 0.228,
+      "step": 15043
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.703329333755899e-06,
+      "loss": 0.2299,
+      "step": 15044
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.7021442749086e-06,
+      "loss": 0.2075,
+      "step": 15045
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.700959234599913e-06,
+      "loss": 0.2362,
+      "step": 15046
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69977421284676e-06,
+      "loss": 0.2069,
+      "step": 15047
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698589209666074e-06,
+      "loss": 0.2493,
+      "step": 15048
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.697404225074776e-06,
+      "loss": 0.2391,
+      "step": 15049
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.696219259089793e-06,
+      "loss": 0.2071,
+      "step": 15050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.695034311728054e-06,
+      "loss": 0.2304,
+      "step": 15051
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.693849383006483e-06,
+      "loss": 0.2652,
+      "step": 15052
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692664472942007e-06,
+      "loss": 0.2111,
+      "step": 15053
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69147958155155e-06,
+      "loss": 0.2446,
+      "step": 15054
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.690294708852037e-06,
+      "loss": 0.2142,
+      "step": 15055
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.689109854860396e-06,
+      "loss": 0.1994,
+      "step": 15056
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687925019593544e-06,
+      "loss": 0.2139,
+      "step": 15057
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686740203068411e-06,
+      "loss": 0.237,
+      "step": 15058
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68555540530192e-06,
+      "loss": 0.2252,
+      "step": 15059
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684370626310993e-06,
+      "loss": 0.2424,
+      "step": 15060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.683185866112557e-06,
+      "loss": 0.2163,
+      "step": 15061
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68200112472353e-06,
+      "loss": 0.229,
+      "step": 15062
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680816402160839e-06,
+      "loss": 0.2302,
+      "step": 15063
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.679631698441407e-06,
+      "loss": 0.2152,
+      "step": 15064
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.678447013582155e-06,
+      "loss": 0.221,
+      "step": 15065
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677262347600002e-06,
+      "loss": 0.2244,
+      "step": 15066
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676077700511874e-06,
+      "loss": 0.2136,
+      "step": 15067
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674893072334689e-06,
+      "loss": 0.2236,
+      "step": 15068
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.673708463085377e-06,
+      "loss": 0.2043,
+      "step": 15069
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672523872780846e-06,
+      "loss": 0.2044,
+      "step": 15070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.671339301438027e-06,
+      "loss": 0.1986,
+      "step": 15071
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670154749073836e-06,
+      "loss": 0.2394,
+      "step": 15072
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.668970215705191e-06,
+      "loss": 0.2347,
+      "step": 15073
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667785701349016e-06,
+      "loss": 0.2042,
+      "step": 15074
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66660120602223e-06,
+      "loss": 0.222,
+      "step": 15075
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.665416729741751e-06,
+      "loss": 0.2287,
+      "step": 15076
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.664232272524497e-06,
+      "loss": 0.2353,
+      "step": 15077
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.663047834387391e-06,
+      "loss": 0.2285,
+      "step": 15078
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661863415347349e-06,
+      "loss": 0.2155,
+      "step": 15079
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66067901542129e-06,
+      "loss": 0.2467,
+      "step": 15080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65949463462613e-06,
+      "loss": 0.2248,
+      "step": 15081
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.658310272978792e-06,
+      "loss": 0.2212,
+      "step": 15082
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657125930496186e-06,
+      "loss": 0.2053,
+      "step": 15083
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.655941607195231e-06,
+      "loss": 0.2097,
+      "step": 15084
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.654757303092849e-06,
+      "loss": 0.2413,
+      "step": 15085
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65357301820595e-06,
+      "loss": 0.2258,
+      "step": 15086
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.652388752551458e-06,
+      "loss": 0.2182,
+      "step": 15087
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.651204506146283e-06,
+      "loss": 0.2348,
+      "step": 15088
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65002027900734e-06,
+      "loss": 0.2265,
+      "step": 15089
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.64883607115155e-06,
+      "loss": 0.2173,
+      "step": 15090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.647651882595825e-06,
+      "loss": 0.2375,
+      "step": 15091
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.64646771335708e-06,
+      "loss": 0.2321,
+      "step": 15092
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645283563452229e-06,
+      "loss": 0.2007,
+      "step": 15093
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644099432898188e-06,
+      "loss": 0.2384,
+      "step": 15094
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642915321711871e-06,
+      "loss": 0.2347,
+      "step": 15095
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.641731229910196e-06,
+      "loss": 0.2501,
+      "step": 15096
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.640547157510066e-06,
+      "loss": 0.2071,
+      "step": 15097
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639363104528405e-06,
+      "loss": 0.2353,
+      "step": 15098
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.638179070982117e-06,
+      "loss": 0.2437,
+      "step": 15099
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636995056888122e-06,
+      "loss": 0.2065,
+      "step": 15100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.635811062263328e-06,
+      "loss": 0.218,
+      "step": 15101
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63462708712465e-06,
+      "loss": 0.213,
+      "step": 15102
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633443131489e-06,
+      "loss": 0.2342,
+      "step": 15103
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632259195373287e-06,
+      "loss": 0.2239,
+      "step": 15104
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631075278794426e-06,
+      "loss": 0.2456,
+      "step": 15105
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.629891381769322e-06,
+      "loss": 0.2179,
+      "step": 15106
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.628707504314896e-06,
+      "loss": 0.2139,
+      "step": 15107
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627523646448047e-06,
+      "loss": 0.2355,
+      "step": 15108
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.626339808185696e-06,
+      "loss": 0.2327,
+      "step": 15109
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625155989544745e-06,
+      "loss": 0.2178,
+      "step": 15110
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.623972190542105e-06,
+      "loss": 0.2387,
+      "step": 15111
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.622788411194686e-06,
+      "loss": 0.2268,
+      "step": 15112
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621604651519398e-06,
+      "loss": 0.2394,
+      "step": 15113
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620420911533154e-06,
+      "loss": 0.2442,
+      "step": 15114
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619237191252855e-06,
+      "loss": 0.2503,
+      "step": 15115
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.618053490695413e-06,
+      "loss": 0.2116,
+      "step": 15116
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.616869809877734e-06,
+      "loss": 0.2274,
+      "step": 15117
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61568614881673e-06,
+      "loss": 0.2249,
+      "step": 15118
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.614502507529305e-06,
+      "loss": 0.2115,
+      "step": 15119
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613318886032367e-06,
+      "loss": 0.2136,
+      "step": 15120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.612135284342824e-06,
+      "loss": 0.2195,
+      "step": 15121
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.610951702477583e-06,
+      "loss": 0.2178,
+      "step": 15122
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609768140453546e-06,
+      "loss": 0.2264,
+      "step": 15123
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.608584598287621e-06,
+      "loss": 0.2152,
+      "step": 15124
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607401075996717e-06,
+      "loss": 0.2156,
+      "step": 15125
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.606217573597738e-06,
+      "loss": 0.2331,
+      "step": 15126
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.605034091107585e-06,
+      "loss": 0.2441,
+      "step": 15127
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.603850628543167e-06,
+      "loss": 0.2334,
+      "step": 15128
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.602667185921387e-06,
+      "loss": 0.21,
+      "step": 15129
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601483763259153e-06,
+      "loss": 0.2246,
+      "step": 15130
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.600300360573363e-06,
+      "loss": 0.2304,
+      "step": 15131
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599116977880924e-06,
+      "loss": 0.2429,
+      "step": 15132
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.597933615198742e-06,
+      "loss": 0.2375,
+      "step": 15133
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.596750272543715e-06,
+      "loss": 0.1937,
+      "step": 15134
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.595566949932752e-06,
+      "loss": 0.2276,
+      "step": 15135
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59438364738275e-06,
+      "loss": 0.2151,
+      "step": 15136
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.593200364910616e-06,
+      "loss": 0.2403,
+      "step": 15137
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.592017102533245e-06,
+      "loss": 0.2058,
+      "step": 15138
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590833860267545e-06,
+      "loss": 0.2384,
+      "step": 15139
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.589650638130414e-06,
+      "loss": 0.2171,
+      "step": 15140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588467436138756e-06,
+      "loss": 0.2203,
+      "step": 15141
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.587284254309473e-06,
+      "loss": 0.2375,
+      "step": 15142
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586101092659462e-06,
+      "loss": 0.2258,
+      "step": 15143
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584917951205623e-06,
+      "loss": 0.236,
+      "step": 15144
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583734829964857e-06,
+      "loss": 0.1918,
+      "step": 15145
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58255172895407e-06,
+      "loss": 0.2305,
+      "step": 15146
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581368648190152e-06,
+      "loss": 0.2306,
+      "step": 15147
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580185587690009e-06,
+      "loss": 0.2478,
+      "step": 15148
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.579002547470534e-06,
+      "loss": 0.1989,
+      "step": 15149
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577819527548628e-06,
+      "loss": 0.2466,
+      "step": 15150
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.57663652794119e-06,
+      "loss": 0.228,
+      "step": 15151
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575453548665118e-06,
+      "loss": 0.2322,
+      "step": 15152
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574270589737311e-06,
+      "loss": 0.2336,
+      "step": 15153
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573087651174663e-06,
+      "loss": 0.2285,
+      "step": 15154
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571904732994073e-06,
+      "loss": 0.2316,
+      "step": 15155
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.570721835212438e-06,
+      "loss": 0.2296,
+      "step": 15156
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569538957846655e-06,
+      "loss": 0.2201,
+      "step": 15157
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56835610091362e-06,
+      "loss": 0.2289,
+      "step": 15158
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567173264430228e-06,
+      "loss": 0.2091,
+      "step": 15159
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565990448413376e-06,
+      "loss": 0.2182,
+      "step": 15160
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564807652879962e-06,
+      "loss": 0.2284,
+      "step": 15161
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563624877846874e-06,
+      "loss": 0.2446,
+      "step": 15162
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56244212333101e-06,
+      "loss": 0.2428,
+      "step": 15163
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.561259389349269e-06,
+      "loss": 0.2377,
+      "step": 15164
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.560076675918537e-06,
+      "loss": 0.2265,
+      "step": 15165
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.558893983055711e-06,
+      "loss": 0.2365,
+      "step": 15166
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.557711310777689e-06,
+      "loss": 0.2183,
+      "step": 15167
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55652865910136e-06,
+      "loss": 0.2388,
+      "step": 15168
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55534602804362e-06,
+      "loss": 0.2529,
+      "step": 15169
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.554163417621357e-06,
+      "loss": 0.2184,
+      "step": 15170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.552980827851467e-06,
+      "loss": 0.2196,
+      "step": 15171
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.551798258750843e-06,
+      "loss": 0.2243,
+      "step": 15172
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550615710336373e-06,
+      "loss": 0.2246,
+      "step": 15173
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549433182624955e-06,
+      "loss": 0.2192,
+      "step": 15174
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.548250675633476e-06,
+      "loss": 0.2194,
+      "step": 15175
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547068189378826e-06,
+      "loss": 0.2084,
+      "step": 15176
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.545885723877895e-06,
+      "loss": 0.2316,
+      "step": 15177
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.544703279147576e-06,
+      "loss": 0.198,
+      "step": 15178
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.543520855204757e-06,
+      "loss": 0.2239,
+      "step": 15179
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542338452066333e-06,
+      "loss": 0.2244,
+      "step": 15180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.541156069749186e-06,
+      "loss": 0.2173,
+      "step": 15181
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.539973708270208e-06,
+      "loss": 0.2237,
+      "step": 15182
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53879136764629e-06,
+      "loss": 0.2603,
+      "step": 15183
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.537609047894318e-06,
+      "loss": 0.2377,
+      "step": 15184
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536426749031184e-06,
+      "loss": 0.2241,
+      "step": 15185
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535244471073773e-06,
+      "loss": 0.2113,
+      "step": 15186
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534062214038973e-06,
+      "loss": 0.2241,
+      "step": 15187
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532879977943673e-06,
+      "loss": 0.231,
+      "step": 15188
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.531697762804755e-06,
+      "loss": 0.2183,
+      "step": 15189
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.530515568639111e-06,
+      "loss": 0.2282,
+      "step": 15190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.529333395463625e-06,
+      "loss": 0.22,
+      "step": 15191
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528151243295188e-06,
+      "loss": 0.2307,
+      "step": 15192
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526969112150677e-06,
+      "loss": 0.2474,
+      "step": 15193
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525787002046984e-06,
+      "loss": 0.2211,
+      "step": 15194
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.524604913000994e-06,
+      "loss": 0.2296,
+      "step": 15195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.523422845029591e-06,
+      "loss": 0.2296,
+      "step": 15196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522240798149658e-06,
+      "loss": 0.2395,
+      "step": 15197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521058772378083e-06,
+      "loss": 0.2221,
+      "step": 15198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.519876767731746e-06,
+      "loss": 0.2241,
+      "step": 15199
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.518694784227533e-06,
+      "loss": 0.2184,
+      "step": 15200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517512821882334e-06,
+      "loss": 0.2248,
+      "step": 15201
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.516330880713019e-06,
+      "loss": 0.2259,
+      "step": 15202
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515148960736479e-06,
+      "loss": 0.204,
+      "step": 15203
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.513967061969594e-06,
+      "loss": 0.2587,
+      "step": 15204
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.512785184429247e-06,
+      "loss": 0.2279,
+      "step": 15205
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51160332813232e-06,
+      "loss": 0.2285,
+      "step": 15206
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.510421493095697e-06,
+      "loss": 0.2376,
+      "step": 15207
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509239679336256e-06,
+      "loss": 0.2332,
+      "step": 15208
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50805788687088e-06,
+      "loss": 0.2208,
+      "step": 15209
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.506876115716448e-06,
+      "loss": 0.227,
+      "step": 15210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.505694365889842e-06,
+      "loss": 0.2196,
+      "step": 15211
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504512637407943e-06,
+      "loss": 0.2144,
+      "step": 15212
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503330930287628e-06,
+      "loss": 0.2591,
+      "step": 15213
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.502149244545783e-06,
+      "loss": 0.2494,
+      "step": 15214
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.500967580199278e-06,
+      "loss": 0.2371,
+      "step": 15215
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.499785937264996e-06,
+      "loss": 0.2273,
+      "step": 15216
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498604315759815e-06,
+      "loss": 0.2582,
+      "step": 15217
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497422715700614e-06,
+      "loss": 0.2061,
+      "step": 15218
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.496241137104275e-06,
+      "loss": 0.2361,
+      "step": 15219
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49505957998767e-06,
+      "loss": 0.239,
+      "step": 15220
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.493878044367675e-06,
+      "loss": 0.2126,
+      "step": 15221
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.492696530261173e-06,
+      "loss": 0.2297,
+      "step": 15222
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.491515037685039e-06,
+      "loss": 0.2197,
+      "step": 15223
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.490333566656149e-06,
+      "loss": 0.2096,
+      "step": 15224
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.489152117191378e-06,
+      "loss": 0.2349,
+      "step": 15225
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.487970689307601e-06,
+      "loss": 0.221,
+      "step": 15226
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486789283021702e-06,
+      "loss": 0.2014,
+      "step": 15227
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.485607898350543e-06,
+      "loss": 0.2393,
+      "step": 15228
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484426535311007e-06,
+      "loss": 0.2469,
+      "step": 15229
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.483245193919965e-06,
+      "loss": 0.233,
+      "step": 15230
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482063874194297e-06,
+      "loss": 0.2357,
+      "step": 15231
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480882576150872e-06,
+      "loss": 0.2506,
+      "step": 15232
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479701299806563e-06,
+      "loss": 0.2003,
+      "step": 15233
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.478520045178247e-06,
+      "loss": 0.2145,
+      "step": 15234
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477338812282797e-06,
+      "loss": 0.2378,
+      "step": 15235
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476157601137083e-06,
+      "loss": 0.2452,
+      "step": 15236
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474976411757978e-06,
+      "loss": 0.2097,
+      "step": 15237
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.473795244162354e-06,
+      "loss": 0.2378,
+      "step": 15238
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472614098367086e-06,
+      "loss": 0.2331,
+      "step": 15239
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.471432974389043e-06,
+      "loss": 0.2129,
+      "step": 15240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.470251872245098e-06,
+      "loss": 0.2094,
+      "step": 15241
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46907079195212e-06,
+      "loss": 0.2232,
+      "step": 15242
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467889733526977e-06,
+      "loss": 0.2378,
+      "step": 15243
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466708696986543e-06,
+      "loss": 0.2045,
+      "step": 15244
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465527682347687e-06,
+      "loss": 0.215,
+      "step": 15245
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46434668962728e-06,
+      "loss": 0.2092,
+      "step": 15246
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46316571884219e-06,
+      "loss": 0.2523,
+      "step": 15247
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461984770009286e-06,
+      "loss": 0.2316,
+      "step": 15248
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.460803843145436e-06,
+      "loss": 0.2089,
+      "step": 15249
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45962293826751e-06,
+      "loss": 0.2336,
+      "step": 15250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.458442055392377e-06,
+      "loss": 0.2425,
+      "step": 15251
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.4572611945369e-06,
+      "loss": 0.2254,
+      "step": 15252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456080355717951e-06,
+      "loss": 0.2278,
+      "step": 15253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.454899538952399e-06,
+      "loss": 0.2318,
+      "step": 15254
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453718744257103e-06,
+      "loss": 0.2133,
+      "step": 15255
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452537971648935e-06,
+      "loss": 0.2309,
+      "step": 15256
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45135722114476e-06,
+      "loss": 0.2328,
+      "step": 15257
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450176492761446e-06,
+      "loss": 0.2285,
+      "step": 15258
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.448995786515855e-06,
+      "loss": 0.2247,
+      "step": 15259
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.447815102424854e-06,
+      "loss": 0.2088,
+      "step": 15260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446634440505306e-06,
+      "loss": 0.2498,
+      "step": 15261
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44545380077408e-06,
+      "loss": 0.2155,
+      "step": 15262
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444273183248039e-06,
+      "loss": 0.2405,
+      "step": 15263
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.443092587944043e-06,
+      "loss": 0.2134,
+      "step": 15264
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.441912014878958e-06,
+      "loss": 0.2305,
+      "step": 15265
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44073146406965e-06,
+      "loss": 0.2368,
+      "step": 15266
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43955093553298e-06,
+      "loss": 0.2305,
+      "step": 15267
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43837042928581e-06,
+      "loss": 0.2226,
+      "step": 15268
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.437189945345003e-06,
+      "loss": 0.2315,
+      "step": 15269
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43600948372742e-06,
+      "loss": 0.2375,
+      "step": 15270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.434829044449921e-06,
+      "loss": 0.1998,
+      "step": 15271
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.433648627529373e-06,
+      "loss": 0.2262,
+      "step": 15272
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432468232982633e-06,
+      "loss": 0.2362,
+      "step": 15273
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431287860826565e-06,
+      "loss": 0.2139,
+      "step": 15274
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.430107511078027e-06,
+      "loss": 0.2134,
+      "step": 15275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428927183753876e-06,
+      "loss": 0.2339,
+      "step": 15276
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.427746878870978e-06,
+      "loss": 0.2064,
+      "step": 15277
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42656659644619e-06,
+      "loss": 0.227,
+      "step": 15278
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425386336496373e-06,
+      "loss": 0.2092,
+      "step": 15279
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424206099038383e-06,
+      "loss": 0.2048,
+      "step": 15280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42302588408908e-06,
+      "loss": 0.2486,
+      "step": 15281
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42184569166532e-06,
+      "loss": 0.2227,
+      "step": 15282
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420665521783962e-06,
+      "loss": 0.2309,
+      "step": 15283
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.419485374461865e-06,
+      "loss": 0.2243,
+      "step": 15284
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418305249715888e-06,
+      "loss": 0.2313,
+      "step": 15285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417125147562883e-06,
+      "loss": 0.2236,
+      "step": 15286
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.415945068019709e-06,
+      "loss": 0.2315,
+      "step": 15287
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.414765011103222e-06,
+      "loss": 0.2276,
+      "step": 15288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.413584976830279e-06,
+      "loss": 0.2143,
+      "step": 15289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.412404965217735e-06,
+      "loss": 0.2442,
+      "step": 15290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411224976282446e-06,
+      "loss": 0.2416,
+      "step": 15291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.410045010041263e-06,
+      "loss": 0.2204,
+      "step": 15292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40886506651105e-06,
+      "loss": 0.2203,
+      "step": 15293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.407685145708652e-06,
+      "loss": 0.2319,
+      "step": 15294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406505247650926e-06,
+      "loss": 0.2307,
+      "step": 15295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.405325372354724e-06,
+      "loss": 0.2327,
+      "step": 15296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.404145519836906e-06,
+      "loss": 0.2222,
+      "step": 15297
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402965690114317e-06,
+      "loss": 0.212,
+      "step": 15298
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401785883203814e-06,
+      "loss": 0.2208,
+      "step": 15299
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.400606099122247e-06,
+      "loss": 0.2496,
+      "step": 15300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.399426337886473e-06,
+      "loss": 0.215,
+      "step": 15301
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.398246599513338e-06,
+      "loss": 0.213,
+      "step": 15302
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397066884019696e-06,
+      "loss": 0.2528,
+      "step": 15303
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.395887191422397e-06,
+      "loss": 0.2068,
+      "step": 15304
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.394707521738294e-06,
+      "loss": 0.2063,
+      "step": 15305
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393527874984239e-06,
+      "loss": 0.2332,
+      "step": 15306
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.392348251177076e-06,
+      "loss": 0.234,
+      "step": 15307
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39116865033366e-06,
+      "loss": 0.2405,
+      "step": 15308
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.389989072470835e-06,
+      "loss": 0.2102,
+      "step": 15309
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.388809517605456e-06,
+      "loss": 0.2398,
+      "step": 15310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387629985754366e-06,
+      "loss": 0.2099,
+      "step": 15311
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.386450476934418e-06,
+      "loss": 0.2484,
+      "step": 15312
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.385270991162463e-06,
+      "loss": 0.2146,
+      "step": 15313
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.384091528455342e-06,
+      "loss": 0.2081,
+      "step": 15314
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.382912088829904e-06,
+      "loss": 0.2357,
+      "step": 15315
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.381732672302999e-06,
+      "loss": 0.2436,
+      "step": 15316
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380553278891472e-06,
+      "loss": 0.216,
+      "step": 15317
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.379373908612169e-06,
+      "loss": 0.2101,
+      "step": 15318
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378194561481938e-06,
+      "loss": 0.2439,
+      "step": 15319
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377015237517626e-06,
+      "loss": 0.2131,
+      "step": 15320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375835936736072e-06,
+      "loss": 0.2315,
+      "step": 15321
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.374656659154127e-06,
+      "loss": 0.2366,
+      "step": 15322
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.373477404788632e-06,
+      "loss": 0.1933,
+      "step": 15323
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.372298173656437e-06,
+      "loss": 0.2258,
+      "step": 15324
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37111896577438e-06,
+      "loss": 0.2334,
+      "step": 15325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36993978115931e-06,
+      "loss": 0.2307,
+      "step": 15326
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.368760619828068e-06,
+      "loss": 0.2155,
+      "step": 15327
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367581481797494e-06,
+      "loss": 0.2511,
+      "step": 15328
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366402367084439e-06,
+      "loss": 0.2453,
+      "step": 15329
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.365223275705738e-06,
+      "loss": 0.2486,
+      "step": 15330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.364044207678235e-06,
+      "loss": 0.2036,
+      "step": 15331
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.362865163018774e-06,
+      "loss": 0.2236,
+      "step": 15332
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3616861417442e-06,
+      "loss": 0.2212,
+      "step": 15333
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.360507143871344e-06,
+      "loss": 0.2316,
+      "step": 15334
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359328169417052e-06,
+      "loss": 0.2456,
+      "step": 15335
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.358149218398168e-06,
+      "loss": 0.2267,
+      "step": 15336
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356970290831528e-06,
+      "loss": 0.2164,
+      "step": 15337
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35579138673397e-06,
+      "loss": 0.2349,
+      "step": 15338
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.354612506122338e-06,
+      "loss": 0.245,
+      "step": 15339
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.353433649013472e-06,
+      "loss": 0.2205,
+      "step": 15340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.352254815424204e-06,
+      "loss": 0.2098,
+      "step": 15341
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.351076005371378e-06,
+      "loss": 0.2349,
+      "step": 15342
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349897218871833e-06,
+      "loss": 0.2373,
+      "step": 15343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348718455942402e-06,
+      "loss": 0.229,
+      "step": 15344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347539716599927e-06,
+      "loss": 0.2366,
+      "step": 15345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346361000861245e-06,
+      "loss": 0.2219,
+      "step": 15346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345182308743191e-06,
+      "loss": 0.2288,
+      "step": 15347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3440036402626e-06,
+      "loss": 0.2227,
+      "step": 15348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342824995436308e-06,
+      "loss": 0.2571,
+      "step": 15349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341646374281155e-06,
+      "loss": 0.2167,
+      "step": 15350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.340467776813972e-06,
+      "loss": 0.2191,
+      "step": 15351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339289203051599e-06,
+      "loss": 0.2191,
+      "step": 15352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.338110653010865e-06,
+      "loss": 0.2198,
+      "step": 15353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33693212670861e-06,
+      "loss": 0.2422,
+      "step": 15354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.335753624161664e-06,
+      "loss": 0.2121,
+      "step": 15355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.334575145386864e-06,
+      "loss": 0.2626,
+      "step": 15356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33339669040104e-06,
+      "loss": 0.2066,
+      "step": 15357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.332218259221027e-06,
+      "loss": 0.2312,
+      "step": 15358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33103985186366e-06,
+      "loss": 0.2102,
+      "step": 15359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.329861468345768e-06,
+      "loss": 0.2323,
+      "step": 15360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328683108684182e-06,
+      "loss": 0.2312,
+      "step": 15361
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.327504772895735e-06,
+      "loss": 0.2173,
+      "step": 15362
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326326460997263e-06,
+      "loss": 0.2455,
+      "step": 15363
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32514817300559e-06,
+      "loss": 0.2224,
+      "step": 15364
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323969908937551e-06,
+      "loss": 0.1983,
+      "step": 15365
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.322791668809974e-06,
+      "loss": 0.219,
+      "step": 15366
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32161345263969e-06,
+      "loss": 0.2065,
+      "step": 15367
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.320435260443532e-06,
+      "loss": 0.2255,
+      "step": 15368
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.319257092238324e-06,
+      "loss": 0.221,
+      "step": 15369
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318078948040897e-06,
+      "loss": 0.2429,
+      "step": 15370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31690082786808e-06,
+      "loss": 0.213,
+      "step": 15371
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315722731736705e-06,
+      "loss": 0.2158,
+      "step": 15372
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.314544659663593e-06,
+      "loss": 0.2225,
+      "step": 15373
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.313366611665576e-06,
+      "loss": 0.2362,
+      "step": 15374
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312188587759478e-06,
+      "loss": 0.2387,
+      "step": 15375
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311010587962128e-06,
+      "loss": 0.2117,
+      "step": 15376
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.309832612290353e-06,
+      "loss": 0.2287,
+      "step": 15377
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308654660760978e-06,
+      "loss": 0.2217,
+      "step": 15378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307476733390832e-06,
+      "loss": 0.243,
+      "step": 15379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306298830196736e-06,
+      "loss": 0.2316,
+      "step": 15380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305120951195517e-06,
+      "loss": 0.2232,
+      "step": 15381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.303943096404e-06,
+      "loss": 0.2323,
+      "step": 15382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302765265839012e-06,
+      "loss": 0.2331,
+      "step": 15383
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301587459517376e-06,
+      "loss": 0.2215,
+      "step": 15384
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.300409677455915e-06,
+      "loss": 0.2434,
+      "step": 15385
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299231919671452e-06,
+      "loss": 0.2197,
+      "step": 15386
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.298054186180808e-06,
+      "loss": 0.2308,
+      "step": 15387
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.29687647700081e-06,
+      "loss": 0.2325,
+      "step": 15388
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295698792148276e-06,
+      "loss": 0.2136,
+      "step": 15389
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294521131640034e-06,
+      "loss": 0.2341,
+      "step": 15390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.2933434954929e-06,
+      "loss": 0.2344,
+      "step": 15391
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292165883723698e-06,
+      "loss": 0.2496,
+      "step": 15392
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.290988296349248e-06,
+      "loss": 0.2194,
+      "step": 15393
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289810733386373e-06,
+      "loss": 0.2356,
+      "step": 15394
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.288633194851894e-06,
+      "loss": 0.2557,
+      "step": 15395
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287455680762626e-06,
+      "loss": 0.228,
+      "step": 15396
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.28627819113539e-06,
+      "loss": 0.2295,
+      "step": 15397
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285100725987009e-06,
+      "loss": 0.2012,
+      "step": 15398
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.283923285334304e-06,
+      "loss": 0.2339,
+      "step": 15399
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.282745869194085e-06,
+      "loss": 0.2372,
+      "step": 15400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.281568477583174e-06,
+      "loss": 0.2479,
+      "step": 15401
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.280391110518392e-06,
+      "loss": 0.2215,
+      "step": 15402
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.279213768016551e-06,
+      "loss": 0.2483,
+      "step": 15403
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278036450094473e-06,
+      "loss": 0.242,
+      "step": 15404
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276859156768973e-06,
+      "loss": 0.2319,
+      "step": 15405
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.275681888056868e-06,
+      "loss": 0.2133,
+      "step": 15406
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.274504643974973e-06,
+      "loss": 0.2228,
+      "step": 15407
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.273327424540104e-06,
+      "loss": 0.251,
+      "step": 15408
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272150229769076e-06,
+      "loss": 0.2496,
+      "step": 15409
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270973059678708e-06,
+      "loss": 0.2288,
+      "step": 15410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.269795914285811e-06,
+      "loss": 0.2208,
+      "step": 15411
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.268618793607201e-06,
+      "loss": 0.2336,
+      "step": 15412
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.267441697659692e-06,
+      "loss": 0.2483,
+      "step": 15413
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266264626460095e-06,
+      "loss": 0.2222,
+      "step": 15414
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.265087580025224e-06,
+      "loss": 0.2272,
+      "step": 15415
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263910558371893e-06,
+      "loss": 0.2345,
+      "step": 15416
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.262733561516916e-06,
+      "loss": 0.2314,
+      "step": 15417
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.261556589477105e-06,
+      "loss": 0.1967,
+      "step": 15418
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260379642269269e-06,
+      "loss": 0.2032,
+      "step": 15419
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.259202719910221e-06,
+      "loss": 0.2343,
+      "step": 15420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258025822416773e-06,
+      "loss": 0.2324,
+      "step": 15421
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256848949805737e-06,
+      "loss": 0.2186,
+      "step": 15422
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.255672102093923e-06,
+      "loss": 0.2389,
+      "step": 15423
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.254495279298137e-06,
+      "loss": 0.2295,
+      "step": 15424
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.253318481435198e-06,
+      "loss": 0.2312,
+      "step": 15425
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252141708521904e-06,
+      "loss": 0.2329,
+      "step": 15426
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.250964960575069e-06,
+      "loss": 0.2399,
+      "step": 15427
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.249788237611503e-06,
+      "loss": 0.2439,
+      "step": 15428
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248611539648015e-06,
+      "loss": 0.2267,
+      "step": 15429
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247434866701408e-06,
+      "loss": 0.2187,
+      "step": 15430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246258218788494e-06,
+      "loss": 0.231,
+      "step": 15431
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24508159592608e-06,
+      "loss": 0.2343,
+      "step": 15432
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243904998130972e-06,
+      "loss": 0.216,
+      "step": 15433
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.242728425419978e-06,
+      "loss": 0.2172,
+      "step": 15434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.241551877809903e-06,
+      "loss": 0.2188,
+      "step": 15435
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24037535531755e-06,
+      "loss": 0.2408,
+      "step": 15436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23919885795973e-06,
+      "loss": 0.2246,
+      "step": 15437
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.238022385753248e-06,
+      "loss": 0.2319,
+      "step": 15438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.236845938714901e-06,
+      "loss": 0.2359,
+      "step": 15439
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.235669516861502e-06,
+      "loss": 0.2271,
+      "step": 15440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23449312020985e-06,
+      "loss": 0.2667,
+      "step": 15441
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23331674877675e-06,
+      "loss": 0.2477,
+      "step": 15442
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.232140402579005e-06,
+      "loss": 0.2334,
+      "step": 15443
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23096408163342e-06,
+      "loss": 0.2278,
+      "step": 15444
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.229787785956798e-06,
+      "loss": 0.215,
+      "step": 15445
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.228611515565936e-06,
+      "loss": 0.2104,
+      "step": 15446
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22743527047764e-06,
+      "loss": 0.2535,
+      "step": 15447
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22625905070871e-06,
+      "loss": 0.2333,
+      "step": 15448
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22508285627595e-06,
+      "loss": 0.1851,
+      "step": 15449
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.223906687196159e-06,
+      "loss": 0.237,
+      "step": 15450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222730543486138e-06,
+      "loss": 0.2283,
+      "step": 15451
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221554425162687e-06,
+      "loss": 0.2175,
+      "step": 15452
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.220378332242602e-06,
+      "loss": 0.2253,
+      "step": 15453
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219202264742685e-06,
+      "loss": 0.233,
+      "step": 15454
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.218026222679735e-06,
+      "loss": 0.2224,
+      "step": 15455
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.216850206070552e-06,
+      "loss": 0.2374,
+      "step": 15456
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.215674214931936e-06,
+      "loss": 0.2323,
+      "step": 15457
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214498249280679e-06,
+      "loss": 0.2105,
+      "step": 15458
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.213322309133581e-06,
+      "loss": 0.2398,
+      "step": 15459
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.212146394507441e-06,
+      "loss": 0.2286,
+      "step": 15460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210970505419057e-06,
+      "loss": 0.2121,
+      "step": 15461
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20979464188522e-06,
+      "loss": 0.2126,
+      "step": 15462
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20861880392273e-06,
+      "loss": 0.2319,
+      "step": 15463
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.207442991548386e-06,
+      "loss": 0.2288,
+      "step": 15464
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206267204778974e-06,
+      "loss": 0.2087,
+      "step": 15465
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.205091443631294e-06,
+      "loss": 0.2445,
+      "step": 15466
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.203915708122141e-06,
+      "loss": 0.2101,
+      "step": 15467
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.202739998268312e-06,
+      "loss": 0.2192,
+      "step": 15468
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.201564314086594e-06,
+      "loss": 0.2276,
+      "step": 15469
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200388655593787e-06,
+      "loss": 0.2545,
+      "step": 15470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.199213022806677e-06,
+      "loss": 0.2315,
+      "step": 15471
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198037415742066e-06,
+      "loss": 0.2168,
+      "step": 15472
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.196861834416742e-06,
+      "loss": 0.2434,
+      "step": 15473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195686278847495e-06,
+      "loss": 0.2152,
+      "step": 15474
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.194510749051117e-06,
+      "loss": 0.2291,
+      "step": 15475
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.193335245044401e-06,
+      "loss": 0.2113,
+      "step": 15476
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19215976684414e-06,
+      "loss": 0.2558,
+      "step": 15477
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.190984314467122e-06,
+      "loss": 0.2451,
+      "step": 15478
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18980888793014e-06,
+      "loss": 0.2132,
+      "step": 15479
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188633487249975e-06,
+      "loss": 0.2175,
+      "step": 15480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187458112443425e-06,
+      "loss": 0.1949,
+      "step": 15481
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.186282763527276e-06,
+      "loss": 0.2195,
+      "step": 15482
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185107440518319e-06,
+      "loss": 0.2074,
+      "step": 15483
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.183932143433341e-06,
+      "loss": 0.2235,
+      "step": 15484
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.182756872289129e-06,
+      "loss": 0.2135,
+      "step": 15485
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18158162710247e-06,
+      "loss": 0.2493,
+      "step": 15486
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.180406407890154e-06,
+      "loss": 0.2324,
+      "step": 15487
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.179231214668965e-06,
+      "loss": 0.2211,
+      "step": 15488
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178056047455695e-06,
+      "loss": 0.2118,
+      "step": 15489
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.176880906267123e-06,
+      "loss": 0.2193,
+      "step": 15490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.175705791120042e-06,
+      "loss": 0.2592,
+      "step": 15491
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174530702031228e-06,
+      "loss": 0.2212,
+      "step": 15492
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.173355639017474e-06,
+      "loss": 0.2435,
+      "step": 15493
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.17218060209556e-06,
+      "loss": 0.2145,
+      "step": 15494
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.171005591282274e-06,
+      "loss": 0.2391,
+      "step": 15495
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169830606594398e-06,
+      "loss": 0.2229,
+      "step": 15496
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.168655648048713e-06,
+      "loss": 0.2463,
+      "step": 15497
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167480715662006e-06,
+      "loss": 0.2083,
+      "step": 15498
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.166305809451058e-06,
+      "loss": 0.2306,
+      "step": 15499
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.165130929432656e-06,
+      "loss": 0.2342,
+      "step": 15500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.163956075623574e-06,
+      "loss": 0.2231,
+      "step": 15501
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162781248040597e-06,
+      "loss": 0.233,
+      "step": 15502
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161606446700507e-06,
+      "loss": 0.1947,
+      "step": 15503
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16043167162009e-06,
+      "loss": 0.2151,
+      "step": 15504
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.159256922816117e-06,
+      "loss": 0.2199,
+      "step": 15505
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15808220030537e-06,
+      "loss": 0.1977,
+      "step": 15506
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.156907504104636e-06,
+      "loss": 0.2503,
+      "step": 15507
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155732834230686e-06,
+      "loss": 0.1907,
+      "step": 15508
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154558190700303e-06,
+      "loss": 0.2006,
+      "step": 15509
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.153383573530265e-06,
+      "loss": 0.226,
+      "step": 15510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152208982737353e-06,
+      "loss": 0.2238,
+      "step": 15511
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.151034418338341e-06,
+      "loss": 0.1998,
+      "step": 15512
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.149859880350007e-06,
+      "loss": 0.1984,
+      "step": 15513
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148685368789127e-06,
+      "loss": 0.2071,
+      "step": 15514
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147510883672482e-06,
+      "loss": 0.2456,
+      "step": 15515
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.146336425016849e-06,
+      "loss": 0.2144,
+      "step": 15516
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.145161992839e-06,
+      "loss": 0.2385,
+      "step": 15517
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143987587155712e-06,
+      "loss": 0.2306,
+      "step": 15518
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.142813207983758e-06,
+      "loss": 0.231,
+      "step": 15519
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.141638855339914e-06,
+      "loss": 0.2209,
+      "step": 15520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140464529240955e-06,
+      "loss": 0.2195,
+      "step": 15521
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139290229703658e-06,
+      "loss": 0.2108,
+      "step": 15522
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.138115956744793e-06,
+      "loss": 0.249,
+      "step": 15523
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.136941710381134e-06,
+      "loss": 0.2149,
+      "step": 15524
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135767490629456e-06,
+      "loss": 0.2385,
+      "step": 15525
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134593297506529e-06,
+      "loss": 0.223,
+      "step": 15526
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.133419131029125e-06,
+      "loss": 0.2343,
+      "step": 15527
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132244991214022e-06,
+      "loss": 0.2404,
+      "step": 15528
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.131070878077982e-06,
+      "loss": 0.2293,
+      "step": 15529
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.129896791637786e-06,
+      "loss": 0.2332,
+      "step": 15530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128722731910195e-06,
+      "loss": 0.2268,
+      "step": 15531
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.127548698911983e-06,
+      "loss": 0.2255,
+      "step": 15532
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.126374692659922e-06,
+      "loss": 0.2343,
+      "step": 15533
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12520071317078e-06,
+      "loss": 0.2314,
+      "step": 15534
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.124026760461326e-06,
+      "loss": 0.2265,
+      "step": 15535
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122852834548329e-06,
+      "loss": 0.2323,
+      "step": 15536
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.121678935448557e-06,
+      "loss": 0.2398,
+      "step": 15537
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120505063178778e-06,
+      "loss": 0.2336,
+      "step": 15538
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119331217755764e-06,
+      "loss": 0.258,
+      "step": 15539
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118157399196274e-06,
+      "loss": 0.2165,
+      "step": 15540
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11698360751708e-06,
+      "loss": 0.2133,
+      "step": 15541
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.115809842734949e-06,
+      "loss": 0.2544,
+      "step": 15542
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.114636104866648e-06,
+      "loss": 0.2351,
+      "step": 15543
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113462393928937e-06,
+      "loss": 0.2096,
+      "step": 15544
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112288709938585e-06,
+      "loss": 0.2279,
+      "step": 15545
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111115052912358e-06,
+      "loss": 0.2281,
+      "step": 15546
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109941422867018e-06,
+      "loss": 0.2368,
+      "step": 15547
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.108767819819331e-06,
+      "loss": 0.2561,
+      "step": 15548
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.107594243786059e-06,
+      "loss": 0.2168,
+      "step": 15549
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10642069478397e-06,
+      "loss": 0.2516,
+      "step": 15550
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.105247172829819e-06,
+      "loss": 0.2346,
+      "step": 15551
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.104073677940375e-06,
+      "loss": 0.2177,
+      "step": 15552
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.102900210132398e-06,
+      "loss": 0.2266,
+      "step": 15553
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10172676942265e-06,
+      "loss": 0.2299,
+      "step": 15554
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.100553355827897e-06,
+      "loss": 0.2139,
+      "step": 15555
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099379969364893e-06,
+      "loss": 0.2272,
+      "step": 15556
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.098206610050404e-06,
+      "loss": 0.242,
+      "step": 15557
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.097033277901186e-06,
+      "loss": 0.2232,
+      "step": 15558
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.095859972933998e-06,
+      "loss": 0.2173,
+      "step": 15559
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094686695165604e-06,
+      "loss": 0.2341,
+      "step": 15560
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.093513444612761e-06,
+      "loss": 0.2335,
+      "step": 15561
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09234022129223e-06,
+      "loss": 0.2207,
+      "step": 15562
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091167025220765e-06,
+      "loss": 0.2314,
+      "step": 15563
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089993856415128e-06,
+      "loss": 0.2788,
+      "step": 15564
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.088820714892075e-06,
+      "loss": 0.2068,
+      "step": 15565
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087647600668363e-06,
+      "loss": 0.2363,
+      "step": 15566
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08647451376075e-06,
+      "loss": 0.2287,
+      "step": 15567
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08530145418599e-06,
+      "loss": 0.2218,
+      "step": 15568
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08412842196084e-06,
+      "loss": 0.2029,
+      "step": 15569
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.082955417102061e-06,
+      "loss": 0.2279,
+      "step": 15570
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.081782439626398e-06,
+      "loss": 0.2359,
+      "step": 15571
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.080609489550613e-06,
+      "loss": 0.2219,
+      "step": 15572
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07943656689146e-06,
+      "loss": 0.245,
+      "step": 15573
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07826367166569e-06,
+      "loss": 0.2195,
+      "step": 15574
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.077090803890056e-06,
+      "loss": 0.2342,
+      "step": 15575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075917963581316e-06,
+      "loss": 0.2489,
+      "step": 15576
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07474515075622e-06,
+      "loss": 0.2392,
+      "step": 15577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.073572365431523e-06,
+      "loss": 0.1916,
+      "step": 15578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.072399607623974e-06,
+      "loss": 0.2381,
+      "step": 15579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.071226877350324e-06,
+      "loss": 0.2438,
+      "step": 15580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.070054174627326e-06,
+      "loss": 0.2238,
+      "step": 15581
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.068881499471735e-06,
+      "loss": 0.231,
+      "step": 15582
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.067708851900296e-06,
+      "loss": 0.2412,
+      "step": 15583
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.066536231929762e-06,
+      "loss": 0.2424,
+      "step": 15584
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06536363957688e-06,
+      "loss": 0.2092,
+      "step": 15585
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.064191074858399e-06,
+      "loss": 0.2313,
+      "step": 15586
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.063018537791069e-06,
+      "loss": 0.2296,
+      "step": 15587
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.061846028391642e-06,
+      "loss": 0.2352,
+      "step": 15588
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060673546676865e-06,
+      "loss": 0.2438,
+      "step": 15589
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05950109266348e-06,
+      "loss": 0.2259,
+      "step": 15590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05832866636824e-06,
+      "loss": 0.2487,
+      "step": 15591
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05715626780789e-06,
+      "loss": 0.2258,
+      "step": 15592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05598389699918e-06,
+      "loss": 0.2166,
+      "step": 15593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.054811553958853e-06,
+      "loss": 0.2392,
+      "step": 15594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053639238703653e-06,
+      "loss": 0.209,
+      "step": 15595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.052466951250331e-06,
+      "loss": 0.2307,
+      "step": 15596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.051294691615626e-06,
+      "loss": 0.2323,
+      "step": 15597
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050122459816284e-06,
+      "loss": 0.1962,
+      "step": 15598
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04895025586905e-06,
+      "loss": 0.2345,
+      "step": 15599
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04777807979067e-06,
+      "loss": 0.2204,
+      "step": 15600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046605931597885e-06,
+      "loss": 0.2126,
+      "step": 15601
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.045433811307437e-06,
+      "loss": 0.2246,
+      "step": 15602
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04426171893607e-06,
+      "loss": 0.2499,
+      "step": 15603
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043089654500526e-06,
+      "loss": 0.2124,
+      "step": 15604
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041917618017549e-06,
+      "loss": 0.2392,
+      "step": 15605
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.040745609503878e-06,
+      "loss": 0.2234,
+      "step": 15606
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039573628976254e-06,
+      "loss": 0.2415,
+      "step": 15607
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.038401676451417e-06,
+      "loss": 0.2096,
+      "step": 15608
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.037229751946114e-06,
+      "loss": 0.236,
+      "step": 15609
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.036057855477075e-06,
+      "loss": 0.2393,
+      "step": 15610
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034885987061041e-06,
+      "loss": 0.2282,
+      "step": 15611
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.033714146714757e-06,
+      "loss": 0.2523,
+      "step": 15612
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032542334454955e-06,
+      "loss": 0.2276,
+      "step": 15613
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031370550298378e-06,
+      "loss": 0.2159,
+      "step": 15614
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030198794261762e-06,
+      "loss": 0.1999,
+      "step": 15615
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029027066361844e-06,
+      "loss": 0.2284,
+      "step": 15616
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.027855366615363e-06,
+      "loss": 0.215,
+      "step": 15617
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026683695039054e-06,
+      "loss": 0.2322,
+      "step": 15618
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02551205164965e-06,
+      "loss": 0.2277,
+      "step": 15619
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024340436463894e-06,
+      "loss": 0.2296,
+      "step": 15620
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.023168849498517e-06,
+      "loss": 0.2259,
+      "step": 15621
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.021997290770255e-06,
+      "loss": 0.2164,
+      "step": 15622
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.020825760295841e-06,
+      "loss": 0.2244,
+      "step": 15623
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01965425809201e-06,
+      "loss": 0.2242,
+      "step": 15624
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.018482784175495e-06,
+      "loss": 0.1906,
+      "step": 15625
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01731133856303e-06,
+      "loss": 0.2066,
+      "step": 15626
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.016139921271348e-06,
+      "loss": 0.2024,
+      "step": 15627
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.014968532317184e-06,
+      "loss": 0.2288,
+      "step": 15628
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.013797171717266e-06,
+      "loss": 0.2282,
+      "step": 15629
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012625839488328e-06,
+      "loss": 0.2339,
+      "step": 15630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011454535647102e-06,
+      "loss": 0.2492,
+      "step": 15631
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.010283260210316e-06,
+      "loss": 0.2388,
+      "step": 15632
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.009112013194707e-06,
+      "loss": 0.236,
+      "step": 15633
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.007940794616998e-06,
+      "loss": 0.2082,
+      "step": 15634
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00676960449392e-06,
+      "loss": 0.2388,
+      "step": 15635
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00559844284221e-06,
+      "loss": 0.2042,
+      "step": 15636
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.004427309678584e-06,
+      "loss": 0.2409,
+      "step": 15637
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.003256205019778e-06,
+      "loss": 0.2518,
+      "step": 15638
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002085128882522e-06,
+      "loss": 0.214,
+      "step": 15639
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.000914081283538e-06,
+      "loss": 0.221,
+      "step": 15640
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 0.2504,
+      "step": 15641
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.998572071767306e-06,
+      "loss": 0.2255,
+      "step": 15642
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997401109883507e-06,
+      "loss": 0.2098,
+      "step": 15643
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.996230176604896e-06,
+      "loss": 0.2367,
+      "step": 15644
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995059271948187e-06,
+      "loss": 0.2034,
+      "step": 15645
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993888395930112e-06,
+      "loss": 0.2724,
+      "step": 15646
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.992717548567394e-06,
+      "loss": 0.239,
+      "step": 15647
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991546729876756e-06,
+      "loss": 0.2206,
+      "step": 15648
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99037593987493e-06,
+      "loss": 0.2075,
+      "step": 15649
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98920517857863e-06,
+      "loss": 0.2336,
+      "step": 15650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988034446004582e-06,
+      "loss": 0.239,
+      "step": 15651
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.986863742169508e-06,
+      "loss": 0.2018,
+      "step": 15652
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985693067090132e-06,
+      "loss": 0.2426,
+      "step": 15653
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.984522420783176e-06,
+      "loss": 0.2105,
+      "step": 15654
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983351803265362e-06,
+      "loss": 0.2266,
+      "step": 15655
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98218121455341e-06,
+      "loss": 0.2244,
+      "step": 15656
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98101065466404e-06,
+      "loss": 0.2286,
+      "step": 15657
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979840123613973e-06,
+      "loss": 0.2197,
+      "step": 15658
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97866962141993e-06,
+      "loss": 0.2256,
+      "step": 15659
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97749914809863e-06,
+      "loss": 0.2248,
+      "step": 15660
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.976328703666792e-06,
+      "loss": 0.2068,
+      "step": 15661
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975158288141138e-06,
+      "loss": 0.2234,
+      "step": 15662
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.973987901538379e-06,
+      "loss": 0.2118,
+      "step": 15663
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.972817543875235e-06,
+      "loss": 0.2355,
+      "step": 15664
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971647215168426e-06,
+      "loss": 0.2406,
+      "step": 15665
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.970476915434668e-06,
+      "loss": 0.2363,
+      "step": 15666
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96930664469068e-06,
+      "loss": 0.2143,
+      "step": 15667
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.968136402953171e-06,
+      "loss": 0.2241,
+      "step": 15668
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.966966190238864e-06,
+      "loss": 0.2066,
+      "step": 15669
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.965796006564471e-06,
+      "loss": 0.2143,
+      "step": 15670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96462585194671e-06,
+      "loss": 0.2319,
+      "step": 15671
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.963455726402292e-06,
+      "loss": 0.2516,
+      "step": 15672
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.962285629947933e-06,
+      "loss": 0.2332,
+      "step": 15673
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.961115562600344e-06,
+      "loss": 0.2252,
+      "step": 15674
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.959945524376245e-06,
+      "loss": 0.227,
+      "step": 15675
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95877551529234e-06,
+      "loss": 0.2316,
+      "step": 15676
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.957605535365348e-06,
+      "loss": 0.2373,
+      "step": 15677
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956435584611979e-06,
+      "loss": 0.2282,
+      "step": 15678
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955265663048942e-06,
+      "loss": 0.2259,
+      "step": 15679
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954095770692951e-06,
+      "loss": 0.2217,
+      "step": 15680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952925907560717e-06,
+      "loss": 0.2308,
+      "step": 15681
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95175607366895e-06,
+      "loss": 0.1981,
+      "step": 15682
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950586269034359e-06,
+      "loss": 0.2304,
+      "step": 15683
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.949416493673654e-06,
+      "loss": 0.2208,
+      "step": 15684
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948246747603545e-06,
+      "loss": 0.2229,
+      "step": 15685
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.947077030840738e-06,
+      "loss": 0.2406,
+      "step": 15686
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.945907343401946e-06,
+      "loss": 0.2489,
+      "step": 15687
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.944737685303876e-06,
+      "loss": 0.228,
+      "step": 15688
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943568056563231e-06,
+      "loss": 0.2059,
+      "step": 15689
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94239845719672e-06,
+      "loss": 0.2196,
+      "step": 15690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941228887221048e-06,
+      "loss": 0.2033,
+      "step": 15691
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940059346652924e-06,
+      "loss": 0.2173,
+      "step": 15692
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.938889835509055e-06,
+      "loss": 0.2398,
+      "step": 15693
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.937720353806145e-06,
+      "loss": 0.226,
+      "step": 15694
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.936550901560896e-06,
+      "loss": 0.2163,
+      "step": 15695
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.935381478790015e-06,
+      "loss": 0.2373,
+      "step": 15696
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934212085510206e-06,
+      "loss": 0.2375,
+      "step": 15697
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.933042721738172e-06,
+      "loss": 0.2445,
+      "step": 15698
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93187338749062e-06,
+      "loss": 0.224,
+      "step": 15699
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930704082784244e-06,
+      "loss": 0.2326,
+      "step": 15700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92953480763576e-06,
+      "loss": 0.2159,
+      "step": 15701
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.928365562061856e-06,
+      "loss": 0.2236,
+      "step": 15702
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92719634607924e-06,
+      "loss": 0.234,
+      "step": 15703
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926027159704612e-06,
+      "loss": 0.2148,
+      "step": 15704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924858002954676e-06,
+      "loss": 0.2124,
+      "step": 15705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.923688875846125e-06,
+      "loss": 0.2077,
+      "step": 15706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.922519778395665e-06,
+      "loss": 0.2155,
+      "step": 15707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.921350710619995e-06,
+      "loss": 0.2011,
+      "step": 15708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92018167253581e-06,
+      "loss": 0.224,
+      "step": 15709
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919012664159816e-06,
+      "loss": 0.2446,
+      "step": 15710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917843685508702e-06,
+      "loss": 0.232,
+      "step": 15711
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.916674736599172e-06,
+      "loss": 0.2358,
+      "step": 15712
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91550581744792e-06,
+      "loss": 0.2475,
+      "step": 15713
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.914336928071647e-06,
+      "loss": 0.2187,
+      "step": 15714
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913168068487049e-06,
+      "loss": 0.2426,
+      "step": 15715
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.911999238710815e-06,
+      "loss": 0.2284,
+      "step": 15716
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910830438759647e-06,
+      "loss": 0.2033,
+      "step": 15717
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90966166865024e-06,
+      "loss": 0.236,
+      "step": 15718
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.908492928399283e-06,
+      "loss": 0.2305,
+      "step": 15719
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907324218023477e-06,
+      "loss": 0.1902,
+      "step": 15720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.906155537539513e-06,
+      "loss": 0.2594,
+      "step": 15721
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90498688696409e-06,
+      "loss": 0.216,
+      "step": 15722
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.903818266313889e-06,
+      "loss": 0.2031,
+      "step": 15723
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902649675605612e-06,
+      "loss": 0.2342,
+      "step": 15724
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90148111485595e-06,
+      "loss": 0.2074,
+      "step": 15725
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900312584081595e-06,
+      "loss": 0.2035,
+      "step": 15726
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899144083299233e-06,
+      "loss": 0.2096,
+      "step": 15727
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.897975612525565e-06,
+      "loss": 0.2327,
+      "step": 15728
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896807171777272e-06,
+      "loss": 0.2247,
+      "step": 15729
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895638761071045e-06,
+      "loss": 0.242,
+      "step": 15730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.894470380423578e-06,
+      "loss": 0.2432,
+      "step": 15731
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.893302029851557e-06,
+      "loss": 0.2124,
+      "step": 15732
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892133709371673e-06,
+      "loss": 0.2355,
+      "step": 15733
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.890965419000613e-06,
+      "loss": 0.2217,
+      "step": 15734
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.889797158755064e-06,
+      "loss": 0.2505,
+      "step": 15735
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888628928651715e-06,
+      "loss": 0.205,
+      "step": 15736
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887460728707253e-06,
+      "loss": 0.2239,
+      "step": 15737
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886292558938366e-06,
+      "loss": 0.245,
+      "step": 15738
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885124419361735e-06,
+      "loss": 0.2203,
+      "step": 15739
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88395630999405e-06,
+      "loss": 0.212,
+      "step": 15740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882788230852002e-06,
+      "loss": 0.2143,
+      "step": 15741
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.881620181952263e-06,
+      "loss": 0.2205,
+      "step": 15742
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.880452163311522e-06,
+      "loss": 0.2523,
+      "step": 15743
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87928417494647e-06,
+      "loss": 0.2166,
+      "step": 15744
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878116216873781e-06,
+      "loss": 0.2156,
+      "step": 15745
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876948289110145e-06,
+      "loss": 0.2344,
+      "step": 15746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87578039167224e-06,
+      "loss": 0.2237,
+      "step": 15747
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87461252457675e-06,
+      "loss": 0.2102,
+      "step": 15748
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87344468784036e-06,
+      "loss": 0.2327,
+      "step": 15749
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87227688147975e-06,
+      "loss": 0.198,
+      "step": 15750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871109105511595e-06,
+      "loss": 0.232,
+      "step": 15751
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.869941359952585e-06,
+      "loss": 0.2289,
+      "step": 15752
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868773644819392e-06,
+      "loss": 0.2137,
+      "step": 15753
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.867605960128704e-06,
+      "loss": 0.2174,
+      "step": 15754
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.866438305897191e-06,
+      "loss": 0.2105,
+      "step": 15755
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.865270682141538e-06,
+      "loss": 0.2409,
+      "step": 15756
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86410308887842e-06,
+      "loss": 0.2131,
+      "step": 15757
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862935526124516e-06,
+      "loss": 0.244,
+      "step": 15758
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861767993896505e-06,
+      "loss": 0.2413,
+      "step": 15759
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.860600492211063e-06,
+      "loss": 0.2432,
+      "step": 15760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.859433021084868e-06,
+      "loss": 0.2467,
+      "step": 15761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.858265580534591e-06,
+      "loss": 0.2458,
+      "step": 15762
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.857098170576914e-06,
+      "loss": 0.2206,
+      "step": 15763
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.855930791228508e-06,
+      "loss": 0.2371,
+      "step": 15764
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.854763442506053e-06,
+      "loss": 0.256,
+      "step": 15765
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.853596124426219e-06,
+      "loss": 0.2286,
+      "step": 15766
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852428837005684e-06,
+      "loss": 0.2243,
+      "step": 15767
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.851261580261115e-06,
+      "loss": 0.2238,
+      "step": 15768
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85009435420919e-06,
+      "loss": 0.2237,
+      "step": 15769
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84892715886658e-06,
+      "loss": 0.2145,
+      "step": 15770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.847759994249958e-06,
+      "loss": 0.2203,
+      "step": 15771
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846592860375997e-06,
+      "loss": 0.223,
+      "step": 15772
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.845425757261364e-06,
+      "loss": 0.2072,
+      "step": 15773
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.844258684922734e-06,
+      "loss": 0.2131,
+      "step": 15774
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843091643376779e-06,
+      "loss": 0.226,
+      "step": 15775
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841924632640166e-06,
+      "loss": 0.2135,
+      "step": 15776
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.840757652729563e-06,
+      "loss": 0.2185,
+      "step": 15777
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839590703661641e-06,
+      "loss": 0.2152,
+      "step": 15778
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83842378545307e-06,
+      "loss": 0.2239,
+      "step": 15779
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837256898120521e-06,
+      "loss": 0.2314,
+      "step": 15780
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.836090041680655e-06,
+      "loss": 0.2228,
+      "step": 15781
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834923216150141e-06,
+      "loss": 0.2477,
+      "step": 15782
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833756421545652e-06,
+      "loss": 0.2343,
+      "step": 15783
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.832589657883846e-06,
+      "loss": 0.2228,
+      "step": 15784
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831422925181396e-06,
+      "loss": 0.2133,
+      "step": 15785
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.830256223454962e-06,
+      "loss": 0.2279,
+      "step": 15786
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.829089552721213e-06,
+      "loss": 0.2263,
+      "step": 15787
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827922912996816e-06,
+      "loss": 0.2394,
+      "step": 15788
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.826756304298428e-06,
+      "loss": 0.2173,
+      "step": 15789
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825589726642718e-06,
+      "loss": 0.2472,
+      "step": 15790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824423180046347e-06,
+      "loss": 0.2119,
+      "step": 15791
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823256664525985e-06,
+      "loss": 0.2323,
+      "step": 15792
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822090180098283e-06,
+      "loss": 0.2232,
+      "step": 15793
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820923726779916e-06,
+      "loss": 0.2413,
+      "step": 15794
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.819757304587534e-06,
+      "loss": 0.2368,
+      "step": 15795
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818590913537803e-06,
+      "loss": 0.2425,
+      "step": 15796
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.817424553647382e-06,
+      "loss": 0.2052,
+      "step": 15797
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816258224932935e-06,
+      "loss": 0.2349,
+      "step": 15798
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81509192741112e-06,
+      "loss": 0.2011,
+      "step": 15799
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.813925661098595e-06,
+      "loss": 0.2194,
+      "step": 15800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812759426012023e-06,
+      "loss": 0.2153,
+      "step": 15801
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811593222168057e-06,
+      "loss": 0.216,
+      "step": 15802
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810427049583358e-06,
+      "loss": 0.2321,
+      "step": 15803
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.809260908274585e-06,
+      "loss": 0.2162,
+      "step": 15804
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808094798258394e-06,
+      "loss": 0.2251,
+      "step": 15805
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806928719551439e-06,
+      "loss": 0.2642,
+      "step": 15806
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.805762672170385e-06,
+      "loss": 0.2326,
+      "step": 15807
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804596656131877e-06,
+      "loss": 0.2163,
+      "step": 15808
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803430671452574e-06,
+      "loss": 0.2381,
+      "step": 15809
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802264718149134e-06,
+      "loss": 0.2359,
+      "step": 15810
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80109879623821e-06,
+      "loss": 0.2171,
+      "step": 15811
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.799932905736454e-06,
+      "loss": 0.1962,
+      "step": 15812
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.798767046660521e-06,
+      "loss": 0.2115,
+      "step": 15813
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.797601219027066e-06,
+      "loss": 0.2362,
+      "step": 15814
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79643542285274e-06,
+      "loss": 0.2214,
+      "step": 15815
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795269658154193e-06,
+      "loss": 0.2291,
+      "step": 15816
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.794103924948081e-06,
+      "loss": 0.2372,
+      "step": 15817
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.792938223251054e-06,
+      "loss": 0.2367,
+      "step": 15818
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79177255307976e-06,
+      "loss": 0.2174,
+      "step": 15819
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790606914450859e-06,
+      "loss": 0.2278,
+      "step": 15820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.789441307380986e-06,
+      "loss": 0.2194,
+      "step": 15821
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788275731886805e-06,
+      "loss": 0.2397,
+      "step": 15822
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.787110187984954e-06,
+      "loss": 0.2385,
+      "step": 15823
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785944675692087e-06,
+      "loss": 0.2287,
+      "step": 15824
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784779195024852e-06,
+      "loss": 0.2123,
+      "step": 15825
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.783613745999896e-06,
+      "loss": 0.253,
+      "step": 15826
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.782448328633867e-06,
+      "loss": 0.2135,
+      "step": 15827
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.781282942943411e-06,
+      "loss": 0.2274,
+      "step": 15828
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780117588945175e-06,
+      "loss": 0.205,
+      "step": 15829
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778952266655804e-06,
+      "loss": 0.2134,
+      "step": 15830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.777786976091946e-06,
+      "loss": 0.2354,
+      "step": 15831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.776621717270243e-06,
+      "loss": 0.2297,
+      "step": 15832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.775456490207345e-06,
+      "loss": 0.2257,
+      "step": 15833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774291294919887e-06,
+      "loss": 0.2077,
+      "step": 15834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.773126131424519e-06,
+      "loss": 0.1922,
+      "step": 15835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771960999737883e-06,
+      "loss": 0.2241,
+      "step": 15836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77079589987662e-06,
+      "loss": 0.2134,
+      "step": 15837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76963083185738e-06,
+      "loss": 0.2276,
+      "step": 15838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.768465795696794e-06,
+      "loss": 0.2502,
+      "step": 15839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767300791411507e-06,
+      "loss": 0.2479,
+      "step": 15840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.766135819018163e-06,
+      "loss": 0.219,
+      "step": 15841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764970878533402e-06,
+      "loss": 0.2583,
+      "step": 15842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.763805969973863e-06,
+      "loss": 0.2111,
+      "step": 15843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.762641093356185e-06,
+      "loss": 0.2073,
+      "step": 15844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761476248697006e-06,
+      "loss": 0.225,
+      "step": 15845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76031143601297e-06,
+      "loss": 0.2069,
+      "step": 15846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75914665532071e-06,
+      "loss": 0.2172,
+      "step": 15847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757981906636864e-06,
+      "loss": 0.2273,
+      "step": 15848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.756817189978072e-06,
+      "loss": 0.2174,
+      "step": 15849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755652505360967e-06,
+      "loss": 0.226,
+      "step": 15850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754487852802188e-06,
+      "loss": 0.2252,
+      "step": 15851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.753323232318371e-06,
+      "loss": 0.2533,
+      "step": 15852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.752158643926151e-06,
+      "loss": 0.2056,
+      "step": 15853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.750994087642166e-06,
+      "loss": 0.2252,
+      "step": 15854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.749829563483045e-06,
+      "loss": 0.2555,
+      "step": 15855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748665071465425e-06,
+      "loss": 0.251,
+      "step": 15856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74750061160594e-06,
+      "loss": 0.2471,
+      "step": 15857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.746336183921222e-06,
+      "loss": 0.2168,
+      "step": 15858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745171788427909e-06,
+      "loss": 0.2371,
+      "step": 15859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744007425142623e-06,
+      "loss": 0.231,
+      "step": 15860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742843094082005e-06,
+      "loss": 0.2287,
+      "step": 15861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74167879526268e-06,
+      "loss": 0.1982,
+      "step": 15862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740514528701282e-06,
+      "loss": 0.2265,
+      "step": 15863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.739350294414442e-06,
+      "loss": 0.2062,
+      "step": 15864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73818609241879e-06,
+      "loss": 0.2324,
+      "step": 15865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.737021922730953e-06,
+      "loss": 0.1966,
+      "step": 15866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73585778536756e-06,
+      "loss": 0.2337,
+      "step": 15867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.734693680345244e-06,
+      "loss": 0.1873,
+      "step": 15868
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.733529607680627e-06,
+      "loss": 0.2374,
+      "step": 15869
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732365567390344e-06,
+      "loss": 0.2293,
+      "step": 15870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.731201559491016e-06,
+      "loss": 0.2494,
+      "step": 15871
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730037583999271e-06,
+      "loss": 0.2166,
+      "step": 15872
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728873640931741e-06,
+      "loss": 0.2315,
+      "step": 15873
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.727709730305041e-06,
+      "loss": 0.221,
+      "step": 15874
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726545852135805e-06,
+      "loss": 0.2347,
+      "step": 15875
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.725382006440654e-06,
+      "loss": 0.2498,
+      "step": 15876
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724218193236215e-06,
+      "loss": 0.2482,
+      "step": 15877
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.723054412539108e-06,
+      "loss": 0.2215,
+      "step": 15878
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.721890664365958e-06,
+      "loss": 0.2068,
+      "step": 15879
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720726948733391e-06,
+      "loss": 0.2193,
+      "step": 15880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719563265658028e-06,
+      "loss": 0.1992,
+      "step": 15881
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71839961515649e-06,
+      "loss": 0.2329,
+      "step": 15882
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.717235997245398e-06,
+      "loss": 0.2342,
+      "step": 15883
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716072411941376e-06,
+      "loss": 0.2411,
+      "step": 15884
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.714908859261042e-06,
+      "loss": 0.2282,
+      "step": 15885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.713745339221019e-06,
+      "loss": 0.2169,
+      "step": 15886
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712581851837924e-06,
+      "loss": 0.2076,
+      "step": 15887
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.711418397128378e-06,
+      "loss": 0.2352,
+      "step": 15888
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710254975108997e-06,
+      "loss": 0.2051,
+      "step": 15889
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709091585796401e-06,
+      "loss": 0.2371,
+      "step": 15890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.707928229207208e-06,
+      "loss": 0.1926,
+      "step": 15891
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706764905358037e-06,
+      "loss": 0.2072,
+      "step": 15892
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.705601614265507e-06,
+      "loss": 0.2225,
+      "step": 15893
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.704438355946227e-06,
+      "loss": 0.2359,
+      "step": 15894
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703275130416817e-06,
+      "loss": 0.2279,
+      "step": 15895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702111937693894e-06,
+      "loss": 0.2182,
+      "step": 15896
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700948777794073e-06,
+      "loss": 0.2221,
+      "step": 15897
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.699785650733966e-06,
+      "loss": 0.2059,
+      "step": 15898
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.698622556530194e-06,
+      "loss": 0.2164,
+      "step": 15899
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697459495199357e-06,
+      "loss": 0.2412,
+      "step": 15900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69629646675808e-06,
+      "loss": 0.2479,
+      "step": 15901
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69513347122297e-06,
+      "loss": 0.1968,
+      "step": 15902
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693970508610644e-06,
+      "loss": 0.2575,
+      "step": 15903
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.692807578937712e-06,
+      "loss": 0.2494,
+      "step": 15904
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.691644682220783e-06,
+      "loss": 0.2297,
+      "step": 15905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690481818476468e-06,
+      "loss": 0.2123,
+      "step": 15906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68931898772138e-06,
+      "loss": 0.2266,
+      "step": 15907
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688156189972126e-06,
+      "loss": 0.2399,
+      "step": 15908
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68699342524532e-06,
+      "loss": 0.2274,
+      "step": 15909
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685830693557566e-06,
+      "loss": 0.2221,
+      "step": 15910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684667994925477e-06,
+      "loss": 0.2575,
+      "step": 15911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.683505329365659e-06,
+      "loss": 0.2426,
+      "step": 15912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.682342696894718e-06,
+      "loss": 0.2279,
+      "step": 15913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.681180097529261e-06,
+      "loss": 0.2092,
+      "step": 15914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680017531285894e-06,
+      "loss": 0.2252,
+      "step": 15915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.678854998181231e-06,
+      "loss": 0.2141,
+      "step": 15916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677692498231867e-06,
+      "loss": 0.213,
+      "step": 15917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676530031454411e-06,
+      "loss": 0.2234,
+      "step": 15918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67536759786547e-06,
+      "loss": 0.2347,
+      "step": 15919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67420519748165e-06,
+      "loss": 0.2165,
+      "step": 15920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673042830319548e-06,
+      "loss": 0.2195,
+      "step": 15921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67188049639577e-06,
+      "loss": 0.1832,
+      "step": 15922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67071819572692e-06,
+      "loss": 0.2128,
+      "step": 15923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669555928329601e-06,
+      "loss": 0.2171,
+      "step": 15924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.668393694220418e-06,
+      "loss": 0.2258,
+      "step": 15925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667231493415963e-06,
+      "loss": 0.2377,
+      "step": 15926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666069325932845e-06,
+      "loss": 0.2408,
+      "step": 15927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.664907191787659e-06,
+      "loss": 0.233,
+      "step": 15928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663745090997008e-06,
+      "loss": 0.251,
+      "step": 15929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66258302357749e-06,
+      "loss": 0.2366,
+      "step": 15930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661420989545706e-06,
+      "loss": 0.2336,
+      "step": 15931
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660258988918254e-06,
+      "loss": 0.22,
+      "step": 15932
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65909702171173e-06,
+      "loss": 0.2233,
+      "step": 15933
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657935087942736e-06,
+      "loss": 0.208,
+      "step": 15934
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.656773187627863e-06,
+      "loss": 0.2158,
+      "step": 15935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655611320783715e-06,
+      "loss": 0.2305,
+      "step": 15936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65444948742688e-06,
+      "loss": 0.2464,
+      "step": 15937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.653287687573963e-06,
+      "loss": 0.2131,
+      "step": 15938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.652125921241549e-06,
+      "loss": 0.2003,
+      "step": 15939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.650964188446238e-06,
+      "loss": 0.221,
+      "step": 15940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.649802489204624e-06,
+      "loss": 0.1985,
+      "step": 15941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.6486408235333e-06,
+      "loss": 0.2414,
+      "step": 15942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.647479191448862e-06,
+      "loss": 0.2349,
+      "step": 15943
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6463175929679e-06,
+      "loss": 0.2317,
+      "step": 15944
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.645156028107005e-06,
+      "loss": 0.243,
+      "step": 15945
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.643994496882771e-06,
+      "loss": 0.2425,
+      "step": 15946
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642832999311788e-06,
+      "loss": 0.23,
+      "step": 15947
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.641671535410653e-06,
+      "loss": 0.2193,
+      "step": 15948
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.640510105195945e-06,
+      "loss": 0.2198,
+      "step": 15949
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639348708684264e-06,
+      "loss": 0.1984,
+      "step": 15950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638187345892194e-06,
+      "loss": 0.2267,
+      "step": 15951
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.637026016836331e-06,
+      "loss": 0.2201,
+      "step": 15952
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635864721533253e-06,
+      "loss": 0.2191,
+      "step": 15953
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.634703459999555e-06,
+      "loss": 0.2455,
+      "step": 15954
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63354223225182e-06,
+      "loss": 0.223,
+      "step": 15955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632381038306638e-06,
+      "loss": 0.2017,
+      "step": 15956
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631219878180596e-06,
+      "loss": 0.2328,
+      "step": 15957
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.630058751890277e-06,
+      "loss": 0.2259,
+      "step": 15958
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628897659452272e-06,
+      "loss": 0.2259,
+      "step": 15959
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62773660088316e-06,
+      "loss": 0.2317,
+      "step": 15960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62657557619953e-06,
+      "loss": 0.2258,
+      "step": 15961
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625414585417963e-06,
+      "loss": 0.2105,
+      "step": 15962
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.624253628555044e-06,
+      "loss": 0.2432,
+      "step": 15963
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623092705627358e-06,
+      "loss": 0.211,
+      "step": 15964
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.621931816651487e-06,
+      "loss": 0.2403,
+      "step": 15965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.620770961644011e-06,
+      "loss": 0.2407,
+      "step": 15966
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.619610140621512e-06,
+      "loss": 0.2222,
+      "step": 15967
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61844935360057e-06,
+      "loss": 0.2391,
+      "step": 15968
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.617288600597769e-06,
+      "loss": 0.2385,
+      "step": 15969
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.616127881629689e-06,
+      "loss": 0.2298,
+      "step": 15970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.614967196712908e-06,
+      "loss": 0.2038,
+      "step": 15971
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613806545864004e-06,
+      "loss": 0.2433,
+      "step": 15972
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.612645929099559e-06,
+      "loss": 0.2363,
+      "step": 15973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.611485346436149e-06,
+      "loss": 0.227,
+      "step": 15974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.610324797890356e-06,
+      "loss": 0.2264,
+      "step": 15975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6091642834787516e-06,
+      "loss": 0.238,
+      "step": 15976
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608003803217914e-06,
+      "loss": 0.2454,
+      "step": 15977
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.2373,
+      "step": 15978
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605682945214853e-06,
+      "loss": 0.2317,
+      "step": 15979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.604522567505775e-06,
+      "loss": 0.2115,
+      "step": 15980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.603362224013769e-06,
+      "loss": 0.277,
+      "step": 15981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602201914755408e-06,
+      "loss": 0.2197,
+      "step": 15982
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.601041639747264e-06,
+      "loss": 0.2207,
+      "step": 15983
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599881399005913e-06,
+      "loss": 0.1833,
+      "step": 15984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.598721192547926e-06,
+      "loss": 0.2347,
+      "step": 15985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597561020389878e-06,
+      "loss": 0.24,
+      "step": 15986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596400882548337e-06,
+      "loss": 0.2267,
+      "step": 15987
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.595240779039876e-06,
+      "loss": 0.2237,
+      "step": 15988
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5940807098810685e-06,
+      "loss": 0.2426,
+      "step": 15989
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.592920675088482e-06,
+      "loss": 0.2303,
+      "step": 15990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.591760674678692e-06,
+      "loss": 0.236,
+      "step": 15991
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590600708668258e-06,
+      "loss": 0.2169,
+      "step": 15992
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5894407770737585e-06,
+      "loss": 0.2407,
+      "step": 15993
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588280879911756e-06,
+      "loss": 0.2155,
+      "step": 15994
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58712101719882e-06,
+      "loss": 0.2438,
+      "step": 15995
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585961188951518e-06,
+      "loss": 0.2374,
+      "step": 15996
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.584801395186417e-06,
+      "loss": 0.2341,
+      "step": 15997
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.583641635920087e-06,
+      "loss": 0.2423,
+      "step": 15998
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58248191116909e-06,
+      "loss": 0.2102,
+      "step": 15999
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.581322220949992e-06,
+      "loss": 0.2254,
+      "step": 16000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.580162565279358e-06,
+      "loss": 0.2374,
+      "step": 16001
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.579002944173752e-06,
+      "loss": 0.2247,
+      "step": 16002
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.577843357649743e-06,
+      "loss": 0.2037,
+      "step": 16003
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5766838057238915e-06,
+      "loss": 0.222,
+      "step": 16004
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575524288412759e-06,
+      "loss": 0.2188,
+      "step": 16005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.574364805732908e-06,
+      "loss": 0.1891,
+      "step": 16006
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5732053577009e-06,
+      "loss": 0.221,
+      "step": 16007
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.572045944333299e-06,
+      "loss": 0.2455,
+      "step": 16008
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570886565646668e-06,
+      "loss": 0.2245,
+      "step": 16009
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.569727221657561e-06,
+      "loss": 0.228,
+      "step": 16010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.568567912382544e-06,
+      "loss": 0.227,
+      "step": 16011
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.567408637838174e-06,
+      "loss": 0.2226,
+      "step": 16012
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56624939804101e-06,
+      "loss": 0.2118,
+      "step": 16013
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.565090193007614e-06,
+      "loss": 0.217,
+      "step": 16014
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.563931022754539e-06,
+      "loss": 0.2349,
+      "step": 16015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.562771887298345e-06,
+      "loss": 0.2173,
+      "step": 16016
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56161278665559e-06,
+      "loss": 0.2449,
+      "step": 16017
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.560453720842834e-06,
+      "loss": 0.221,
+      "step": 16018
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.559294689876624e-06,
+      "loss": 0.2248,
+      "step": 16019
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.558135693773522e-06,
+      "loss": 0.2082,
+      "step": 16020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.556976732550082e-06,
+      "loss": 0.1894,
+      "step": 16021
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555817806222857e-06,
+      "loss": 0.26,
+      "step": 16022
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.554658914808404e-06,
+      "loss": 0.2378,
+      "step": 16023
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.553500058323273e-06,
+      "loss": 0.2351,
+      "step": 16024
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.552341236784022e-06,
+      "loss": 0.2419,
+      "step": 16025
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5511824502072e-06,
+      "loss": 0.219,
+      "step": 16026
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.550023698609358e-06,
+      "loss": 0.2532,
+      "step": 16027
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54886498200705e-06,
+      "loss": 0.2289,
+      "step": 16028
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5477063004168285e-06,
+      "loss": 0.2127,
+      "step": 16029
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.546547653855243e-06,
+      "loss": 0.2038,
+      "step": 16030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545389042338844e-06,
+      "loss": 0.2241,
+      "step": 16031
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54423046588418e-06,
+      "loss": 0.2264,
+      "step": 16032
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.543071924507799e-06,
+      "loss": 0.2342,
+      "step": 16033
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.54191341822625e-06,
+      "loss": 0.2328,
+      "step": 16034
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.540754947056083e-06,
+      "loss": 0.236,
+      "step": 16035
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.539596511013843e-06,
+      "loss": 0.235,
+      "step": 16036
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538438110116084e-06,
+      "loss": 0.2048,
+      "step": 16037
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.537279744379343e-06,
+      "loss": 0.2265,
+      "step": 16038
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.536121413820172e-06,
+      "loss": 0.2233,
+      "step": 16039
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534963118455115e-06,
+      "loss": 0.2385,
+      "step": 16040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.53380485830072e-06,
+      "loss": 0.2348,
+      "step": 16041
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.532646633373526e-06,
+      "loss": 0.2236,
+      "step": 16042
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5314884436900815e-06,
+      "loss": 0.2203,
+      "step": 16043
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.530330289266933e-06,
+      "loss": 0.2198,
+      "step": 16044
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5291721701206144e-06,
+      "loss": 0.225,
+      "step": 16045
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528014086267673e-06,
+      "loss": 0.2278,
+      "step": 16046
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.526856037724652e-06,
+      "loss": 0.2106,
+      "step": 16047
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.525698024508095e-06,
+      "loss": 0.2267,
+      "step": 16048
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5245400466345364e-06,
+      "loss": 0.2387,
+      "step": 16049
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523382104120523e-06,
+      "loss": 0.2051,
+      "step": 16050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52222419698259e-06,
+      "loss": 0.2109,
+      "step": 16051
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5210663252372815e-06,
+      "loss": 0.2149,
+      "step": 16052
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.519908488901136e-06,
+      "loss": 0.2299,
+      "step": 16053
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51875068799069e-06,
+      "loss": 0.2295,
+      "step": 16054
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.517592922522481e-06,
+      "loss": 0.223,
+      "step": 16055
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516435192513048e-06,
+      "loss": 0.2476,
+      "step": 16056
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.515277497978932e-06,
+      "loss": 0.2197,
+      "step": 16057
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514119838936662e-06,
+      "loss": 0.2147,
+      "step": 16058
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.512962215402779e-06,
+      "loss": 0.2414,
+      "step": 16059
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.511804627393815e-06,
+      "loss": 0.2274,
+      "step": 16060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510647074926307e-06,
+      "loss": 0.2288,
+      "step": 16061
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.50948955801679e-06,
+      "loss": 0.2199,
+      "step": 16062
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.508332076681797e-06,
+      "loss": 0.2183,
+      "step": 16063
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.507174630937865e-06,
+      "loss": 0.2132,
+      "step": 16064
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506017220801521e-06,
+      "loss": 0.2049,
+      "step": 16065
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504859846289301e-06,
+      "loss": 0.2265,
+      "step": 16066
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503702507417736e-06,
+      "loss": 0.209,
+      "step": 16067
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.502545204203359e-06,
+      "loss": 0.2424,
+      "step": 16068
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.501387936662701e-06,
+      "loss": 0.2018,
+      "step": 16069
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500230704812292e-06,
+      "loss": 0.2348,
+      "step": 16070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499073508668661e-06,
+      "loss": 0.2039,
+      "step": 16071
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497916348248335e-06,
+      "loss": 0.2277,
+      "step": 16072
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.496759223567847e-06,
+      "loss": 0.2469,
+      "step": 16073
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.495602134643723e-06,
+      "loss": 0.2581,
+      "step": 16074
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.494445081492494e-06,
+      "loss": 0.1974,
+      "step": 16075
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4932880641306834e-06,
+      "loss": 0.2412,
+      "step": 16076
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.492131082574821e-06,
+      "loss": 0.2332,
+      "step": 16077
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490974136841432e-06,
+      "loss": 0.2262,
+      "step": 16078
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.489817226947041e-06,
+      "loss": 0.2161,
+      "step": 16079
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.488660352908177e-06,
+      "loss": 0.2113,
+      "step": 16080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487503514741361e-06,
+      "loss": 0.2169,
+      "step": 16081
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486346712463119e-06,
+      "loss": 0.2326,
+      "step": 16082
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.485189946089979e-06,
+      "loss": 0.239,
+      "step": 16083
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484033215638455e-06,
+      "loss": 0.2217,
+      "step": 16084
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.482876521125075e-06,
+      "loss": 0.2155,
+      "step": 16085
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.481719862566361e-06,
+      "loss": 0.2064,
+      "step": 16086
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.480563239978837e-06,
+      "loss": 0.226,
+      "step": 16087
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.47940665337902e-06,
+      "loss": 0.2292,
+      "step": 16088
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.478250102783433e-06,
+      "loss": 0.2207,
+      "step": 16089
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4770935882085946e-06,
+      "loss": 0.2378,
+      "step": 16090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.47593710967103e-06,
+      "loss": 0.2182,
+      "step": 16091
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.474780667187249e-06,
+      "loss": 0.2298,
+      "step": 16092
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4736242607737775e-06,
+      "loss": 0.2228,
+      "step": 16093
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.472467890447131e-06,
+      "loss": 0.2258,
+      "step": 16094
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471311556223828e-06,
+      "loss": 0.2162,
+      "step": 16095
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4701552581203864e-06,
+      "loss": 0.226,
+      "step": 16096
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4689989961533224e-06,
+      "loss": 0.2261,
+      "step": 16097
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.467842770339152e-06,
+      "loss": 0.208,
+      "step": 16098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.466686580694387e-06,
+      "loss": 0.2189,
+      "step": 16099
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.465530427235546e-06,
+      "loss": 0.2248,
+      "step": 16100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.464374309979143e-06,
+      "loss": 0.2212,
+      "step": 16101
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463218228941691e-06,
+      "loss": 0.2197,
+      "step": 16102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.462062184139709e-06,
+      "loss": 0.2121,
+      "step": 16103
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.460906175589702e-06,
+      "loss": 0.2236,
+      "step": 16104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.459750203308186e-06,
+      "loss": 0.2074,
+      "step": 16105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.458594267311673e-06,
+      "loss": 0.2363,
+      "step": 16106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457438367616674e-06,
+      "loss": 0.2135,
+      "step": 16107
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456282504239704e-06,
+      "loss": 0.236,
+      "step": 16108
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4551266771972664e-06,
+      "loss": 0.219,
+      "step": 16109
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453970886505879e-06,
+      "loss": 0.1996,
+      "step": 16110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45281513218204e-06,
+      "loss": 0.216,
+      "step": 16111
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451659414242268e-06,
+      "loss": 0.2006,
+      "step": 16112
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4505037327030675e-06,
+      "loss": 0.2345,
+      "step": 16113
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.449348087580947e-06,
+      "loss": 0.213,
+      "step": 16114
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.448192478892413e-06,
+      "loss": 0.234,
+      "step": 16115
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447036906653973e-06,
+      "loss": 0.2326,
+      "step": 16116
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445881370882133e-06,
+      "loss": 0.2183,
+      "step": 16117
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.444725871593398e-06,
+      "loss": 0.233,
+      "step": 16118
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.443570408804277e-06,
+      "loss": 0.2091,
+      "step": 16119
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442414982531269e-06,
+      "loss": 0.2198,
+      "step": 16120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.441259592790881e-06,
+      "loss": 0.2317,
+      "step": 16121
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4401042395996146e-06,
+      "loss": 0.2134,
+      "step": 16122
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.438948922973981e-06,
+      "loss": 0.2165,
+      "step": 16123
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.437793642930473e-06,
+      "loss": 0.2218,
+      "step": 16124
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.436638399485594e-06,
+      "loss": 0.2265,
+      "step": 16125
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.435483192655851e-06,
+      "loss": 0.2221,
+      "step": 16126
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4343280224577395e-06,
+      "loss": 0.2108,
+      "step": 16127
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433172888907761e-06,
+      "loss": 0.2167,
+      "step": 16128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.432017792022419e-06,
+      "loss": 0.2125,
+      "step": 16129
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.43086273181821e-06,
+      "loss": 0.2331,
+      "step": 16130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.429707708311633e-06,
+      "loss": 0.2216,
+      "step": 16131
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4285527215191865e-06,
+      "loss": 0.2369,
+      "step": 16132
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.427397771457369e-06,
+      "loss": 0.2237,
+      "step": 16133
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426242858142676e-06,
+      "loss": 0.2359,
+      "step": 16134
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42508798159161e-06,
+      "loss": 0.2341,
+      "step": 16135
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423933141820663e-06,
+      "loss": 0.2314,
+      "step": 16136
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42277833884633e-06,
+      "loss": 0.2071,
+      "step": 16137
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4216235726851046e-06,
+      "loss": 0.2337,
+      "step": 16138
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.420468843353486e-06,
+      "loss": 0.1982,
+      "step": 16139
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419314150867964e-06,
+      "loss": 0.2408,
+      "step": 16140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418159495245036e-06,
+      "loss": 0.2476,
+      "step": 16141
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.417004876501197e-06,
+      "loss": 0.2322,
+      "step": 16142
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4158502946529336e-06,
+      "loss": 0.2256,
+      "step": 16143
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.414695749716741e-06,
+      "loss": 0.2276,
+      "step": 16144
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.41354124170911e-06,
+      "loss": 0.2202,
+      "step": 16145
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.412386770646534e-06,
+      "loss": 0.254,
+      "step": 16146
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4112323365455e-06,
+      "loss": 0.2391,
+      "step": 16147
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4100779394225e-06,
+      "loss": 0.2358,
+      "step": 16148
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.408923579294026e-06,
+      "loss": 0.2135,
+      "step": 16149
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4077692561765605e-06,
+      "loss": 0.1998,
+      "step": 16150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.406614970086597e-06,
+      "loss": 0.2039,
+      "step": 16151
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.405460721040618e-06,
+      "loss": 0.2333,
+      "step": 16152
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.404306509055119e-06,
+      "loss": 0.2301,
+      "step": 16153
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4031523341465805e-06,
+      "loss": 0.208,
+      "step": 16154
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40199819633149e-06,
+      "loss": 0.2151,
+      "step": 16155
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.400844095626334e-06,
+      "loss": 0.2168,
+      "step": 16156
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.399690032047597e-06,
+      "loss": 0.2278,
+      "step": 16157
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3985360056117675e-06,
+      "loss": 0.2349,
+      "step": 16158
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.397382016335324e-06,
+      "loss": 0.2006,
+      "step": 16159
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3962280642347515e-06,
+      "loss": 0.2341,
+      "step": 16160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.395074149326535e-06,
+      "loss": 0.2243,
+      "step": 16161
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393920271627159e-06,
+      "loss": 0.2386,
+      "step": 16162
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3927664311531e-06,
+      "loss": 0.2259,
+      "step": 16163
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391612627920844e-06,
+      "loss": 0.2209,
+      "step": 16164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390458861946867e-06,
+      "loss": 0.2064,
+      "step": 16165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.389305133247655e-06,
+      "loss": 0.2145,
+      "step": 16166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3881514418396834e-06,
+      "loss": 0.2284,
+      "step": 16167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386997787739434e-06,
+      "loss": 0.2172,
+      "step": 16168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385844170963388e-06,
+      "loss": 0.213,
+      "step": 16169
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384690591528017e-06,
+      "loss": 0.2043,
+      "step": 16170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383537049449804e-06,
+      "loss": 0.1969,
+      "step": 16171
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.382383544745224e-06,
+      "loss": 0.2198,
+      "step": 16172
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3812300774307555e-06,
+      "loss": 0.2272,
+      "step": 16173
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3800766475228755e-06,
+      "loss": 0.214,
+      "step": 16174
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378923255038056e-06,
+      "loss": 0.1996,
+      "step": 16175
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377769899992779e-06,
+      "loss": 0.2107,
+      "step": 16176
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.376616582403509e-06,
+      "loss": 0.2322,
+      "step": 16177
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.375463302286725e-06,
+      "loss": 0.2059,
+      "step": 16178
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3743100596589e-06,
+      "loss": 0.2136,
+      "step": 16179
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373156854536509e-06,
+      "loss": 0.2369,
+      "step": 16180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.372003686936023e-06,
+      "loss": 0.2153,
+      "step": 16181
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370850556873913e-06,
+      "loss": 0.2216,
+      "step": 16182
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369697464366651e-06,
+      "loss": 0.182,
+      "step": 16183
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368544409430708e-06,
+      "loss": 0.216,
+      "step": 16184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367391392082557e-06,
+      "loss": 0.2286,
+      "step": 16185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.366238412338661e-06,
+      "loss": 0.2404,
+      "step": 16186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365085470215494e-06,
+      "loss": 0.2184,
+      "step": 16187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.363932565729524e-06,
+      "loss": 0.2375,
+      "step": 16188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.362779698897225e-06,
+      "loss": 0.2263,
+      "step": 16189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361626869735051e-06,
+      "loss": 0.2423,
+      "step": 16190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.360474078259479e-06,
+      "loss": 0.2248,
+      "step": 16191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.359321324486974e-06,
+      "loss": 0.2385,
+      "step": 16192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358168608433999e-06,
+      "loss": 0.2202,
+      "step": 16193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.357015930117022e-06,
+      "loss": 0.217,
+      "step": 16194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355863289552507e-06,
+      "loss": 0.2432,
+      "step": 16195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354710686756918e-06,
+      "loss": 0.2197,
+      "step": 16196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3535581217467224e-06,
+      "loss": 0.2277,
+      "step": 16197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352405594538379e-06,
+      "loss": 0.2278,
+      "step": 16198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35125310514835e-06,
+      "loss": 0.1948,
+      "step": 16199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350100653593102e-06,
+      "loss": 0.2091,
+      "step": 16200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.348948239889097e-06,
+      "loss": 0.2147,
+      "step": 16201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.347795864052791e-06,
+      "loss": 0.234,
+      "step": 16202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.346643526100649e-06,
+      "loss": 0.2342,
+      "step": 16203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345491226049127e-06,
+      "loss": 0.2111,
+      "step": 16204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.344338963914686e-06,
+      "loss": 0.2246,
+      "step": 16205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343186739713784e-06,
+      "loss": 0.2415,
+      "step": 16206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.342034553462882e-06,
+      "loss": 0.2326,
+      "step": 16207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34088240517844e-06,
+      "loss": 0.2066,
+      "step": 16208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.33973029487691e-06,
+      "loss": 0.2468,
+      "step": 16209
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3385782225747495e-06,
+      "loss": 0.1801,
+      "step": 16210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.337426188288416e-06,
+      "loss": 0.2381,
+      "step": 16211
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3362741920343675e-06,
+      "loss": 0.2399,
+      "step": 16212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.335122233829057e-06,
+      "loss": 0.2442,
+      "step": 16213
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.333970313688937e-06,
+      "loss": 0.2066,
+      "step": 16214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.332818431630469e-06,
+      "loss": 0.1945,
+      "step": 16215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.331666587670097e-06,
+      "loss": 0.2182,
+      "step": 16216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.330514781824276e-06,
+      "loss": 0.2145,
+      "step": 16217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.329363014109463e-06,
+      "loss": 0.2373,
+      "step": 16218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.328211284542108e-06,
+      "loss": 0.2224,
+      "step": 16219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.327059593138661e-06,
+      "loss": 0.2268,
+      "step": 16220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.325907939915572e-06,
+      "loss": 0.2119,
+      "step": 16221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.324756324889294e-06,
+      "loss": 0.1929,
+      "step": 16222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323604748076274e-06,
+      "loss": 0.1994,
+      "step": 16223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.322453209492967e-06,
+      "loss": 0.2234,
+      "step": 16224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.321301709155813e-06,
+      "loss": 0.2325,
+      "step": 16225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3201502470812655e-06,
+      "loss": 0.2204,
+      "step": 16226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.318998823285771e-06,
+      "loss": 0.1976,
+      "step": 16227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.317847437785781e-06,
+      "loss": 0.2173,
+      "step": 16228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316696090597732e-06,
+      "loss": 0.2157,
+      "step": 16229
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315544781738075e-06,
+      "loss": 0.2346,
+      "step": 16230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.314393511223259e-06,
+      "loss": 0.202,
+      "step": 16231
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.313242279069721e-06,
+      "loss": 0.2374,
+      "step": 16232
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.312091085293911e-06,
+      "loss": 0.2166,
+      "step": 16233
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310939929912271e-06,
+      "loss": 0.2529,
+      "step": 16234
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309788812941246e-06,
+      "loss": 0.2388,
+      "step": 16235
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.308637734397275e-06,
+      "loss": 0.2317,
+      "step": 16236
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307486694296802e-06,
+      "loss": 0.2136,
+      "step": 16237
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3063356926562685e-06,
+      "loss": 0.2345,
+      "step": 16238
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.305184729492116e-06,
+      "loss": 0.2023,
+      "step": 16239
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304033804820785e-06,
+      "loss": 0.2355,
+      "step": 16240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3028829186587156e-06,
+      "loss": 0.2451,
+      "step": 16241
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301732071022346e-06,
+      "loss": 0.2249,
+      "step": 16242
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300581261928112e-06,
+      "loss": 0.2311,
+      "step": 16243
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.299430491392456e-06,
+      "loss": 0.2407,
+      "step": 16244
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.298279759431814e-06,
+      "loss": 0.2513,
+      "step": 16245
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.297129066062623e-06,
+      "loss": 0.2402,
+      "step": 16246
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295978411301323e-06,
+      "loss": 0.2168,
+      "step": 16247
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.294827795164346e-06,
+      "loss": 0.2274,
+      "step": 16248
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.293677217668126e-06,
+      "loss": 0.2286,
+      "step": 16249
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.292526678829102e-06,
+      "loss": 0.2211,
+      "step": 16250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.291376178663708e-06,
+      "loss": 0.244,
+      "step": 16251
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290225717188375e-06,
+      "loss": 0.2103,
+      "step": 16252
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.289075294419538e-06,
+      "loss": 0.2328,
+      "step": 16253
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.287924910373628e-06,
+      "loss": 0.2414,
+      "step": 16254
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.286774565067084e-06,
+      "loss": 0.2353,
+      "step": 16255
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285624258516328e-06,
+      "loss": 0.2303,
+      "step": 16256
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284473990737795e-06,
+      "loss": 0.2026,
+      "step": 16257
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283323761747917e-06,
+      "loss": 0.2341,
+      "step": 16258
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2821735715631205e-06,
+      "loss": 0.2166,
+      "step": 16259
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.281023420199837e-06,
+      "loss": 0.2373,
+      "step": 16260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.279873307674495e-06,
+      "loss": 0.21,
+      "step": 16261
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278723234003523e-06,
+      "loss": 0.2134,
+      "step": 16262
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.277573199203352e-06,
+      "loss": 0.2146,
+      "step": 16263
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276423203290402e-06,
+      "loss": 0.2212,
+      "step": 16264
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2752732462811025e-06,
+      "loss": 0.224,
+      "step": 16265
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.274123328191881e-06,
+      "loss": 0.2388,
+      "step": 16266
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2729734490391645e-06,
+      "loss": 0.2101,
+      "step": 16267
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.271823608839377e-06,
+      "loss": 0.2333,
+      "step": 16268
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.270673807608939e-06,
+      "loss": 0.2337,
+      "step": 16269
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269524045364275e-06,
+      "loss": 0.2084,
+      "step": 16270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268374322121811e-06,
+      "loss": 0.2244,
+      "step": 16271
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267224637897969e-06,
+      "loss": 0.2186,
+      "step": 16272
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266074992709171e-06,
+      "loss": 0.2078,
+      "step": 16273
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26492538657184e-06,
+      "loss": 0.2038,
+      "step": 16274
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263775819502393e-06,
+      "loss": 0.2195,
+      "step": 16275
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262626291517254e-06,
+      "loss": 0.2213,
+      "step": 16276
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2614768026328405e-06,
+      "loss": 0.2098,
+      "step": 16277
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.260327352865573e-06,
+      "loss": 0.233,
+      "step": 16278
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259177942231872e-06,
+      "loss": 0.2226,
+      "step": 16279
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.258028570748154e-06,
+      "loss": 0.2152,
+      "step": 16280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256879238430839e-06,
+      "loss": 0.2149,
+      "step": 16281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.255729945296339e-06,
+      "loss": 0.2198,
+      "step": 16282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.254580691361073e-06,
+      "loss": 0.2287,
+      "step": 16283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.253431476641457e-06,
+      "loss": 0.2447,
+      "step": 16284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.252282301153908e-06,
+      "loss": 0.2225,
+      "step": 16285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251133164914838e-06,
+      "loss": 0.2248,
+      "step": 16286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249984067940664e-06,
+      "loss": 0.242,
+      "step": 16287
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2488350102477975e-06,
+      "loss": 0.2446,
+      "step": 16288
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.247685991852653e-06,
+      "loss": 0.2293,
+      "step": 16289
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.246537012771646e-06,
+      "loss": 0.1977,
+      "step": 16290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.245388073021182e-06,
+      "loss": 0.2249,
+      "step": 16291
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244239172617678e-06,
+      "loss": 0.1941,
+      "step": 16292
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243090311577541e-06,
+      "loss": 0.2274,
+      "step": 16293
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241941489917188e-06,
+      "loss": 0.2213,
+      "step": 16294
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.240792707653019e-06,
+      "loss": 0.1819,
+      "step": 16295
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23964396480145e-06,
+      "loss": 0.2227,
+      "step": 16296
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.238495261378889e-06,
+      "loss": 0.2155,
+      "step": 16297
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23734659740174e-06,
+      "loss": 0.2106,
+      "step": 16298
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236197972886415e-06,
+      "loss": 0.2303,
+      "step": 16299
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23504938784932e-06,
+      "loss": 0.2108,
+      "step": 16300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23390084230686e-06,
+      "loss": 0.2184,
+      "step": 16301
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2327523362754444e-06,
+      "loss": 0.2464,
+      "step": 16302
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.231603869771474e-06,
+      "loss": 0.2368,
+      "step": 16303
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.230455442811355e-06,
+      "loss": 0.2143,
+      "step": 16304
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229307055411493e-06,
+      "loss": 0.205,
+      "step": 16305
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228158707588293e-06,
+      "loss": 0.2101,
+      "step": 16306
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.227010399358154e-06,
+      "loss": 0.2116,
+      "step": 16307
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.225862130737482e-06,
+      "loss": 0.2113,
+      "step": 16308
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224713901742675e-06,
+      "loss": 0.2167,
+      "step": 16309
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.223565712390135e-06,
+      "loss": 0.2355,
+      "step": 16310
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.222417562696264e-06,
+      "loss": 0.228,
+      "step": 16311
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.221269452677464e-06,
+      "loss": 0.2195,
+      "step": 16312
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220121382350135e-06,
+      "loss": 0.2248,
+      "step": 16313
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.218973351730671e-06,
+      "loss": 0.217,
+      "step": 16314
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 0.2383,
+      "step": 16315
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.216677409680942e-06,
+      "loss": 0.223,
+      "step": 16316
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215529498283471e-06,
+      "loss": 0.2277,
+      "step": 16317
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.214381626659462e-06,
+      "loss": 0.2255,
+      "step": 16318
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213233794825305e-06,
+      "loss": 0.2293,
+      "step": 16319
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212086002797403e-06,
+      "loss": 0.2103,
+      "step": 16320
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.210938250592143e-06,
+      "loss": 0.2508,
+      "step": 16321
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2097905382259225e-06,
+      "loss": 0.2372,
+      "step": 16322
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208642865715136e-06,
+      "loss": 0.2124,
+      "step": 16323
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2074952330761804e-06,
+      "loss": 0.2077,
+      "step": 16324
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.206347640325443e-06,
+      "loss": 0.2217,
+      "step": 16325
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.205200087479318e-06,
+      "loss": 0.1982,
+      "step": 16326
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.204052574554197e-06,
+      "loss": 0.2295,
+      "step": 16327
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202905101566471e-06,
+      "loss": 0.2288,
+      "step": 16328
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201757668532534e-06,
+      "loss": 0.2082,
+      "step": 16329
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20061027546877e-06,
+      "loss": 0.2302,
+      "step": 16330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199462922391572e-06,
+      "loss": 0.258,
+      "step": 16331
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.198315609317328e-06,
+      "loss": 0.2319,
+      "step": 16332
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197168336262426e-06,
+      "loss": 0.2142,
+      "step": 16333
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.196021103243258e-06,
+      "loss": 0.2528,
+      "step": 16334
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.194873910276205e-06,
+      "loss": 0.2346,
+      "step": 16335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193726757377656e-06,
+      "loss": 0.2101,
+      "step": 16336
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192579644563995e-06,
+      "loss": 0.2191,
+      "step": 16337
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.191432571851609e-06,
+      "loss": 0.203,
+      "step": 16338
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.190285539256883e-06,
+      "loss": 0.2196,
+      "step": 16339
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1891385467962035e-06,
+      "loss": 0.2302,
+      "step": 16340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18799159448595e-06,
+      "loss": 0.2242,
+      "step": 16341
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186844682342507e-06,
+      "loss": 0.203,
+      "step": 16342
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1856978103822575e-06,
+      "loss": 0.214,
+      "step": 16343
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.184550978621584e-06,
+      "loss": 0.2529,
+      "step": 16344
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183404187076869e-06,
+      "loss": 0.2113,
+      "step": 16345
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182257435764489e-06,
+      "loss": 0.2113,
+      "step": 16346
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.181110724700833e-06,
+      "loss": 0.2266,
+      "step": 16347
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179964053902269e-06,
+      "loss": 0.2144,
+      "step": 16348
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178817423385181e-06,
+      "loss": 0.2243,
+      "step": 16349
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1776708331659485e-06,
+      "loss": 0.2429,
+      "step": 16350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.176524283260948e-06,
+      "loss": 0.2664,
+      "step": 16351
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.175377773686561e-06,
+      "loss": 0.2245,
+      "step": 16352
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1742313044591595e-06,
+      "loss": 0.2619,
+      "step": 16353
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.17308487559512e-06,
+      "loss": 0.2221,
+      "step": 16354
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.171938487110821e-06,
+      "loss": 0.2152,
+      "step": 16355
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.170792139022637e-06,
+      "loss": 0.2091,
+      "step": 16356
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169645831346941e-06,
+      "loss": 0.2358,
+      "step": 16357
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1684995641001055e-06,
+      "loss": 0.2084,
+      "step": 16358
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167353337298506e-06,
+      "loss": 0.2345,
+      "step": 16359
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.166207150958519e-06,
+      "loss": 0.2673,
+      "step": 16360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.165061005096509e-06,
+      "loss": 0.22,
+      "step": 16361
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1639148997288524e-06,
+      "loss": 0.2136,
+      "step": 16362
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.162768834871919e-06,
+      "loss": 0.2404,
+      "step": 16363
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.161622810542078e-06,
+      "loss": 0.216,
+      "step": 16364
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1604768267557e-06,
+      "loss": 0.2265,
+      "step": 16365
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1593308835291556e-06,
+      "loss": 0.2286,
+      "step": 16366
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1581849808788116e-06,
+      "loss": 0.2211,
+      "step": 16367
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15703911882104e-06,
+      "loss": 0.2593,
+      "step": 16368
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1558932973722025e-06,
+      "loss": 0.2327,
+      "step": 16369
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15474751654867e-06,
+      "loss": 0.2277,
+      "step": 16370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153601776366808e-06,
+      "loss": 0.2176,
+      "step": 16371
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.152456076842982e-06,
+      "loss": 0.2281,
+      "step": 16372
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15131041799356e-06,
+      "loss": 0.2397,
+      "step": 16373
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.150164799834902e-06,
+      "loss": 0.2364,
+      "step": 16374
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149019222383373e-06,
+      "loss": 0.2226,
+      "step": 16375
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1478736856553375e-06,
+      "loss": 0.2092,
+      "step": 16376
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.146728189667158e-06,
+      "loss": 0.2373,
+      "step": 16377
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145582734435197e-06,
+      "loss": 0.2029,
+      "step": 16378
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1444373199758185e-06,
+      "loss": 0.2015,
+      "step": 16379
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143291946305379e-06,
+      "loss": 0.1962,
+      "step": 16380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.142146613440241e-06,
+      "loss": 0.2511,
+      "step": 16381
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.141001321396768e-06,
+      "loss": 0.1862,
+      "step": 16382
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1398560701913135e-06,
+      "loss": 0.2324,
+      "step": 16383
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.138710859840244e-06,
+      "loss": 0.2365,
+      "step": 16384
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137565690359908e-06,
+      "loss": 0.2502,
+      "step": 16385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136420561766673e-06,
+      "loss": 0.2197,
+      "step": 16386
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.135275474076887e-06,
+      "loss": 0.2025,
+      "step": 16387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.134130427306911e-06,
+      "loss": 0.213,
+      "step": 16388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1329854214731e-06,
+      "loss": 0.2211,
+      "step": 16389
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1318404565918095e-06,
+      "loss": 0.2011,
+      "step": 16390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130695532679398e-06,
+      "loss": 0.2031,
+      "step": 16391
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.129550649752213e-06,
+      "loss": 0.1902,
+      "step": 16392
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128405807826611e-06,
+      "loss": 0.2185,
+      "step": 16393
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.127261006918945e-06,
+      "loss": 0.2415,
+      "step": 16394
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.12611624704557e-06,
+      "loss": 0.2104,
+      "step": 16395
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.124971528222833e-06,
+      "loss": 0.2217,
+      "step": 16396
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123826850467088e-06,
+      "loss": 0.2315,
+      "step": 16397
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122682213794684e-06,
+      "loss": 0.2274,
+      "step": 16398
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.121537618221976e-06,
+      "loss": 0.2375,
+      "step": 16399
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.120393063765307e-06,
+      "loss": 0.1998,
+      "step": 16400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119248550441027e-06,
+      "loss": 0.2118,
+      "step": 16401
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.118104078265488e-06,
+      "loss": 0.2144,
+      "step": 16402
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116959647255034e-06,
+      "loss": 0.1826,
+      "step": 16403
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.115815257426013e-06,
+      "loss": 0.2213,
+      "step": 16404
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.114670908794772e-06,
+      "loss": 0.2141,
+      "step": 16405
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113526601377656e-06,
+      "loss": 0.2328,
+      "step": 16406
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1123823351910126e-06,
+      "loss": 0.2095,
+      "step": 16407
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.111238110251184e-06,
+      "loss": 0.2366,
+      "step": 16408
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110093926574515e-06,
+      "loss": 0.2065,
+      "step": 16409
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.10894978417735e-06,
+      "loss": 0.2377,
+      "step": 16410
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.107805683076034e-06,
+      "loss": 0.2118,
+      "step": 16411
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.106661623286903e-06,
+      "loss": 0.2162,
+      "step": 16412
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1055176048263085e-06,
+      "loss": 0.2197,
+      "step": 16413
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.104373627710582e-06,
+      "loss": 0.2094,
+      "step": 16414
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1032296919560675e-06,
+      "loss": 0.2152,
+      "step": 16415
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102085797579106e-06,
+      "loss": 0.2363,
+      "step": 16416
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100941944596035e-06,
+      "loss": 0.2122,
+      "step": 16417
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.099798133023197e-06,
+      "loss": 0.2179,
+      "step": 16418
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.098654362876927e-06,
+      "loss": 0.2563,
+      "step": 16419
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097510634173563e-06,
+      "loss": 0.2319,
+      "step": 16420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096366946929441e-06,
+      "loss": 0.2454,
+      "step": 16421
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.095223301160901e-06,
+      "loss": 0.2119,
+      "step": 16422
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.09407969688428e-06,
+      "loss": 0.2212,
+      "step": 16423
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092936134115907e-06,
+      "loss": 0.2236,
+      "step": 16424
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091792612872119e-06,
+      "loss": 0.1968,
+      "step": 16425
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.090649133169256e-06,
+      "loss": 0.2458,
+      "step": 16426
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.089505695023642e-06,
+      "loss": 0.2102,
+      "step": 16427
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.088362298451614e-06,
+      "loss": 0.209,
+      "step": 16428
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.087218943469509e-06,
+      "loss": 0.2235,
+      "step": 16429
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.08607563009365e-06,
+      "loss": 0.2121,
+      "step": 16430
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084932358340375e-06,
+      "loss": 0.2093,
+      "step": 16431
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083789128226011e-06,
+      "loss": 0.2273,
+      "step": 16432
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.082645939766888e-06,
+      "loss": 0.2333,
+      "step": 16433
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.08150279297934e-06,
+      "loss": 0.2121,
+      "step": 16434
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.080359687879689e-06,
+      "loss": 0.1999,
+      "step": 16435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079216624484267e-06,
+      "loss": 0.2212,
+      "step": 16436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0780736028094e-06,
+      "loss": 0.2308,
+      "step": 16437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.076930622871416e-06,
+      "loss": 0.2265,
+      "step": 16438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0757876846866454e-06,
+      "loss": 0.2173,
+      "step": 16439
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0746447882714066e-06,
+      "loss": 0.191,
+      "step": 16440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.073501933642028e-06,
+      "loss": 0.2375,
+      "step": 16441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.072359120814833e-06,
+      "loss": 0.2146,
+      "step": 16442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071216349806147e-06,
+      "loss": 0.2301,
+      "step": 16443
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.070073620632294e-06,
+      "loss": 0.2201,
+      "step": 16444
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.068930933309595e-06,
+      "loss": 0.2052,
+      "step": 16445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.067788287854373e-06,
+      "loss": 0.2004,
+      "step": 16446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.066645684282949e-06,
+      "loss": 0.2281,
+      "step": 16447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065503122611645e-06,
+      "loss": 0.2065,
+      "step": 16448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0643606028567794e-06,
+      "loss": 0.2523,
+      "step": 16449
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063218125034677e-06,
+      "loss": 0.2399,
+      "step": 16450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.062075689161652e-06,
+      "loss": 0.2265,
+      "step": 16451
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.060933295254027e-06,
+      "loss": 0.2261,
+      "step": 16452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.059790943328113e-06,
+      "loss": 0.2484,
+      "step": 16453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.058648633400233e-06,
+      "loss": 0.2373,
+      "step": 16454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.057506365486702e-06,
+      "loss": 0.2243,
+      "step": 16455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0563641396038374e-06,
+      "loss": 0.217,
+      "step": 16456
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.055221955767956e-06,
+      "loss": 0.2111,
+      "step": 16457
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.054079813995368e-06,
+      "loss": 0.2102,
+      "step": 16458
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0529377143023925e-06,
+      "loss": 0.237,
+      "step": 16459
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05179565670534e-06,
+      "loss": 0.2173,
+      "step": 16460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050653641220528e-06,
+      "loss": 0.228,
+      "step": 16461
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049511667864264e-06,
+      "loss": 0.228,
+      "step": 16462
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048369736652862e-06,
+      "loss": 0.2148,
+      "step": 16463
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.047227847602636e-06,
+      "loss": 0.2178,
+      "step": 16464
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046086000729896e-06,
+      "loss": 0.2387,
+      "step": 16465
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.044944196050947e-06,
+      "loss": 0.2206,
+      "step": 16466
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.043802433582103e-06,
+      "loss": 0.2106,
+      "step": 16467
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.042660713339674e-06,
+      "loss": 0.2244,
+      "step": 16468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041519035339964e-06,
+      "loss": 0.2342,
+      "step": 16469
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040377399599283e-06,
+      "loss": 0.2311,
+      "step": 16470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.03923580613394e-06,
+      "loss": 0.2181,
+      "step": 16471
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.038094254960239e-06,
+      "loss": 0.2205,
+      "step": 16472
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036952746094488e-06,
+      "loss": 0.225,
+      "step": 16473
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.035811279552991e-06,
+      "loss": 0.2233,
+      "step": 16474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.034669855352052e-06,
+      "loss": 0.242,
+      "step": 16475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0335284735079774e-06,
+      "loss": 0.2175,
+      "step": 16476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.032387134037071e-06,
+      "loss": 0.2313,
+      "step": 16477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031245836955634e-06,
+      "loss": 0.2436,
+      "step": 16478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.030104582279969e-06,
+      "loss": 0.2163,
+      "step": 16479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.028963370026376e-06,
+      "loss": 0.1976,
+      "step": 16480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027822200211156e-06,
+      "loss": 0.2355,
+      "step": 16481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026681072850614e-06,
+      "loss": 0.258,
+      "step": 16482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0255399879610455e-06,
+      "loss": 0.216,
+      "step": 16483
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0243989455587545e-06,
+      "loss": 0.2448,
+      "step": 16484
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.023257945660034e-06,
+      "loss": 0.2199,
+      "step": 16485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022116988281185e-06,
+      "loss": 0.2237,
+      "step": 16486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.020976073438503e-06,
+      "loss": 0.2319,
+      "step": 16487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0198352011482894e-06,
+      "loss": 0.2266,
+      "step": 16488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018694371426839e-06,
+      "loss": 0.2328,
+      "step": 16489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.017553584290443e-06,
+      "loss": 0.2237,
+      "step": 16490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0164128397554e-06,
+      "loss": 0.212,
+      "step": 16491
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.015272137838007e-06,
+      "loss": 0.2386,
+      "step": 16492
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.014131478554554e-06,
+      "loss": 0.2291,
+      "step": 16493
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0129908619213315e-06,
+      "loss": 0.1995,
+      "step": 16494
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011850287954637e-06,
+      "loss": 0.2086,
+      "step": 16495
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.010709756670763e-06,
+      "loss": 0.194,
+      "step": 16496
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.009569268085996e-06,
+      "loss": 0.2246,
+      "step": 16497
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008428822216631e-06,
+      "loss": 0.224,
+      "step": 16498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.007288419078957e-06,
+      "loss": 0.2242,
+      "step": 16499
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006148058689264e-06,
+      "loss": 0.2135,
+      "step": 16500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.00500774106384e-06,
+      "loss": 0.213,
+      "step": 16501
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003867466218972e-06,
+      "loss": 0.2112,
+      "step": 16502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002727234170951e-06,
+      "loss": 0.2239,
+      "step": 16503
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.001587044936061e-06,
+      "loss": 0.2354,
+      "step": 16504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.000446898530595e-06,
+      "loss": 0.2424,
+      "step": 16505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.99930679497083e-06,
+      "loss": 0.216,
+      "step": 16506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998166734273057e-06,
+      "loss": 0.2266,
+      "step": 16507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9970267164535565e-06,
+      "loss": 0.2049,
+      "step": 16508
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.995886741528617e-06,
+      "loss": 0.2355,
+      "step": 16509
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994746809514517e-06,
+      "loss": 0.243,
+      "step": 16510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993606920427543e-06,
+      "loss": 0.2168,
+      "step": 16511
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.99246707428398e-06,
+      "loss": 0.2182,
+      "step": 16512
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.991327271100101e-06,
+      "loss": 0.2007,
+      "step": 16513
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9901875108921944e-06,
+      "loss": 0.222,
+      "step": 16514
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9890477936765375e-06,
+      "loss": 0.2215,
+      "step": 16515
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.987908119469412e-06,
+      "loss": 0.2309,
+      "step": 16516
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986768488287095e-06,
+      "loss": 0.2029,
+      "step": 16517
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.985628900145869e-06,
+      "loss": 0.2414,
+      "step": 16518
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.984489355062004e-06,
+      "loss": 0.2281,
+      "step": 16519
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983349853051782e-06,
+      "loss": 0.2298,
+      "step": 16520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.982210394131481e-06,
+      "loss": 0.189,
+      "step": 16521
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.981070978317374e-06,
+      "loss": 0.2353,
+      "step": 16522
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.979931605625741e-06,
+      "loss": 0.2082,
+      "step": 16523
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978792276072852e-06,
+      "loss": 0.2291,
+      "step": 16524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.977652989674983e-06,
+      "loss": 0.2143,
+      "step": 16525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976513746448408e-06,
+      "loss": 0.1967,
+      "step": 16526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975374546409399e-06,
+      "loss": 0.2468,
+      "step": 16527
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974235389574233e-06,
+      "loss": 0.2481,
+      "step": 16528
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9730962759591745e-06,
+      "loss": 0.2404,
+      "step": 16529
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.971957205580497e-06,
+      "loss": 0.2198,
+      "step": 16530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970818178454478e-06,
+      "loss": 0.2373,
+      "step": 16531
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.969679194597376e-06,
+      "loss": 0.24,
+      "step": 16532
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.968540254025467e-06,
+      "loss": 0.2262,
+      "step": 16533
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.96740135675502e-06,
+      "loss": 0.2326,
+      "step": 16534
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966262502802299e-06,
+      "loss": 0.2174,
+      "step": 16535
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9651236921835745e-06,
+      "loss": 0.2232,
+      "step": 16536
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.963984924915111e-06,
+      "loss": 0.2216,
+      "step": 16537
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.962846201013178e-06,
+      "loss": 0.2316,
+      "step": 16538
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.961707520494041e-06,
+      "loss": 0.2091,
+      "step": 16539
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9605688833739614e-06,
+      "loss": 0.199,
+      "step": 16540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.959430289669204e-06,
+      "loss": 0.235,
+      "step": 16541
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958291739396036e-06,
+      "loss": 0.2361,
+      "step": 16542
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957153232570717e-06,
+      "loss": 0.2279,
+      "step": 16543
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956014769209516e-06,
+      "loss": 0.2186,
+      "step": 16544
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.954876349328685e-06,
+      "loss": 0.2256,
+      "step": 16545
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9537379729444935e-06,
+      "loss": 0.22,
+      "step": 16546
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952599640073194e-06,
+      "loss": 0.2169,
+      "step": 16547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951461350731054e-06,
+      "loss": 0.2147,
+      "step": 16548
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9503231049343275e-06,
+      "loss": 0.2313,
+      "step": 16549
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.94918490269928e-06,
+      "loss": 0.2174,
+      "step": 16550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.948046744042161e-06,
+      "loss": 0.2011,
+      "step": 16551
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.946908628979232e-06,
+      "loss": 0.2274,
+      "step": 16552
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.945770557526752e-06,
+      "loss": 0.2166,
+      "step": 16553
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944632529700974e-06,
+      "loss": 0.2154,
+      "step": 16554
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.943494545518159e-06,
+      "loss": 0.2155,
+      "step": 16555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942356604994553e-06,
+      "loss": 0.2005,
+      "step": 16556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.941218708146423e-06,
+      "loss": 0.2399,
+      "step": 16557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940080854990008e-06,
+      "loss": 0.2088,
+      "step": 16558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.93894304554157e-06,
+      "loss": 0.2446,
+      "step": 16559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.937805279817359e-06,
+      "loss": 0.2217,
+      "step": 16560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936667557833627e-06,
+      "loss": 0.2281,
+      "step": 16561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.935529879606629e-06,
+      "loss": 0.2375,
+      "step": 16562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9343922451526105e-06,
+      "loss": 0.2096,
+      "step": 16563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933254654487824e-06,
+      "loss": 0.2176,
+      "step": 16564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9321171076285175e-06,
+      "loss": 0.2004,
+      "step": 16565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930979604590943e-06,
+      "loss": 0.2485,
+      "step": 16566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.929842145391344e-06,
+      "loss": 0.2078,
+      "step": 16567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928704730045971e-06,
+      "loss": 0.2317,
+      "step": 16568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.92756735857107e-06,
+      "loss": 0.2164,
+      "step": 16569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926430030982887e-06,
+      "loss": 0.2441,
+      "step": 16570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9252927472976716e-06,
+      "loss": 0.2223,
+      "step": 16571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9241555075316624e-06,
+      "loss": 0.2083,
+      "step": 16572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9230183117011075e-06,
+      "loss": 0.2346,
+      "step": 16573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921881159822248e-06,
+      "loss": 0.2302,
+      "step": 16574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.920744051911329e-06,
+      "loss": 0.2023,
+      "step": 16575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919606987984593e-06,
+      "loss": 0.2321,
+      "step": 16576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.918469968058281e-06,
+      "loss": 0.2427,
+      "step": 16577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917332992148637e-06,
+      "loss": 0.2382,
+      "step": 16578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916196060271898e-06,
+      "loss": 0.2386,
+      "step": 16579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.915059172444305e-06,
+      "loss": 0.2429,
+      "step": 16580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.913922328682097e-06,
+      "loss": 0.2437,
+      "step": 16581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.912785529001516e-06,
+      "loss": 0.2287,
+      "step": 16582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.911648773418796e-06,
+      "loss": 0.2058,
+      "step": 16583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9105120619501784e-06,
+      "loss": 0.2272,
+      "step": 16584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909375394611895e-06,
+      "loss": 0.2045,
+      "step": 16585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.908238771420184e-06,
+      "loss": 0.2497,
+      "step": 16586
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907102192391282e-06,
+      "loss": 0.2105,
+      "step": 16587
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9059656575414225e-06,
+      "loss": 0.2286,
+      "step": 16588
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.904829166886844e-06,
+      "loss": 0.2325,
+      "step": 16589
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9036927204437745e-06,
+      "loss": 0.239,
+      "step": 16590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.90255631822845e-06,
+      "loss": 0.2146,
+      "step": 16591
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901419960257101e-06,
+      "loss": 0.2274,
+      "step": 16592
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900283646545961e-06,
+      "loss": 0.232,
+      "step": 16593
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.899147377111264e-06,
+      "loss": 0.2077,
+      "step": 16594
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.898011151969233e-06,
+      "loss": 0.208,
+      "step": 16595
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.896874971136104e-06,
+      "loss": 0.1741,
+      "step": 16596
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.895738834628107e-06,
+      "loss": 0.2187,
+      "step": 16597
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894602742461466e-06,
+      "loss": 0.2214,
+      "step": 16598
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.893466694652408e-06,
+      "loss": 0.2337,
+      "step": 16599
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.892330691217166e-06,
+      "loss": 0.2121,
+      "step": 16600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8911947321719665e-06,
+      "loss": 0.2384,
+      "step": 16601
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.890058817533029e-06,
+      "loss": 0.2105,
+      "step": 16602
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888922947316583e-06,
+      "loss": 0.2102,
+      "step": 16603
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.887787121538853e-06,
+      "loss": 0.2318,
+      "step": 16604
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8866513402160685e-06,
+      "loss": 0.2179,
+      "step": 16605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.885515603364443e-06,
+      "loss": 0.1948,
+      "step": 16606
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884379911000206e-06,
+      "loss": 0.2518,
+      "step": 16607
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883244263139578e-06,
+      "loss": 0.2096,
+      "step": 16608
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882108659798781e-06,
+      "loss": 0.2292,
+      "step": 16609
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.880973100994038e-06,
+      "loss": 0.2004,
+      "step": 16610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.879837586741564e-06,
+      "loss": 0.2219,
+      "step": 16611
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878702117057585e-06,
+      "loss": 0.257,
+      "step": 16612
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.877566691958315e-06,
+      "loss": 0.2399,
+      "step": 16613
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.876431311459973e-06,
+      "loss": 0.2258,
+      "step": 16614
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.87529597557878e-06,
+      "loss": 0.238,
+      "step": 16615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8741606843309505e-06,
+      "loss": 0.2027,
+      "step": 16616
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.873025437732703e-06,
+      "loss": 0.2122,
+      "step": 16617
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871890235800252e-06,
+      "loss": 0.2182,
+      "step": 16618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870755078549811e-06,
+      "loss": 0.2279,
+      "step": 16619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.869619965997598e-06,
+      "loss": 0.2234,
+      "step": 16620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868484898159829e-06,
+      "loss": 0.2158,
+      "step": 16621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.867349875052709e-06,
+      "loss": 0.2127,
+      "step": 16622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.866214896692462e-06,
+      "loss": 0.2245,
+      "step": 16623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8650799630952895e-06,
+      "loss": 0.221,
+      "step": 16624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863945074277406e-06,
+      "loss": 0.2041,
+      "step": 16625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.862810230255024e-06,
+      "loss": 0.2212,
+      "step": 16626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.861675431044354e-06,
+      "loss": 0.2284,
+      "step": 16627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.860540676661607e-06,
+      "loss": 0.2414,
+      "step": 16628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859405967122985e-06,
+      "loss": 0.2184,
+      "step": 16629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858271302444704e-06,
+      "loss": 0.2193,
+      "step": 16630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857136682642966e-06,
+      "loss": 0.2279,
+      "step": 16631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.856002107733981e-06,
+      "loss": 0.2513,
+      "step": 16632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8548675777339556e-06,
+      "loss": 0.2639,
+      "step": 16633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853733092659094e-06,
+      "loss": 0.2103,
+      "step": 16634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.852598652525601e-06,
+      "loss": 0.2128,
+      "step": 16635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851464257349685e-06,
+      "loss": 0.2027,
+      "step": 16636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8503299071475434e-06,
+      "loss": 0.2202,
+      "step": 16637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849195601935381e-06,
+      "loss": 0.2322,
+      "step": 16638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8480613417294055e-06,
+      "loss": 0.2242,
+      "step": 16639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.84692712654581e-06,
+      "loss": 0.2253,
+      "step": 16640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.845792956400803e-06,
+      "loss": 0.237,
+      "step": 16641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844658831310581e-06,
+      "loss": 0.231,
+      "step": 16642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.843524751291346e-06,
+      "loss": 0.2444,
+      "step": 16643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8423907163592975e-06,
+      "loss": 0.2359,
+      "step": 16644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841256726530631e-06,
+      "loss": 0.253,
+      "step": 16645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.840122781821548e-06,
+      "loss": 0.2108,
+      "step": 16646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.838988882248243e-06,
+      "loss": 0.207,
+      "step": 16647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837855027826915e-06,
+      "loss": 0.2307,
+      "step": 16648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8367212185737604e-06,
+      "loss": 0.2208,
+      "step": 16649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.835587454504973e-06,
+      "loss": 0.2383,
+      "step": 16650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834453735636748e-06,
+      "loss": 0.2134,
+      "step": 16651
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833320061985278e-06,
+      "loss": 0.2176,
+      "step": 16652
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832186433566756e-06,
+      "loss": 0.2141,
+      "step": 16653
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8310528503973774e-06,
+      "loss": 0.2556,
+      "step": 16654
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.829919312493334e-06,
+      "loss": 0.2236,
+      "step": 16655
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.828785819870816e-06,
+      "loss": 0.2183,
+      "step": 16656
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.827652372546014e-06,
+      "loss": 0.2127,
+      "step": 16657
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.82651897053512e-06,
+      "loss": 0.2272,
+      "step": 16658
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825385613854321e-06,
+      "loss": 0.1912,
+      "step": 16659
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8242523025198114e-06,
+      "loss": 0.218,
+      "step": 16660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823119036547772e-06,
+      "loss": 0.2453,
+      "step": 16661
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.821985815954395e-06,
+      "loss": 0.22,
+      "step": 16662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.820852640755868e-06,
+      "loss": 0.2326,
+      "step": 16663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.819719510968375e-06,
+      "loss": 0.2242,
+      "step": 16664
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8185864266081e-06,
+      "loss": 0.2023,
+      "step": 16665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8174533876912305e-06,
+      "loss": 0.214,
+      "step": 16666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8163203942339526e-06,
+      "loss": 0.2258,
+      "step": 16667
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.815187446252447e-06,
+      "loss": 0.2278,
+      "step": 16668
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.814054543762897e-06,
+      "loss": 0.2107,
+      "step": 16669
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.812921686781486e-06,
+      "loss": 0.2023,
+      "step": 16670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.811788875324398e-06,
+      "loss": 0.1946,
+      "step": 16671
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.81065610940781e-06,
+      "loss": 0.1899,
+      "step": 16672
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.809523389047902e-06,
+      "loss": 0.2216,
+      "step": 16673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808390714260857e-06,
+      "loss": 0.223,
+      "step": 16674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8072580850628536e-06,
+      "loss": 0.2252,
+      "step": 16675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806125501470074e-06,
+      "loss": 0.2189,
+      "step": 16676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8049929634986865e-06,
+      "loss": 0.228,
+      "step": 16677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803860471164878e-06,
+      "loss": 0.2412,
+      "step": 16678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.802728024484817e-06,
+      "loss": 0.2248,
+      "step": 16679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801595623474684e-06,
+      "loss": 0.2066,
+      "step": 16680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800463268150652e-06,
+      "loss": 0.2256,
+      "step": 16681
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.799330958528898e-06,
+      "loss": 0.2159,
+      "step": 16682
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7981986946255975e-06,
+      "loss": 0.234,
+      "step": 16683
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.79706647645692e-06,
+      "loss": 0.2051,
+      "step": 16684
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795934304039037e-06,
+      "loss": 0.1944,
+      "step": 16685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7948021773881235e-06,
+      "loss": 0.2364,
+      "step": 16686
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793670096520351e-06,
+      "loss": 0.2104,
+      "step": 16687
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.79253806145189e-06,
+      "loss": 0.2337,
+      "step": 16688
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.791406072198912e-06,
+      "loss": 0.2307,
+      "step": 16689
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.790274128777583e-06,
+      "loss": 0.238,
+      "step": 16690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789142231204071e-06,
+      "loss": 0.2197,
+      "step": 16691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.788010379494545e-06,
+      "loss": 0.2149,
+      "step": 16692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.786878573665174e-06,
+      "loss": 0.227,
+      "step": 16693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785746813732128e-06,
+      "loss": 0.2002,
+      "step": 16694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7846150997115645e-06,
+      "loss": 0.2383,
+      "step": 16695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7834834316196555e-06,
+      "loss": 0.2461,
+      "step": 16696
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7823518094725636e-06,
+      "loss": 0.2146,
+      "step": 16697
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.781220233286454e-06,
+      "loss": 0.2239,
+      "step": 16698
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78008870307749e-06,
+      "loss": 0.2061,
+      "step": 16699
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7789572188618326e-06,
+      "loss": 0.2263,
+      "step": 16700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.777825780655645e-06,
+      "loss": 0.2179,
+      "step": 16701
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.776694388475093e-06,
+      "loss": 0.2289,
+      "step": 16702
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.775563042336331e-06,
+      "loss": 0.2427,
+      "step": 16703
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7744317422555185e-06,
+      "loss": 0.2268,
+      "step": 16704
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773300488248819e-06,
+      "loss": 0.2262,
+      "step": 16705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772169280332395e-06,
+      "loss": 0.2336,
+      "step": 16706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.771038118522395e-06,
+      "loss": 0.2051,
+      "step": 16707
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769907002834983e-06,
+      "loss": 0.2269,
+      "step": 16708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768775933286314e-06,
+      "loss": 0.2312,
+      "step": 16709
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.767644909892548e-06,
+      "loss": 0.2405,
+      "step": 16710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.766513932669835e-06,
+      "loss": 0.2289,
+      "step": 16711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765383001634331e-06,
+      "loss": 0.2288,
+      "step": 16712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.764252116802192e-06,
+      "loss": 0.2081,
+      "step": 16713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763121278189573e-06,
+      "loss": 0.2091,
+      "step": 16714
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.761990485812628e-06,
+      "loss": 0.2112,
+      "step": 16715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.760859739687502e-06,
+      "loss": 0.2195,
+      "step": 16716
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759729039830353e-06,
+      "loss": 0.2225,
+      "step": 16717
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758598386257328e-06,
+      "loss": 0.2307,
+      "step": 16718
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.757467778984579e-06,
+      "loss": 0.258,
+      "step": 16719
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.756337218028256e-06,
+      "loss": 0.2275,
+      "step": 16720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755206703404508e-06,
+      "loss": 0.2119,
+      "step": 16721
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.754076235129486e-06,
+      "loss": 0.2289,
+      "step": 16722
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75294581321933e-06,
+      "loss": 0.2532,
+      "step": 16723
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751815437690193e-06,
+      "loss": 0.2472,
+      "step": 16724
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.750685108558221e-06,
+      "loss": 0.2044,
+      "step": 16725
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749554825839559e-06,
+      "loss": 0.2036,
+      "step": 16726
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.748424589550352e-06,
+      "loss": 0.2263,
+      "step": 16727
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.747294399706743e-06,
+      "loss": 0.2154,
+      "step": 16728
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746164256324881e-06,
+      "loss": 0.2303,
+      "step": 16729
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.745034159420899e-06,
+      "loss": 0.2442,
+      "step": 16730
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743904109010946e-06,
+      "loss": 0.2344,
+      "step": 16731
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7427741051111625e-06,
+      "loss": 0.2263,
+      "step": 16732
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741644147737693e-06,
+      "loss": 0.2053,
+      "step": 16733
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.740514236906671e-06,
+      "loss": 0.2457,
+      "step": 16734
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.739384372634241e-06,
+      "loss": 0.2311,
+      "step": 16735
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7382545549365385e-06,
+      "loss": 0.2449,
+      "step": 16736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737124783829707e-06,
+      "loss": 0.2054,
+      "step": 16737
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735995059329881e-06,
+      "loss": 0.2194,
+      "step": 16738
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7348653814531975e-06,
+      "loss": 0.2536,
+      "step": 16739
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.733735750215792e-06,
+      "loss": 0.2018,
+      "step": 16740
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732606165633802e-06,
+      "loss": 0.2126,
+      "step": 16741
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.731476627723366e-06,
+      "loss": 0.2189,
+      "step": 16742
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.730347136500608e-06,
+      "loss": 0.2525,
+      "step": 16743
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729217691981672e-06,
+      "loss": 0.2208,
+      "step": 16744
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.728088294182684e-06,
+      "loss": 0.2202,
+      "step": 16745
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.72695894311978e-06,
+      "loss": 0.2158,
+      "step": 16746
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.72582963880909e-06,
+      "loss": 0.2531,
+      "step": 16747
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7247003812667465e-06,
+      "loss": 0.244,
+      "step": 16748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.723571170508883e-06,
+      "loss": 0.1965,
+      "step": 16749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.722442006551621e-06,
+      "loss": 0.2167,
+      "step": 16750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.721312889411095e-06,
+      "loss": 0.2194,
+      "step": 16751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7201838191034315e-06,
+      "loss": 0.2273,
+      "step": 16752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.719054795644759e-06,
+      "loss": 0.2169,
+      "step": 16753
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.717925819051208e-06,
+      "loss": 0.2311,
+      "step": 16754
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.716796889338901e-06,
+      "loss": 0.2346,
+      "step": 16755
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.715668006523963e-06,
+      "loss": 0.2415,
+      "step": 16756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.714539170622519e-06,
+      "loss": 0.2331,
+      "step": 16757
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7134103816506935e-06,
+      "loss": 0.248,
+      "step": 16758
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7122816396246115e-06,
+      "loss": 0.2336,
+      "step": 16759
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.711152944560398e-06,
+      "loss": 0.2168,
+      "step": 16760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.710024296474169e-06,
+      "loss": 0.2352,
+      "step": 16761
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.708895695382051e-06,
+      "loss": 0.2356,
+      "step": 16762
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7077671413001644e-06,
+      "loss": 0.2088,
+      "step": 16763
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.706638634244629e-06,
+      "loss": 0.2249,
+      "step": 16764
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.705510174231566e-06,
+      "loss": 0.1965,
+      "step": 16765
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70438176127709e-06,
+      "loss": 0.232,
+      "step": 16766
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.703253395397323e-06,
+      "loss": 0.2276,
+      "step": 16767
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.702125076608385e-06,
+      "loss": 0.2167,
+      "step": 16768
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700996804926388e-06,
+      "loss": 0.2222,
+      "step": 16769
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699868580367447e-06,
+      "loss": 0.2159,
+      "step": 16770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.698740402947681e-06,
+      "loss": 0.2133,
+      "step": 16771
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697612272683207e-06,
+      "loss": 0.2444,
+      "step": 16772
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.696484189590136e-06,
+      "loss": 0.2479,
+      "step": 16773
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6953561536845805e-06,
+      "loss": 0.2357,
+      "step": 16774
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694228164982654e-06,
+      "loss": 0.2055,
+      "step": 16775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693100223500474e-06,
+      "loss": 0.1966,
+      "step": 16776
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.691972329254144e-06,
+      "loss": 0.2216,
+      "step": 16777
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6908444822597805e-06,
+      "loss": 0.2222,
+      "step": 16778
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.689716682533492e-06,
+      "loss": 0.2194,
+      "step": 16779
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6885889300913865e-06,
+      "loss": 0.2347,
+      "step": 16780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687461224949579e-06,
+      "loss": 0.2229,
+      "step": 16781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.686333567124169e-06,
+      "loss": 0.211,
+      "step": 16782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.685205956631271e-06,
+      "loss": 0.2034,
+      "step": 16783
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6840783934869855e-06,
+      "loss": 0.268,
+      "step": 16784
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682950877707421e-06,
+      "loss": 0.2018,
+      "step": 16785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6818234093086866e-06,
+      "loss": 0.1858,
+      "step": 16786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680695988306883e-06,
+      "loss": 0.2079,
+      "step": 16787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.679568614718118e-06,
+      "loss": 0.2443,
+      "step": 16788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6784412885584905e-06,
+      "loss": 0.2091,
+      "step": 16789
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6773140098441055e-06,
+      "loss": 0.2349,
+      "step": 16790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.676186778591066e-06,
+      "loss": 0.2202,
+      "step": 16791
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.675059594815472e-06,
+      "loss": 0.2218,
+      "step": 16792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673932458533426e-06,
+      "loss": 0.2206,
+      "step": 16793
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.672805369761029e-06,
+      "loss": 0.2122,
+      "step": 16794
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.671678328514376e-06,
+      "loss": 0.2401,
+      "step": 16795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670551334809566e-06,
+      "loss": 0.2217,
+      "step": 16796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6694243886627e-06,
+      "loss": 0.2315,
+      "step": 16797
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.668297490089874e-06,
+      "loss": 0.2154,
+      "step": 16798
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6671706391071854e-06,
+      "loss": 0.2459,
+      "step": 16799
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.666043835730729e-06,
+      "loss": 0.2297,
+      "step": 16800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6649170799766e-06,
+      "loss": 0.2419,
+      "step": 16801
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663790371860894e-06,
+      "loss": 0.2121,
+      "step": 16802
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.662663711399705e-06,
+      "loss": 0.2166,
+      "step": 16803
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.661537098609128e-06,
+      "loss": 0.1896,
+      "step": 16804
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.660410533505251e-06,
+      "loss": 0.2147,
+      "step": 16805
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6592840161041685e-06,
+      "loss": 0.2046,
+      "step": 16806
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.658157546421973e-06,
+      "loss": 0.2223,
+      "step": 16807
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657031124474756e-06,
+      "loss": 0.2317,
+      "step": 16808
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655904750278602e-06,
+      "loss": 0.2109,
+      "step": 16809
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6547784238496016e-06,
+      "loss": 0.226,
+      "step": 16810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653652145203847e-06,
+      "loss": 0.2086,
+      "step": 16811
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6525259143574225e-06,
+      "loss": 0.2055,
+      "step": 16812
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651399731326416e-06,
+      "loss": 0.2262,
+      "step": 16813
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650273596126915e-06,
+      "loss": 0.2419,
+      "step": 16814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649147508775007e-06,
+      "loss": 0.2374,
+      "step": 16815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.648021469286772e-06,
+      "loss": 0.2399,
+      "step": 16816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.646895477678297e-06,
+      "loss": 0.2102,
+      "step": 16817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645769533965667e-06,
+      "loss": 0.2487,
+      "step": 16818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644643638164962e-06,
+      "loss": 0.2218,
+      "step": 16819
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.643517790292271e-06,
+      "loss": 0.2033,
+      "step": 16820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.642391990363668e-06,
+      "loss": 0.2493,
+      "step": 16821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.641266238395238e-06,
+      "loss": 0.2187,
+      "step": 16822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.640140534403059e-06,
+      "loss": 0.2174,
+      "step": 16823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639014878403209e-06,
+      "loss": 0.219,
+      "step": 16824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637889270411772e-06,
+      "loss": 0.236,
+      "step": 16825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.636763710444822e-06,
+      "loss": 0.2456,
+      "step": 16826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.635638198518443e-06,
+      "loss": 0.2197,
+      "step": 16827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634512734648702e-06,
+      "loss": 0.2519,
+      "step": 16828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633387318851683e-06,
+      "loss": 0.218,
+      "step": 16829
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6322619511434575e-06,
+      "loss": 0.2247,
+      "step": 16830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6311366315401024e-06,
+      "loss": 0.2108,
+      "step": 16831
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.630011360057691e-06,
+      "loss": 0.1944,
+      "step": 16832
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628886136712296e-06,
+      "loss": 0.2029,
+      "step": 16833
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627760961519993e-06,
+      "loss": 0.2137,
+      "step": 16834
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6266358344968504e-06,
+      "loss": 0.23,
+      "step": 16835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.62551075565894e-06,
+      "loss": 0.2311,
+      "step": 16836
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.624385725022333e-06,
+      "loss": 0.2313,
+      "step": 16837
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6232607426031026e-06,
+      "loss": 0.1897,
+      "step": 16838
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622135808417313e-06,
+      "loss": 0.206,
+      "step": 16839
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6210109224810345e-06,
+      "loss": 0.2191,
+      "step": 16840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.619886084810336e-06,
+      "loss": 0.2335,
+      "step": 16841
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6187612954212845e-06,
+      "loss": 0.2265,
+      "step": 16842
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.617636554329948e-06,
+      "loss": 0.2445,
+      "step": 16843
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.616511861552388e-06,
+      "loss": 0.1992,
+      "step": 16844
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.615387217104674e-06,
+      "loss": 0.2019,
+      "step": 16845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.614262621002867e-06,
+      "loss": 0.2122,
+      "step": 16846
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.613138073263039e-06,
+      "loss": 0.2228,
+      "step": 16847
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.61201357390124e-06,
+      "loss": 0.2148,
+      "step": 16848
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.610889122933543e-06,
+      "loss": 0.2305,
+      "step": 16849
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.609764720376002e-06,
+      "loss": 0.2229,
+      "step": 16850
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608640366244683e-06,
+      "loss": 0.2455,
+      "step": 16851
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6075160605556435e-06,
+      "loss": 0.2386,
+      "step": 16852
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6063918033249455e-06,
+      "loss": 0.2269,
+      "step": 16853
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60526759456865e-06,
+      "loss": 0.2061,
+      "step": 16854
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.604143434302807e-06,
+      "loss": 0.221,
+      "step": 16855
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.603019322543482e-06,
+      "loss": 0.2163,
+      "step": 16856
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60189525930673e-06,
+      "loss": 0.2107,
+      "step": 16857
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.600771244608604e-06,
+      "loss": 0.2025,
+      "step": 16858
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599647278465164e-06,
+      "loss": 0.2311,
+      "step": 16859
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598523360892464e-06,
+      "loss": 0.221,
+      "step": 16860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597399491906555e-06,
+      "loss": 0.2259,
+      "step": 16861
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.596275671523491e-06,
+      "loss": 0.1904,
+      "step": 16862
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595151899759325e-06,
+      "loss": 0.21,
+      "step": 16863
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.594028176630109e-06,
+      "loss": 0.1854,
+      "step": 16864
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592904502151899e-06,
+      "loss": 0.2097,
+      "step": 16865
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.591780876340738e-06,
+      "loss": 0.2102,
+      "step": 16866
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5906572992126795e-06,
+      "loss": 0.2335,
+      "step": 16867
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589533770783774e-06,
+      "loss": 0.2024,
+      "step": 16868
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588410291070066e-06,
+      "loss": 0.2337,
+      "step": 16869
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.587286860087611e-06,
+      "loss": 0.2164,
+      "step": 16870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586163477852447e-06,
+      "loss": 0.1958,
+      "step": 16871
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585040144380626e-06,
+      "loss": 0.2176,
+      "step": 16872
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.583916859688191e-06,
+      "loss": 0.2285,
+      "step": 16873
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5827936237911925e-06,
+      "loss": 0.2386,
+      "step": 16874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.581670436705667e-06,
+      "loss": 0.222,
+      "step": 16875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580547298447661e-06,
+      "loss": 0.256,
+      "step": 16876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.579424209033221e-06,
+      "loss": 0.218,
+      "step": 16877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.578301168478384e-06,
+      "loss": 0.2233,
+      "step": 16878
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577178176799194e-06,
+      "loss": 0.1992,
+      "step": 16879
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.576055234011691e-06,
+      "loss": 0.2096,
+      "step": 16880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574932340131917e-06,
+      "loss": 0.2248,
+      "step": 16881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.573809495175913e-06,
+      "loss": 0.2319,
+      "step": 16882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572686699159713e-06,
+      "loss": 0.2468,
+      "step": 16883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.571563952099356e-06,
+      "loss": 0.2257,
+      "step": 16884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5704412540108816e-06,
+      "loss": 0.239,
+      "step": 16885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569318604910327e-06,
+      "loss": 0.239,
+      "step": 16886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568196004813727e-06,
+      "loss": 0.2393,
+      "step": 16887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.567073453737116e-06,
+      "loss": 0.2221,
+      "step": 16888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.565950951696527e-06,
+      "loss": 0.2164,
+      "step": 16889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.564828498707996e-06,
+      "loss": 0.2158,
+      "step": 16890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563706094787555e-06,
+      "loss": 0.2025,
+      "step": 16891
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.562583739951238e-06,
+      "loss": 0.2136,
+      "step": 16892
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.56146143421508e-06,
+      "loss": 0.2352,
+      "step": 16893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.560339177595103e-06,
+      "loss": 0.2538,
+      "step": 16894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559216970107345e-06,
+      "loss": 0.2402,
+      "step": 16895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.558094811767832e-06,
+      "loss": 0.2191,
+      "step": 16896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556972702592595e-06,
+      "loss": 0.2343,
+      "step": 16897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555850642597663e-06,
+      "loss": 0.2246,
+      "step": 16898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.55472863179906e-06,
+      "loss": 0.2346,
+      "step": 16899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.553606670212819e-06,
+      "loss": 0.2395,
+      "step": 16900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552484757854958e-06,
+      "loss": 0.2246,
+      "step": 16901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.551362894741505e-06,
+      "loss": 0.2283,
+      "step": 16902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.550241080888488e-06,
+      "loss": 0.2204,
+      "step": 16903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549119316311929e-06,
+      "loss": 0.2234,
+      "step": 16904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547997601027851e-06,
+      "loss": 0.2057,
+      "step": 16905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546875935052277e-06,
+      "loss": 0.2401,
+      "step": 16906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545754318401228e-06,
+      "loss": 0.2077,
+      "step": 16907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.544632751090727e-06,
+      "loss": 0.2528,
+      "step": 16908
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5435112331367955e-06,
+      "loss": 0.2221,
+      "step": 16909
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54238976455545e-06,
+      "loss": 0.2062,
+      "step": 16910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54126834536271e-06,
+      "loss": 0.2082,
+      "step": 16911
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.540146975574595e-06,
+      "loss": 0.249,
+      "step": 16912
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5390256552071275e-06,
+      "loss": 0.2199,
+      "step": 16913
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537904384276314e-06,
+      "loss": 0.2113,
+      "step": 16914
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536783162798177e-06,
+      "loss": 0.2057,
+      "step": 16915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.535661990788733e-06,
+      "loss": 0.2041,
+      "step": 16916
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.534540868263994e-06,
+      "loss": 0.2237,
+      "step": 16917
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533419795239976e-06,
+      "loss": 0.2218,
+      "step": 16918
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.53229877173269e-06,
+      "loss": 0.2083,
+      "step": 16919
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531177797758155e-06,
+      "loss": 0.2402,
+      "step": 16920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.530056873332374e-06,
+      "loss": 0.2115,
+      "step": 16921
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.528935998471364e-06,
+      "loss": 0.2315,
+      "step": 16922
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.527815173191134e-06,
+      "loss": 0.2118,
+      "step": 16923
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.526694397507696e-06,
+      "loss": 0.2188,
+      "step": 16924
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5255736714370586e-06,
+      "loss": 0.2183,
+      "step": 16925
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524452994995229e-06,
+      "loss": 0.2372,
+      "step": 16926
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.523332368198216e-06,
+      "loss": 0.2109,
+      "step": 16927
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522211791062023e-06,
+      "loss": 0.2282,
+      "step": 16928
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52109126360266e-06,
+      "loss": 0.2076,
+      "step": 16929
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.519970785836131e-06,
+      "loss": 0.1924,
+      "step": 16930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518850357778443e-06,
+      "loss": 0.2318,
+      "step": 16931
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517729979445602e-06,
+      "loss": 0.2104,
+      "step": 16932
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516609650853605e-06,
+      "loss": 0.2151,
+      "step": 16933
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.515489372018458e-06,
+      "loss": 0.2454,
+      "step": 16934
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5143691429561654e-06,
+      "loss": 0.218,
+      "step": 16935
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513248963682727e-06,
+      "loss": 0.2596,
+      "step": 16936
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.512128834214142e-06,
+      "loss": 0.2376,
+      "step": 16937
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.51100875456641e-06,
+      "loss": 0.2299,
+      "step": 16938
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.509888724755538e-06,
+      "loss": 0.2216,
+      "step": 16939
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.508768744797513e-06,
+      "loss": 0.209,
+      "step": 16940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507648814708339e-06,
+      "loss": 0.2434,
+      "step": 16941
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.506528934504013e-06,
+      "loss": 0.2317,
+      "step": 16942
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50540910420053e-06,
+      "loss": 0.2171,
+      "step": 16943
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.504289323813886e-06,
+      "loss": 0.2291,
+      "step": 16944
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.503169593360077e-06,
+      "loss": 0.2028,
+      "step": 16945
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.502049912855095e-06,
+      "loss": 0.2043,
+      "step": 16946
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500930282314936e-06,
+      "loss": 0.2181,
+      "step": 16947
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.499810701755595e-06,
+      "loss": 0.2335,
+      "step": 16948
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.498691171193058e-06,
+      "loss": 0.2174,
+      "step": 16949
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49757169064332e-06,
+      "loss": 0.231,
+      "step": 16950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4964522601223704e-06,
+      "loss": 0.2606,
+      "step": 16951
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495332879646202e-06,
+      "loss": 0.1954,
+      "step": 16952
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.494213549230803e-06,
+      "loss": 0.2421,
+      "step": 16953
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493094268892161e-06,
+      "loss": 0.217,
+      "step": 16954
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491975038646261e-06,
+      "loss": 0.205,
+      "step": 16955
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490855858509093e-06,
+      "loss": 0.2224,
+      "step": 16956
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.489736728496644e-06,
+      "loss": 0.2123,
+      "step": 16957
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.488617648624898e-06,
+      "loss": 0.2162,
+      "step": 16958
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487498618909845e-06,
+      "loss": 0.233,
+      "step": 16959
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.486379639367461e-06,
+      "loss": 0.2165,
+      "step": 16960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.485260710013734e-06,
+      "loss": 0.2214,
+      "step": 16961
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484141830864648e-06,
+      "loss": 0.2095,
+      "step": 16962
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483023001936184e-06,
+      "loss": 0.2397,
+      "step": 16963
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4819042232443244e-06,
+      "loss": 0.2344,
+      "step": 16964
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.480785494805046e-06,
+      "loss": 0.2282,
+      "step": 16965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479666816634336e-06,
+      "loss": 0.212,
+      "step": 16966
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.478548188748164e-06,
+      "loss": 0.218,
+      "step": 16967
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477429611162513e-06,
+      "loss": 0.2293,
+      "step": 16968
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476311083893363e-06,
+      "loss": 0.2594,
+      "step": 16969
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4751926069566915e-06,
+      "loss": 0.2185,
+      "step": 16970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474074180368471e-06,
+      "loss": 0.1935,
+      "step": 16971
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.472955804144676e-06,
+      "loss": 0.2322,
+      "step": 16972
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4718374783012865e-06,
+      "loss": 0.2074,
+      "step": 16973
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.470719202854274e-06,
+      "loss": 0.2167,
+      "step": 16974
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.469600977819615e-06,
+      "loss": 0.2046,
+      "step": 16975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468482803213277e-06,
+      "loss": 0.2201,
+      "step": 16976
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.467364679051236e-06,
+      "loss": 0.218,
+      "step": 16977
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.46624660534946e-06,
+      "loss": 0.2267,
+      "step": 16978
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4651285821239276e-06,
+      "loss": 0.2201,
+      "step": 16979
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464010609390597e-06,
+      "loss": 0.2229,
+      "step": 16980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462892687165443e-06,
+      "loss": 0.2281,
+      "step": 16981
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.461774815464438e-06,
+      "loss": 0.2054,
+      "step": 16982
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460656994303542e-06,
+      "loss": 0.2505,
+      "step": 16983
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.459539223698728e-06,
+      "loss": 0.1936,
+      "step": 16984
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458421503665958e-06,
+      "loss": 0.2392,
+      "step": 16985
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4573038342212e-06,
+      "loss": 0.2035,
+      "step": 16986
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456186215380421e-06,
+      "loss": 0.2141,
+      "step": 16987
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45506864715958e-06,
+      "loss": 0.2301,
+      "step": 16988
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.2214,
+      "step": 16989
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452833662641573e-06,
+      "loss": 0.2149,
+      "step": 16990
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451716246376334e-06,
+      "loss": 0.2119,
+      "step": 16991
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450598880794885e-06,
+      "loss": 0.2302,
+      "step": 16992
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.449481565913185e-06,
+      "loss": 0.2187,
+      "step": 16993
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448364301747193e-06,
+      "loss": 0.2318,
+      "step": 16994
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.44724708831287e-06,
+      "loss": 0.2137,
+      "step": 16995
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446129925626174e-06,
+      "loss": 0.1946,
+      "step": 16996
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445012813703064e-06,
+      "loss": 0.2016,
+      "step": 16997
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.443895752559498e-06,
+      "loss": 0.2316,
+      "step": 16998
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442778742211426e-06,
+      "loss": 0.2199,
+      "step": 16999
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.441661782674809e-06,
+      "loss": 0.2222,
+      "step": 17000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4405448739656e-06,
+      "loss": 0.2311,
+      "step": 17001
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439428016099752e-06,
+      "loss": 0.223,
+      "step": 17002
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438311209093222e-06,
+      "loss": 0.1932,
+      "step": 17003
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4371944529619564e-06,
+      "loss": 0.2017,
+      "step": 17004
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.436077747721916e-06,
+      "loss": 0.229,
+      "step": 17005
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434961093389042e-06,
+      "loss": 0.1889,
+      "step": 17006
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.433844489979289e-06,
+      "loss": 0.2206,
+      "step": 17007
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.432727937508605e-06,
+      "loss": 0.2273,
+      "step": 17008
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431611435992942e-06,
+      "loss": 0.2025,
+      "step": 17009
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.430494985448246e-06,
+      "loss": 0.2271,
+      "step": 17010
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429378585890463e-06,
+      "loss": 0.2466,
+      "step": 17011
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.428262237335543e-06,
+      "loss": 0.2094,
+      "step": 17012
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42714593979943e-06,
+      "loss": 0.2498,
+      "step": 17013
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426029693298073e-06,
+      "loss": 0.2127,
+      "step": 17014
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.424913497847409e-06,
+      "loss": 0.2416,
+      "step": 17015
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4237973534633866e-06,
+      "loss": 0.2232,
+      "step": 17016
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.422681260161949e-06,
+      "loss": 0.2194,
+      "step": 17017
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.421565217959041e-06,
+      "loss": 0.2227,
+      "step": 17018
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4204492268705974e-06,
+      "loss": 0.2175,
+      "step": 17019
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419333286912562e-06,
+      "loss": 0.2121,
+      "step": 17020
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418217398100879e-06,
+      "loss": 0.2086,
+      "step": 17021
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.417101560451481e-06,
+      "loss": 0.2182,
+      "step": 17022
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.41598577398031e-06,
+      "loss": 0.2063,
+      "step": 17023
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4148700387033035e-06,
+      "loss": 0.2277,
+      "step": 17024
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413754354636401e-06,
+      "loss": 0.229,
+      "step": 17025
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.412638721795536e-06,
+      "loss": 0.2247,
+      "step": 17026
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.411523140196645e-06,
+      "loss": 0.2138,
+      "step": 17027
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.410407609855662e-06,
+      "loss": 0.2176,
+      "step": 17028
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409292130788525e-06,
+      "loss": 0.2027,
+      "step": 17029
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4081767030111645e-06,
+      "loss": 0.2175,
+      "step": 17030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.407061326539513e-06,
+      "loss": 0.2276,
+      "step": 17031
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.405946001389507e-06,
+      "loss": 0.236,
+      "step": 17032
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.404830727577072e-06,
+      "loss": 0.2412,
+      "step": 17033
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403715505118139e-06,
+      "loss": 0.222,
+      "step": 17034
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.402600334028639e-06,
+      "loss": 0.217,
+      "step": 17035
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.401485214324503e-06,
+      "loss": 0.216,
+      "step": 17036
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.400370146021662e-06,
+      "loss": 0.2238,
+      "step": 17037
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399255129136035e-06,
+      "loss": 0.2191,
+      "step": 17038
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.398140163683555e-06,
+      "loss": 0.205,
+      "step": 17039
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3970252496801475e-06,
+      "loss": 0.2189,
+      "step": 17040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39591038714174e-06,
+      "loss": 0.2067,
+      "step": 17041
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.394795576084252e-06,
+      "loss": 0.2372,
+      "step": 17042
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39368081652361e-06,
+      "loss": 0.2239,
+      "step": 17043
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.392566108475738e-06,
+      "loss": 0.2309,
+      "step": 17044
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.391451451956561e-06,
+      "loss": 0.223,
+      "step": 17045
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390336846981995e-06,
+      "loss": 0.1995,
+      "step": 17046
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389222293567964e-06,
+      "loss": 0.2083,
+      "step": 17047
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.388107791730389e-06,
+      "loss": 0.2291,
+      "step": 17048
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386993341485188e-06,
+      "loss": 0.2518,
+      "step": 17049
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.38587894284828e-06,
+      "loss": 0.2127,
+      "step": 17050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384764595835584e-06,
+      "loss": 0.211,
+      "step": 17051
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.383650300463017e-06,
+      "loss": 0.2373,
+      "step": 17052
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3825360567464976e-06,
+      "loss": 0.2264,
+      "step": 17053
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381421864701938e-06,
+      "loss": 0.2201,
+      "step": 17054
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.380307724345254e-06,
+      "loss": 0.2163,
+      "step": 17055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.379193635692362e-06,
+      "loss": 0.219,
+      "step": 17056
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.378079598759177e-06,
+      "loss": 0.2134,
+      "step": 17057
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37696561356161e-06,
+      "loss": 0.238,
+      "step": 17058
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.375851680115572e-06,
+      "loss": 0.2202,
+      "step": 17059
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.374737798436975e-06,
+      "loss": 0.2433,
+      "step": 17060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373623968541729e-06,
+      "loss": 0.239,
+      "step": 17061
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372510190445746e-06,
+      "loss": 0.2165,
+      "step": 17062
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.371396464164934e-06,
+      "loss": 0.2172,
+      "step": 17063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370282789715203e-06,
+      "loss": 0.2125,
+      "step": 17064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369169167112459e-06,
+      "loss": 0.2151,
+      "step": 17065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3680555963726085e-06,
+      "loss": 0.2245,
+      "step": 17066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.36694207751156e-06,
+      "loss": 0.2116,
+      "step": 17067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365828610545217e-06,
+      "loss": 0.2317,
+      "step": 17068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364715195489487e-06,
+      "loss": 0.2159,
+      "step": 17069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.363601832360271e-06,
+      "loss": 0.2166,
+      "step": 17070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362488521173478e-06,
+      "loss": 0.2341,
+      "step": 17071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361375261945002e-06,
+      "loss": 0.2276,
+      "step": 17072
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.360262054690748e-06,
+      "loss": 0.2096,
+      "step": 17073
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3591488994266194e-06,
+      "loss": 0.2203,
+      "step": 17074
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.358035796168514e-06,
+      "loss": 0.2146,
+      "step": 17075
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356922744932335e-06,
+      "loss": 0.2218,
+      "step": 17076
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.355809745733977e-06,
+      "loss": 0.2229,
+      "step": 17077
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35469679858934e-06,
+      "loss": 0.2311,
+      "step": 17078
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353583903514322e-06,
+      "loss": 0.2385,
+      "step": 17079
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.352471060524821e-06,
+      "loss": 0.2333,
+      "step": 17080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351358269636727e-06,
+      "loss": 0.2327,
+      "step": 17081
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350245530865942e-06,
+      "loss": 0.2229,
+      "step": 17082
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.349132844228355e-06,
+      "loss": 0.2266,
+      "step": 17083
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348020209739867e-06,
+      "loss": 0.2044,
+      "step": 17084
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.346907627416362e-06,
+      "loss": 0.237,
+      "step": 17085
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.345795097273737e-06,
+      "loss": 0.2379,
+      "step": 17086
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.344682619327883e-06,
+      "loss": 0.2454,
+      "step": 17087
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.343570193594688e-06,
+      "loss": 0.2326,
+      "step": 17088
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342457820090045e-06,
+      "loss": 0.2215,
+      "step": 17089
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.341345498829843e-06,
+      "loss": 0.2406,
+      "step": 17090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340233229829969e-06,
+      "loss": 0.2145,
+      "step": 17091
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.339121013106313e-06,
+      "loss": 0.2317,
+      "step": 17092
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338008848674759e-06,
+      "loss": 0.2147,
+      "step": 17093
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336896736551194e-06,
+      "loss": 0.2169,
+      "step": 17094
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.335784676751504e-06,
+      "loss": 0.2236,
+      "step": 17095
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.334672669291577e-06,
+      "loss": 0.2077,
+      "step": 17096
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333560714187293e-06,
+      "loss": 0.2215,
+      "step": 17097
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332448811454535e-06,
+      "loss": 0.229,
+      "step": 17098
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331336961109185e-06,
+      "loss": 0.2376,
+      "step": 17099
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.330225163167125e-06,
+      "loss": 0.2392,
+      "step": 17100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.32911341764424e-06,
+      "loss": 0.2142,
+      "step": 17101
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3280017245564025e-06,
+      "loss": 0.2499,
+      "step": 17102
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326890083919501e-06,
+      "loss": 0.2201,
+      "step": 17103
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.325778495749408e-06,
+      "loss": 0.2073,
+      "step": 17104
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324666960062004e-06,
+      "loss": 0.1962,
+      "step": 17105
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3235554768731645e-06,
+      "loss": 0.198,
+      "step": 17106
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.322444046198767e-06,
+      "loss": 0.2264,
+      "step": 17107
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321332668054689e-06,
+      "loss": 0.2209,
+      "step": 17108
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320221342456801e-06,
+      "loss": 0.2369,
+      "step": 17109
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3191100694209815e-06,
+      "loss": 0.2163,
+      "step": 17110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317998848963105e-06,
+      "loss": 0.2045,
+      "step": 17111
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316887681099038e-06,
+      "loss": 0.2209,
+      "step": 17112
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3157765658446555e-06,
+      "loss": 0.2214,
+      "step": 17113
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.314665503215831e-06,
+      "loss": 0.1881,
+      "step": 17114
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.313554493228431e-06,
+      "loss": 0.2536,
+      "step": 17115
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.312443535898328e-06,
+      "loss": 0.1972,
+      "step": 17116
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31133263124139e-06,
+      "loss": 0.2255,
+      "step": 17117
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.310221779273486e-06,
+      "loss": 0.215,
+      "step": 17118
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3091109800104845e-06,
+      "loss": 0.2377,
+      "step": 17119
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.308000233468247e-06,
+      "loss": 0.188,
+      "step": 17120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.306889539662645e-06,
+      "loss": 0.2264,
+      "step": 17121
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30577889860954e-06,
+      "loss": 0.2006,
+      "step": 17122
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3046683103247994e-06,
+      "loss": 0.2377,
+      "step": 17123
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30355777482429e-06,
+      "loss": 0.2104,
+      "step": 17124
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.302447292123866e-06,
+      "loss": 0.2302,
+      "step": 17125
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3013368622393955e-06,
+      "loss": 0.205,
+      "step": 17126
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.300226485186736e-06,
+      "loss": 0.2227,
+      "step": 17127
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.299116160981751e-06,
+      "loss": 0.2276,
+      "step": 17128
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2980058896402995e-06,
+      "loss": 0.2161,
+      "step": 17129
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.296895671178243e-06,
+      "loss": 0.214,
+      "step": 17130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295785505611436e-06,
+      "loss": 0.2295,
+      "step": 17131
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.294675392955738e-06,
+      "loss": 0.2174,
+      "step": 17132
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.293565333227005e-06,
+      "loss": 0.2265,
+      "step": 17133
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.292455326441096e-06,
+      "loss": 0.226,
+      "step": 17134
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291345372613864e-06,
+      "loss": 0.2056,
+      "step": 17135
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290235471761163e-06,
+      "loss": 0.2377,
+      "step": 17136
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289125623898853e-06,
+      "loss": 0.2184,
+      "step": 17137
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.288015829042777e-06,
+      "loss": 0.1867,
+      "step": 17138
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.286906087208792e-06,
+      "loss": 0.2576,
+      "step": 17139
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285796398412751e-06,
+      "loss": 0.2085,
+      "step": 17140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.284686762670503e-06,
+      "loss": 0.2239,
+      "step": 17141
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283577179997902e-06,
+      "loss": 0.1944,
+      "step": 17142
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28246765041079e-06,
+      "loss": 0.2132,
+      "step": 17143
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2813581739250204e-06,
+      "loss": 0.2518,
+      "step": 17144
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280248750556441e-06,
+      "loss": 0.2188,
+      "step": 17145
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2791393803209e-06,
+      "loss": 0.2205,
+      "step": 17146
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2780300632342395e-06,
+      "loss": 0.2535,
+      "step": 17147
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276920799312307e-06,
+      "loss": 0.2449,
+      "step": 17148
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.275811588570947e-06,
+      "loss": 0.2246,
+      "step": 17149
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274702431026009e-06,
+      "loss": 0.2421,
+      "step": 17150
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.273593326693327e-06,
+      "loss": 0.2298,
+      "step": 17151
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.272484275588748e-06,
+      "loss": 0.207,
+      "step": 17152
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271375277728115e-06,
+      "loss": 0.1982,
+      "step": 17153
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270266333127266e-06,
+      "loss": 0.2172,
+      "step": 17154
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.269157441802043e-06,
+      "loss": 0.2185,
+      "step": 17155
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2680486037682845e-06,
+      "loss": 0.2276,
+      "step": 17156
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266939819041829e-06,
+      "loss": 0.2189,
+      "step": 17157
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.265831087638519e-06,
+      "loss": 0.1899,
+      "step": 17158
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264722409574184e-06,
+      "loss": 0.2111,
+      "step": 17159
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.263613784864664e-06,
+      "loss": 0.2355,
+      "step": 17160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262505213525796e-06,
+      "loss": 0.2009,
+      "step": 17161
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.261396695573416e-06,
+      "loss": 0.2391,
+      "step": 17162
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260288231023355e-06,
+      "loss": 0.2645,
+      "step": 17163
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259179819891448e-06,
+      "loss": 0.223,
+      "step": 17164
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258071462193524e-06,
+      "loss": 0.2137,
+      "step": 17165
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.256963157945417e-06,
+      "loss": 0.2258,
+      "step": 17166
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2558549071629595e-06,
+      "loss": 0.2316,
+      "step": 17167
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.254746709861981e-06,
+      "loss": 0.2091,
+      "step": 17168
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253638566058312e-06,
+      "loss": 0.2252,
+      "step": 17169
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.252530475767778e-06,
+      "loss": 0.2159,
+      "step": 17170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.25142243900621e-06,
+      "loss": 0.1909,
+      "step": 17171
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250314455789434e-06,
+      "loss": 0.2117,
+      "step": 17172
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249206526133278e-06,
+      "loss": 0.231,
+      "step": 17173
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248098650053567e-06,
+      "loss": 0.2023,
+      "step": 17174
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246990827566124e-06,
+      "loss": 0.2614,
+      "step": 17175
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245883058686779e-06,
+      "loss": 0.2398,
+      "step": 17176
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.244775343431346e-06,
+      "loss": 0.2287,
+      "step": 17177
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.243667681815655e-06,
+      "loss": 0.2111,
+      "step": 17178
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242560073855523e-06,
+      "loss": 0.1903,
+      "step": 17179
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241452519566775e-06,
+      "loss": 0.2315,
+      "step": 17180
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240345018965231e-06,
+      "loss": 0.2242,
+      "step": 17181
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239237572066709e-06,
+      "loss": 0.1987,
+      "step": 17182
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.238130178887028e-06,
+      "loss": 0.2096,
+      "step": 17183
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.237022839442005e-06,
+      "loss": 0.2378,
+      "step": 17184
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.235915553747462e-06,
+      "loss": 0.2185,
+      "step": 17185
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2348083218192105e-06,
+      "loss": 0.2218,
+      "step": 17186
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2337011436730665e-06,
+      "loss": 0.2262,
+      "step": 17187
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.232594019324847e-06,
+      "loss": 0.2139,
+      "step": 17188
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2314869487903665e-06,
+      "loss": 0.215,
+      "step": 17189
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2303799320854395e-06,
+      "loss": 0.2144,
+      "step": 17190
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.229272969225874e-06,
+      "loss": 0.1949,
+      "step": 17191
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228166060227488e-06,
+      "loss": 0.2463,
+      "step": 17192
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227059205106085e-06,
+      "loss": 0.2241,
+      "step": 17193
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22595240387748e-06,
+      "loss": 0.2214,
+      "step": 17194
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.224845656557483e-06,
+      "loss": 0.2299,
+      "step": 17195
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2237389631619e-06,
+      "loss": 0.2134,
+      "step": 17196
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222632323706546e-06,
+      "loss": 0.2244,
+      "step": 17197
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2215257382072195e-06,
+      "loss": 0.2303,
+      "step": 17198
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22041920667973e-06,
+      "loss": 0.2441,
+      "step": 17199
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.219312729139884e-06,
+      "loss": 0.2191,
+      "step": 17200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.218206305603489e-06,
+      "loss": 0.2082,
+      "step": 17201
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.217099936086345e-06,
+      "loss": 0.2245,
+      "step": 17202
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.215993620604262e-06,
+      "loss": 0.2306,
+      "step": 17203
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.214887359173031e-06,
+      "loss": 0.2163,
+      "step": 17204
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213781151808462e-06,
+      "loss": 0.2368,
+      "step": 17205
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212674998526355e-06,
+      "loss": 0.2204,
+      "step": 17206
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.211568899342509e-06,
+      "loss": 0.2135,
+      "step": 17207
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.210462854272727e-06,
+      "loss": 0.2359,
+      "step": 17208
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209356863332803e-06,
+      "loss": 0.2221,
+      "step": 17209
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.208250926538537e-06,
+      "loss": 0.2079,
+      "step": 17210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2071450439057275e-06,
+      "loss": 0.2347,
+      "step": 17211
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.20603921545017e-06,
+      "loss": 0.2278,
+      "step": 17212
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.204933441187662e-06,
+      "loss": 0.213,
+      "step": 17213
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2038277211339946e-06,
+      "loss": 0.1966,
+      "step": 17214
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.202722055304964e-06,
+      "loss": 0.2195,
+      "step": 17215
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.201616443716368e-06,
+      "loss": 0.2313,
+      "step": 17216
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200510886383988e-06,
+      "loss": 0.2201,
+      "step": 17217
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199405383323624e-06,
+      "loss": 0.206,
+      "step": 17218
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.198299934551068e-06,
+      "loss": 0.2283,
+      "step": 17219
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197194540082106e-06,
+      "loss": 0.2129,
+      "step": 17220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196089199932529e-06,
+      "loss": 0.2316,
+      "step": 17221
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194983914118127e-06,
+      "loss": 0.2076,
+      "step": 17222
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.193878682654686e-06,
+      "loss": 0.231,
+      "step": 17223
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.192773505557995e-06,
+      "loss": 0.2474,
+      "step": 17224
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.191668382843838e-06,
+      "loss": 0.2137,
+      "step": 17225
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190563314528004e-06,
+      "loss": 0.219,
+      "step": 17226
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.189458300626275e-06,
+      "loss": 0.2183,
+      "step": 17227
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1883533411544335e-06,
+      "loss": 0.2432,
+      "step": 17228
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187248436128273e-06,
+      "loss": 0.2177,
+      "step": 17229
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186143585563562e-06,
+      "loss": 0.2201,
+      "step": 17230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.18503878947609e-06,
+      "loss": 0.214,
+      "step": 17231
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183934047881636e-06,
+      "loss": 0.2264,
+      "step": 17232
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.182829360795979e-06,
+      "loss": 0.2464,
+      "step": 17233
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181724728234901e-06,
+      "loss": 0.217,
+      "step": 17234
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.180620150214182e-06,
+      "loss": 0.2108,
+      "step": 17235
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179515626749594e-06,
+      "loss": 0.2301,
+      "step": 17236
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178411157856918e-06,
+      "loss": 0.2287,
+      "step": 17237
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.17730674355193e-06,
+      "loss": 0.2246,
+      "step": 17238
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176202383850405e-06,
+      "loss": 0.2261,
+      "step": 17239
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1750980787681195e-06,
+      "loss": 0.233,
+      "step": 17240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173993828320843e-06,
+      "loss": 0.2236,
+      "step": 17241
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172889632524357e-06,
+      "loss": 0.2167,
+      "step": 17242
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.171785491394424e-06,
+      "loss": 0.2241,
+      "step": 17243
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170681404946819e-06,
+      "loss": 0.2174,
+      "step": 17244
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.169577373197315e-06,
+      "loss": 0.2092,
+      "step": 17245
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.168473396161679e-06,
+      "loss": 0.2503,
+      "step": 17246
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1673694738556855e-06,
+      "loss": 0.2261,
+      "step": 17247
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166265606295097e-06,
+      "loss": 0.2506,
+      "step": 17248
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.165161793495684e-06,
+      "loss": 0.218,
+      "step": 17249
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.164058035473211e-06,
+      "loss": 0.2179,
+      "step": 17250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.16295433224345e-06,
+      "loss": 0.1971,
+      "step": 17251
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.16185068382216e-06,
+      "loss": 0.2332,
+      "step": 17252
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160747090225108e-06,
+      "loss": 0.2034,
+      "step": 17253
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159643551468057e-06,
+      "loss": 0.2479,
+      "step": 17254
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.158540067566776e-06,
+      "loss": 0.2362,
+      "step": 17255
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.157436638537018e-06,
+      "loss": 0.2365,
+      "step": 17256
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156333264394547e-06,
+      "loss": 0.2114,
+      "step": 17257
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155229945155128e-06,
+      "loss": 0.2491,
+      "step": 17258
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.154126680834516e-06,
+      "loss": 0.1999,
+      "step": 17259
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.15302347144847e-06,
+      "loss": 0.2246,
+      "step": 17260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1519203170127515e-06,
+      "loss": 0.22,
+      "step": 17261
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.150817217543117e-06,
+      "loss": 0.2102,
+      "step": 17262
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149714173055323e-06,
+      "loss": 0.2389,
+      "step": 17263
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1486111835651255e-06,
+      "loss": 0.2292,
+      "step": 17264
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147508249088278e-06,
+      "loss": 0.2272,
+      "step": 17265
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.146405369640537e-06,
+      "loss": 0.225,
+      "step": 17266
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.145302545237655e-06,
+      "loss": 0.2291,
+      "step": 17267
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.144199775895388e-06,
+      "loss": 0.2463,
+      "step": 17268
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143097061629483e-06,
+      "loss": 0.2099,
+      "step": 17269
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.141994402455694e-06,
+      "loss": 0.2294,
+      "step": 17270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14089179838977e-06,
+      "loss": 0.2293,
+      "step": 17271
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13978924944746e-06,
+      "loss": 0.2053,
+      "step": 17272
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.138686755644514e-06,
+      "loss": 0.1969,
+      "step": 17273
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.137584316996683e-06,
+      "loss": 0.2227,
+      "step": 17274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136481933519709e-06,
+      "loss": 0.1962,
+      "step": 17275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.135379605229341e-06,
+      "loss": 0.253,
+      "step": 17276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.134277332141325e-06,
+      "loss": 0.2272,
+      "step": 17277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.133175114271405e-06,
+      "loss": 0.2174,
+      "step": 17278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132072951635328e-06,
+      "loss": 0.2364,
+      "step": 17279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.130970844248833e-06,
+      "loss": 0.2339,
+      "step": 17280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129868792127665e-06,
+      "loss": 0.208,
+      "step": 17281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.128766795287568e-06,
+      "loss": 0.2171,
+      "step": 17282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.127664853744276e-06,
+      "loss": 0.2242,
+      "step": 17283
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126562967513534e-06,
+      "loss": 0.2141,
+      "step": 17284
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1254611366110814e-06,
+      "loss": 0.2241,
+      "step": 17285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124359361052659e-06,
+      "loss": 0.2183,
+      "step": 17286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123257640853996e-06,
+      "loss": 0.2171,
+      "step": 17287
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.122155976030839e-06,
+      "loss": 0.2258,
+      "step": 17288
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.121054366598919e-06,
+      "loss": 0.2153,
+      "step": 17289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.119952812573975e-06,
+      "loss": 0.251,
+      "step": 17290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.118851313971736e-06,
+      "loss": 0.2254,
+      "step": 17291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.11774987080794e-06,
+      "loss": 0.2452,
+      "step": 17292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.11664848309832e-06,
+      "loss": 0.2468,
+      "step": 17293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.115547150858607e-06,
+      "loss": 0.2219,
+      "step": 17294
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.114445874104538e-06,
+      "loss": 0.2362,
+      "step": 17295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113344652851833e-06,
+      "loss": 0.2121,
+      "step": 17296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.112243487116232e-06,
+      "loss": 0.2047,
+      "step": 17297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.111142376913455e-06,
+      "loss": 0.2111,
+      "step": 17298
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110041322259236e-06,
+      "loss": 0.2317,
+      "step": 17299
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1089403231693015e-06,
+      "loss": 0.2225,
+      "step": 17300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.107839379659377e-06,
+      "loss": 0.198,
+      "step": 17301
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.106738491745192e-06,
+      "loss": 0.2171,
+      "step": 17302
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.10563765944247e-06,
+      "loss": 0.2327,
+      "step": 17303
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.10453688276693e-06,
+      "loss": 0.1988,
+      "step": 17304
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.103436161734302e-06,
+      "loss": 0.2035,
+      "step": 17305
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.10233549636031e-06,
+      "loss": 0.2219,
+      "step": 17306
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1012348866606694e-06,
+      "loss": 0.2318,
+      "step": 17307
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.100134332651109e-06,
+      "loss": 0.2067,
+      "step": 17308
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0990338343473395e-06,
+      "loss": 0.216,
+      "step": 17309
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.097933391765087e-06,
+      "loss": 0.2069,
+      "step": 17310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.096833004920069e-06,
+      "loss": 0.2116,
+      "step": 17311
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095732673828002e-06,
+      "loss": 0.2342,
+      "step": 17312
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094632398504607e-06,
+      "loss": 0.2232,
+      "step": 17313
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.093532178965598e-06,
+      "loss": 0.2187,
+      "step": 17314
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092432015226688e-06,
+      "loss": 0.2249,
+      "step": 17315
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091331907303595e-06,
+      "loss": 0.2285,
+      "step": 17316
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090231855212033e-06,
+      "loss": 0.2021,
+      "step": 17317
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0891318589677165e-06,
+      "loss": 0.2001,
+      "step": 17318
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0880319185863526e-06,
+      "loss": 0.2245,
+      "step": 17319
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0869320340836564e-06,
+      "loss": 0.2214,
+      "step": 17320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.085832205475344e-06,
+      "loss": 0.205,
+      "step": 17321
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084732432777114e-06,
+      "loss": 0.1944,
+      "step": 17322
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.083632716004682e-06,
+      "loss": 0.2002,
+      "step": 17323
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.082533055173757e-06,
+      "loss": 0.2202,
+      "step": 17324
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0814334503000436e-06,
+      "loss": 0.2043,
+      "step": 17325
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.080333901399252e-06,
+      "loss": 0.2071,
+      "step": 17326
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.079234408487085e-06,
+      "loss": 0.2352,
+      "step": 17327
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078134971579248e-06,
+      "loss": 0.2197,
+      "step": 17328
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.07703559069145e-06,
+      "loss": 0.2317,
+      "step": 17329
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.07593626583939e-06,
+      "loss": 0.2145,
+      "step": 17330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.074836997038771e-06,
+      "loss": 0.2278,
+      "step": 17331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073737784305296e-06,
+      "loss": 0.2327,
+      "step": 17332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072638627654666e-06,
+      "loss": 0.219,
+      "step": 17333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071539527102587e-06,
+      "loss": 0.2183,
+      "step": 17334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.070440482664748e-06,
+      "loss": 0.2395,
+      "step": 17335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0693414943568554e-06,
+      "loss": 0.2234,
+      "step": 17336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068242562194603e-06,
+      "loss": 0.2124,
+      "step": 17337
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0671436861936895e-06,
+      "loss": 0.247,
+      "step": 17338
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.066044866369811e-06,
+      "loss": 0.219,
+      "step": 17339
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064946102738667e-06,
+      "loss": 0.2238,
+      "step": 17340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0638473953159445e-06,
+      "loss": 0.2194,
+      "step": 17341
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062748744117344e-06,
+      "loss": 0.2386,
+      "step": 17342
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0616501491585555e-06,
+      "loss": 0.2048,
+      "step": 17343
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.060551610455273e-06,
+      "loss": 0.2093,
+      "step": 17344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05945312802319e-06,
+      "loss": 0.2115,
+      "step": 17345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0583547018779905e-06,
+      "loss": 0.2001,
+      "step": 17346
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.05725633203537e-06,
+      "loss": 0.214,
+      "step": 17347
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056158018511021e-06,
+      "loss": 0.2192,
+      "step": 17348
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0550597613206205e-06,
+      "loss": 0.2087,
+      "step": 17349
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.053961560479865e-06,
+      "loss": 0.2246,
+      "step": 17350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.052863416004439e-06,
+      "loss": 0.2248,
+      "step": 17351
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051765327910029e-06,
+      "loss": 0.2211,
+      "step": 17352
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0506672962123184e-06,
+      "loss": 0.2354,
+      "step": 17353
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049569320926991e-06,
+      "loss": 0.2004,
+      "step": 17354
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048471402069734e-06,
+      "loss": 0.2238,
+      "step": 17355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.047373539656229e-06,
+      "loss": 0.1973,
+      "step": 17356
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.046275733702155e-06,
+      "loss": 0.2096,
+      "step": 17357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045177984223194e-06,
+      "loss": 0.2358,
+      "step": 17358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.044080291235028e-06,
+      "loss": 0.2073,
+      "step": 17359
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.042982654753336e-06,
+      "loss": 0.2135,
+      "step": 17360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041885074793801e-06,
+      "loss": 0.2327,
+      "step": 17361
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0407875513720915e-06,
+      "loss": 0.2179,
+      "step": 17362
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0396900845038906e-06,
+      "loss": 0.2099,
+      "step": 17363
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038592674204873e-06,
+      "loss": 0.2247,
+      "step": 17364
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037495320490713e-06,
+      "loss": 0.2212,
+      "step": 17365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.036398023377087e-06,
+      "loss": 0.212,
+      "step": 17366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.035300782879668e-06,
+      "loss": 0.2129,
+      "step": 17367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0342035990141325e-06,
+      "loss": 0.2458,
+      "step": 17368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.033106471796146e-06,
+      "loss": 0.2224,
+      "step": 17369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0320094012413855e-06,
+      "loss": 0.228,
+      "step": 17370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0309123873655196e-06,
+      "loss": 0.2043,
+      "step": 17371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0298154301842175e-06,
+      "loss": 0.2406,
+      "step": 17372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.028718529713151e-06,
+      "loss": 0.2253,
+      "step": 17373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027621685967986e-06,
+      "loss": 0.2151,
+      "step": 17374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0265248989643895e-06,
+      "loss": 0.1946,
+      "step": 17375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.025428168718026e-06,
+      "loss": 0.2073,
+      "step": 17376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024331495244565e-06,
+      "loss": 0.2103,
+      "step": 17377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023234878559669e-06,
+      "loss": 0.2274,
+      "step": 17378
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.022138318679007e-06,
+      "loss": 0.2146,
+      "step": 17379
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.021041815618235e-06,
+      "loss": 0.2405,
+      "step": 17380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0199453693930185e-06,
+      "loss": 0.2166,
+      "step": 17381
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.018848980019021e-06,
+      "loss": 0.2602,
+      "step": 17382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017752647511902e-06,
+      "loss": 0.2137,
+      "step": 17383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.016656371887323e-06,
+      "loss": 0.2282,
+      "step": 17384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.01556015316094e-06,
+      "loss": 0.2134,
+      "step": 17385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014463991348413e-06,
+      "loss": 0.2096,
+      "step": 17386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0133678864654044e-06,
+      "loss": 0.2373,
+      "step": 17387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0122718385275615e-06,
+      "loss": 0.2185,
+      "step": 17388
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0111758475505455e-06,
+      "loss": 0.2129,
+      "step": 17389
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.010079913550011e-06,
+      "loss": 0.219,
+      "step": 17390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.008984036541616e-06,
+      "loss": 0.2203,
+      "step": 17391
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007888216541007e-06,
+      "loss": 0.207,
+      "step": 17392
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.006792453563841e-06,
+      "loss": 0.2239,
+      "step": 17393
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.00569674762577e-06,
+      "loss": 0.2202,
+      "step": 17394
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.004601098742446e-06,
+      "loss": 0.2012,
+      "step": 17395
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.003505506929516e-06,
+      "loss": 0.2312,
+      "step": 17396
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002409972202629e-06,
+      "loss": 0.2203,
+      "step": 17397
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001314494577438e-06,
+      "loss": 0.1761,
+      "step": 17398
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.000219074069589e-06,
+      "loss": 0.2315,
+      "step": 17399
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.999123710694732e-06,
+      "loss": 0.2278,
+      "step": 17400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998028404468506e-06,
+      "loss": 0.2261,
+      "step": 17401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.996933155406562e-06,
+      "loss": 0.219,
+      "step": 17402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99583796352454e-06,
+      "loss": 0.2139,
+      "step": 17403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994742828838087e-06,
+      "loss": 0.2216,
+      "step": 17404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.993647751362846e-06,
+      "loss": 0.1928,
+      "step": 17405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9925527311144585e-06,
+      "loss": 0.2145,
+      "step": 17406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.991457768108566e-06,
+      "loss": 0.2225,
+      "step": 17407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.990362862360809e-06,
+      "loss": 0.2247,
+      "step": 17408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989268013886826e-06,
+      "loss": 0.2245,
+      "step": 17409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.988173222702258e-06,
+      "loss": 0.2155,
+      "step": 17410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.987078488822744e-06,
+      "loss": 0.2464,
+      "step": 17411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985983812263917e-06,
+      "loss": 0.2178,
+      "step": 17412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.984889193041419e-06,
+      "loss": 0.2415,
+      "step": 17413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983794631170879e-06,
+      "loss": 0.2237,
+      "step": 17414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.982700126667933e-06,
+      "loss": 0.2157,
+      "step": 17415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981605679548219e-06,
+      "loss": 0.2128,
+      "step": 17416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.980511289827367e-06,
+      "loss": 0.2289,
+      "step": 17417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979416957521011e-06,
+      "loss": 0.2256,
+      "step": 17418
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.978322682644782e-06,
+      "loss": 0.2347,
+      "step": 17419
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977228465214309e-06,
+      "loss": 0.2132,
+      "step": 17420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9761343052452225e-06,
+      "loss": 0.2358,
+      "step": 17421
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.975040202753154e-06,
+      "loss": 0.2312,
+      "step": 17422
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9739461577537316e-06,
+      "loss": 0.2353,
+      "step": 17423
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972852170262578e-06,
+      "loss": 0.2276,
+      "step": 17424
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9717582402953226e-06,
+      "loss": 0.2069,
+      "step": 17425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.970664367867592e-06,
+      "loss": 0.2283,
+      "step": 17426
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969570552995014e-06,
+      "loss": 0.2006,
+      "step": 17427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.968476795693205e-06,
+      "loss": 0.2427,
+      "step": 17428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9673830959777945e-06,
+      "loss": 0.2024,
+      "step": 17429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9662894538644e-06,
+      "loss": 0.2581,
+      "step": 17430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965195869368648e-06,
+      "loss": 0.2064,
+      "step": 17431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.964102342506155e-06,
+      "loss": 0.2354,
+      "step": 17432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.963008873292543e-06,
+      "loss": 0.2231,
+      "step": 17433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961915461743434e-06,
+      "loss": 0.1945,
+      "step": 17434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9608221078744425e-06,
+      "loss": 0.2206,
+      "step": 17435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959728811701184e-06,
+      "loss": 0.2083,
+      "step": 17436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958635573239282e-06,
+      "loss": 0.2493,
+      "step": 17437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.957542392504346e-06,
+      "loss": 0.2156,
+      "step": 17438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956449269511997e-06,
+      "loss": 0.1985,
+      "step": 17439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955356204277844e-06,
+      "loss": 0.2216,
+      "step": 17440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.954263196817504e-06,
+      "loss": 0.2277,
+      "step": 17441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953170247146586e-06,
+      "loss": 0.2202,
+      "step": 17442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952077355280704e-06,
+      "loss": 0.2309,
+      "step": 17443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.950984521235465e-06,
+      "loss": 0.2141,
+      "step": 17444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.949891745026488e-06,
+      "loss": 0.2346,
+      "step": 17445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.948799026669373e-06,
+      "loss": 0.2433,
+      "step": 17446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947706366179733e-06,
+      "loss": 0.2212,
+      "step": 17447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.946613763573174e-06,
+      "loss": 0.2106,
+      "step": 17448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.945521218865303e-06,
+      "loss": 0.2283,
+      "step": 17449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.94442873207173e-06,
+      "loss": 0.2205,
+      "step": 17450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.943336303208053e-06,
+      "loss": 0.1902,
+      "step": 17451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942243932289879e-06,
+      "loss": 0.1899,
+      "step": 17452
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941151619332819e-06,
+      "loss": 0.2162,
+      "step": 17453
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940059364352464e-06,
+      "loss": 0.2083,
+      "step": 17454
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93896716736442e-06,
+      "loss": 0.2114,
+      "step": 17455
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.937875028384289e-06,
+      "loss": 0.2161,
+      "step": 17456
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936782947427673e-06,
+      "loss": 0.2056,
+      "step": 17457
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9356909245101666e-06,
+      "loss": 0.2188,
+      "step": 17458
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93459895964737e-06,
+      "loss": 0.2232,
+      "step": 17459
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.933507052854885e-06,
+      "loss": 0.2132,
+      "step": 17460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932415204148303e-06,
+      "loss": 0.2425,
+      "step": 17461
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931323413543224e-06,
+      "loss": 0.2384,
+      "step": 17462
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930231681055239e-06,
+      "loss": 0.2464,
+      "step": 17463
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929140006699945e-06,
+      "loss": 0.2245,
+      "step": 17464
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.928048390492935e-06,
+      "loss": 0.2027,
+      "step": 17465
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.926956832449806e-06,
+      "loss": 0.2199,
+      "step": 17466
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.925865332586143e-06,
+      "loss": 0.2159,
+      "step": 17467
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.924773890917539e-06,
+      "loss": 0.2156,
+      "step": 17468
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923682507459585e-06,
+      "loss": 0.2252,
+      "step": 17469
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92259118222787e-06,
+      "loss": 0.2107,
+      "step": 17470
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.921499915237983e-06,
+      "loss": 0.2296,
+      "step": 17471
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920408706505509e-06,
+      "loss": 0.225,
+      "step": 17472
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919317556046041e-06,
+      "loss": 0.2405,
+      "step": 17473
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.91822646387516e-06,
+      "loss": 0.2268,
+      "step": 17474
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917135430008451e-06,
+      "loss": 0.2136,
+      "step": 17475
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916044454461501e-06,
+      "loss": 0.2317,
+      "step": 17476
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.914953537249892e-06,
+      "loss": 0.2292,
+      "step": 17477
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.91386267838921e-06,
+      "loss": 0.2179,
+      "step": 17478
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9127718778950325e-06,
+      "loss": 0.2304,
+      "step": 17479
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9116811357829426e-06,
+      "loss": 0.2402,
+      "step": 17480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9105904520685185e-06,
+      "loss": 0.2348,
+      "step": 17481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.90949982676734e-06,
+      "loss": 0.2051,
+      "step": 17482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.908409259894987e-06,
+      "loss": 0.2119,
+      "step": 17483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.907318751467038e-06,
+      "loss": 0.2391,
+      "step": 17484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906228301499067e-06,
+      "loss": 0.2319,
+      "step": 17485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.905137910006652e-06,
+      "loss": 0.2304,
+      "step": 17486
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9040475770053675e-06,
+      "loss": 0.2177,
+      "step": 17487
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9029573025107875e-06,
+      "loss": 0.2187,
+      "step": 17488
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.90186708653849e-06,
+      "loss": 0.209,
+      "step": 17489
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900776929104041e-06,
+      "loss": 0.2327,
+      "step": 17490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.899686830223015e-06,
+      "loss": 0.2404,
+      "step": 17491
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898596789910986e-06,
+      "loss": 0.2512,
+      "step": 17492
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89750680818352e-06,
+      "loss": 0.2071,
+      "step": 17493
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.896416885056184e-06,
+      "loss": 0.2352,
+      "step": 17494
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8953270205445536e-06,
+      "loss": 0.2212,
+      "step": 17495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.894237214664193e-06,
+      "loss": 0.2101,
+      "step": 17496
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893147467430669e-06,
+      "loss": 0.197,
+      "step": 17497
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892057778859546e-06,
+      "loss": 0.2278,
+      "step": 17498
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.890968148966391e-06,
+      "loss": 0.2317,
+      "step": 17499
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.88987857776677e-06,
+      "loss": 0.2362,
+      "step": 17500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.888789065276244e-06,
+      "loss": 0.1904,
+      "step": 17501
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887699611510374e-06,
+      "loss": 0.2174,
+      "step": 17502
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886610216484725e-06,
+      "loss": 0.2146,
+      "step": 17503
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885520880214857e-06,
+      "loss": 0.2155,
+      "step": 17504
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8844316027163315e-06,
+      "loss": 0.2195,
+      "step": 17505
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883342384004707e-06,
+      "loss": 0.2147,
+      "step": 17506
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.882253224095541e-06,
+      "loss": 0.2273,
+      "step": 17507
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8811641230043894e-06,
+      "loss": 0.2078,
+      "step": 17508
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8800750807468095e-06,
+      "loss": 0.2175,
+      "step": 17509
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878986097338361e-06,
+      "loss": 0.2629,
+      "step": 17510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.877897172794595e-06,
+      "loss": 0.2215,
+      "step": 17511
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87680830713107e-06,
+      "loss": 0.2101,
+      "step": 17512
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.875719500363335e-06,
+      "loss": 0.2179,
+      "step": 17513
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.874630752506943e-06,
+      "loss": 0.254,
+      "step": 17514
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873542063577447e-06,
+      "loss": 0.2242,
+      "step": 17515
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.872453433590401e-06,
+      "loss": 0.2294,
+      "step": 17516
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.871364862561348e-06,
+      "loss": 0.2187,
+      "step": 17517
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.870276350505843e-06,
+      "loss": 0.218,
+      "step": 17518
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.869187897439436e-06,
+      "loss": 0.2301,
+      "step": 17519
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868099503377667e-06,
+      "loss": 0.2246,
+      "step": 17520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8670111683360855e-06,
+      "loss": 0.2219,
+      "step": 17521
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.865922892330238e-06,
+      "loss": 0.226,
+      "step": 17522
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8648346753756725e-06,
+      "loss": 0.2098,
+      "step": 17523
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.863746517487929e-06,
+      "loss": 0.2254,
+      "step": 17524
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.86265841868255e-06,
+      "loss": 0.2173,
+      "step": 17525
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861570378975082e-06,
+      "loss": 0.2028,
+      "step": 17526
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.860482398381064e-06,
+      "loss": 0.2536,
+      "step": 17527
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859394476916041e-06,
+      "loss": 0.2232,
+      "step": 17528
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8583066145955446e-06,
+      "loss": 0.2134,
+      "step": 17529
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85721881143512e-06,
+      "loss": 0.2109,
+      "step": 17530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856131067450305e-06,
+      "loss": 0.2306,
+      "step": 17531
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.855043382656638e-06,
+      "loss": 0.2013,
+      "step": 17532
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.853955757069651e-06,
+      "loss": 0.2185,
+      "step": 17533
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852868190704883e-06,
+      "loss": 0.224,
+      "step": 17534
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.851780683577868e-06,
+      "loss": 0.1935,
+      "step": 17535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850693235704138e-06,
+      "loss": 0.2167,
+      "step": 17536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.849605847099229e-06,
+      "loss": 0.2137,
+      "step": 17537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.848518517778672e-06,
+      "loss": 0.2421,
+      "step": 17538
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847431247758002e-06,
+      "loss": 0.2117,
+      "step": 17539
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846344037052743e-06,
+      "loss": 0.2323,
+      "step": 17540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8452568856784284e-06,
+      "loss": 0.2348,
+      "step": 17541
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844169793650587e-06,
+      "loss": 0.2132,
+      "step": 17542
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843082760984749e-06,
+      "loss": 0.2155,
+      "step": 17543
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841995787696438e-06,
+      "loss": 0.2228,
+      "step": 17544
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.840908873801183e-06,
+      "loss": 0.1783,
+      "step": 17545
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.839822019314509e-06,
+      "loss": 0.252,
+      "step": 17546
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.838735224251937e-06,
+      "loss": 0.2225,
+      "step": 17547
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.837648488628995e-06,
+      "loss": 0.2417,
+      "step": 17548
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.836561812461205e-06,
+      "loss": 0.2189,
+      "step": 17549
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.835475195764087e-06,
+      "loss": 0.2298,
+      "step": 17550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834388638553166e-06,
+      "loss": 0.233,
+      "step": 17551
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833302140843957e-06,
+      "loss": 0.2105,
+      "step": 17552
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832215702651987e-06,
+      "loss": 0.2121,
+      "step": 17553
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.831129323992771e-06,
+      "loss": 0.229,
+      "step": 17554
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830043004881822e-06,
+      "loss": 0.2112,
+      "step": 17555
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.828956745334667e-06,
+      "loss": 0.2073,
+      "step": 17556
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827870545366815e-06,
+      "loss": 0.2295,
+      "step": 17557
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8267844049937835e-06,
+      "loss": 0.2175,
+      "step": 17558
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8256983242310835e-06,
+      "loss": 0.2306,
+      "step": 17559
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824612303094236e-06,
+      "loss": 0.2369,
+      "step": 17560
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.823526341598749e-06,
+      "loss": 0.2462,
+      "step": 17561
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82244043976013e-06,
+      "loss": 0.233,
+      "step": 17562
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8213545975939e-06,
+      "loss": 0.2327,
+      "step": 17563
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8202688151155625e-06,
+      "loss": 0.2197,
+      "step": 17564
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.819183092340626e-06,
+      "loss": 0.2247,
+      "step": 17565
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818097429284606e-06,
+      "loss": 0.2111,
+      "step": 17566
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.817011825962999e-06,
+      "loss": 0.2122,
+      "step": 17567
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815926282391324e-06,
+      "loss": 0.2343,
+      "step": 17568
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.814840798585082e-06,
+      "loss": 0.1982,
+      "step": 17569
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.813755374559773e-06,
+      "loss": 0.2168,
+      "step": 17570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8126700103309134e-06,
+      "loss": 0.1993,
+      "step": 17571
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.81158470591399e-06,
+      "loss": 0.2182,
+      "step": 17572
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.810499461324519e-06,
+      "loss": 0.2396,
+      "step": 17573
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.809414276577994e-06,
+      "loss": 0.2259,
+      "step": 17574
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808329151689922e-06,
+      "loss": 0.2419,
+      "step": 17575
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8072440866757986e-06,
+      "loss": 0.2188,
+      "step": 17576
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806159081551122e-06,
+      "loss": 0.2079,
+      "step": 17577
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8050741363313965e-06,
+      "loss": 0.2134,
+      "step": 17578
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803989251032112e-06,
+      "loss": 0.2094,
+      "step": 17579
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802904425668772e-06,
+      "loss": 0.2107,
+      "step": 17580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8018196602568685e-06,
+      "loss": 0.2335,
+      "step": 17581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.800734954811894e-06,
+      "loss": 0.2153,
+      "step": 17582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.799650309349348e-06,
+      "loss": 0.2104,
+      "step": 17583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.798565723884717e-06,
+      "loss": 0.1939,
+      "step": 17584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.797481198433505e-06,
+      "loss": 0.2403,
+      "step": 17585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796396733011189e-06,
+      "loss": 0.2203,
+      "step": 17586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795312327633269e-06,
+      "loss": 0.2221,
+      "step": 17587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794227982315231e-06,
+      "loss": 0.2357,
+      "step": 17588
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793143697072561e-06,
+      "loss": 0.2079,
+      "step": 17589
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7920594719207545e-06,
+      "loss": 0.2012,
+      "step": 17590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.790975306875291e-06,
+      "loss": 0.2366,
+      "step": 17591
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789891201951663e-06,
+      "loss": 0.2108,
+      "step": 17592
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788807157165355e-06,
+      "loss": 0.2135,
+      "step": 17593
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.787723172531844e-06,
+      "loss": 0.2206,
+      "step": 17594
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786639248066623e-06,
+      "loss": 0.222,
+      "step": 17595
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785555383785173e-06,
+      "loss": 0.229,
+      "step": 17596
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784471579702969e-06,
+      "loss": 0.2307,
+      "step": 17597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.783387835835505e-06,
+      "loss": 0.2389,
+      "step": 17598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.782304152198246e-06,
+      "loss": 0.2166,
+      "step": 17599
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.781220528806683e-06,
+      "loss": 0.2477,
+      "step": 17600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7801369656762864e-06,
+      "loss": 0.2442,
+      "step": 17601
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779053462822541e-06,
+      "loss": 0.2217,
+      "step": 17602
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.777970020260921e-06,
+      "loss": 0.2379,
+      "step": 17603
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776886638006898e-06,
+      "loss": 0.224,
+      "step": 17604
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.775803316075954e-06,
+      "loss": 0.2053,
+      "step": 17605
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774720054483556e-06,
+      "loss": 0.2187,
+      "step": 17606
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773636853245186e-06,
+      "loss": 0.2448,
+      "step": 17607
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.772553712376311e-06,
+      "loss": 0.2358,
+      "step": 17608
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7714706318924e-06,
+      "loss": 0.2114,
+      "step": 17609
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77038761180893e-06,
+      "loss": 0.2203,
+      "step": 17610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76930465214137e-06,
+      "loss": 0.2142,
+      "step": 17611
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.768221752905185e-06,
+      "loss": 0.234,
+      "step": 17612
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.767138914115842e-06,
+      "loss": 0.2087,
+      "step": 17613
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766056135788816e-06,
+      "loss": 0.2337,
+      "step": 17614
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764973417939567e-06,
+      "loss": 0.2267,
+      "step": 17615
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.763890760583561e-06,
+      "loss": 0.2081,
+      "step": 17616
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7628081637362675e-06,
+      "loss": 0.1875,
+      "step": 17617
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761725627413144e-06,
+      "loss": 0.2218,
+      "step": 17618
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.760643151629659e-06,
+      "loss": 0.2196,
+      "step": 17619
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759560736401273e-06,
+      "loss": 0.2295,
+      "step": 17620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758478381743442e-06,
+      "loss": 0.2227,
+      "step": 17621
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.757396087671634e-06,
+      "loss": 0.2393,
+      "step": 17622
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.756313854201303e-06,
+      "loss": 0.2361,
+      "step": 17623
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.755231681347917e-06,
+      "loss": 0.2064,
+      "step": 17624
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.754149569126919e-06,
+      "loss": 0.2046,
+      "step": 17625
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753067517553777e-06,
+      "loss": 0.2285,
+      "step": 17626
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.751985526643945e-06,
+      "loss": 0.2299,
+      "step": 17627
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.750903596412874e-06,
+      "loss": 0.2243,
+      "step": 17628
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7498217268760225e-06,
+      "loss": 0.2269,
+      "step": 17629
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748739918048845e-06,
+      "loss": 0.2157,
+      "step": 17630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747658169946786e-06,
+      "loss": 0.2175,
+      "step": 17631
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746576482585308e-06,
+      "loss": 0.2096,
+      "step": 17632
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745494855979852e-06,
+      "loss": 0.2113,
+      "step": 17633
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744413290145878e-06,
+      "loss": 0.2424,
+      "step": 17634
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.74333178509883e-06,
+      "loss": 0.2264,
+      "step": 17635
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742250340854152e-06,
+      "loss": 0.2259,
+      "step": 17636
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.741168957427304e-06,
+      "loss": 0.2096,
+      "step": 17637
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740087634833715e-06,
+      "loss": 0.2196,
+      "step": 17638
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.739006373088844e-06,
+      "loss": 0.2059,
+      "step": 17639
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.737925172208129e-06,
+      "loss": 0.2291,
+      "step": 17640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7368440322070194e-06,
+      "loss": 0.2057,
+      "step": 17641
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735762953100956e-06,
+      "loss": 0.2182,
+      "step": 17642
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.734681934905375e-06,
+      "loss": 0.2243,
+      "step": 17643
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.733600977635727e-06,
+      "loss": 0.2322,
+      "step": 17644
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.732520081307447e-06,
+      "loss": 0.2053,
+      "step": 17645
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731439245935977e-06,
+      "loss": 0.2296,
+      "step": 17646
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7303584715367555e-06,
+      "loss": 0.2278,
+      "step": 17647
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7292777581252155e-06,
+      "loss": 0.2267,
+      "step": 17648
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.728197105716802e-06,
+      "loss": 0.2312,
+      "step": 17649
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.727116514326947e-06,
+      "loss": 0.235,
+      "step": 17650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.726035983971086e-06,
+      "loss": 0.2318,
+      "step": 17651
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7249555146646486e-06,
+      "loss": 0.2165,
+      "step": 17652
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7238751064230765e-06,
+      "loss": 0.2165,
+      "step": 17653
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.722794759261798e-06,
+      "loss": 0.2104,
+      "step": 17654
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.721714473196242e-06,
+      "loss": 0.2329,
+      "step": 17655
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.720634248241848e-06,
+      "loss": 0.2191,
+      "step": 17656
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.719554084414034e-06,
+      "loss": 0.2269,
+      "step": 17657
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.718473981728241e-06,
+      "loss": 0.2405,
+      "step": 17658
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7173939401998914e-06,
+      "loss": 0.24,
+      "step": 17659
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.716313959844408e-06,
+      "loss": 0.2091,
+      "step": 17660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.715234040677229e-06,
+      "loss": 0.2191,
+      "step": 17661
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.714154182713767e-06,
+      "loss": 0.2158,
+      "step": 17662
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 0.2186,
+      "step": 17663
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.71199465045972e-06,
+      "loss": 0.2251,
+      "step": 17664
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.710914976199977e-06,
+      "loss": 0.2327,
+      "step": 17665
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7098353632056516e-06,
+      "loss": 0.2,
+      "step": 17666
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.708755811492159e-06,
+      "loss": 0.2335,
+      "step": 17667
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.707676321074931e-06,
+      "loss": 0.2102,
+      "step": 17668
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.706596891969378e-06,
+      "loss": 0.2217,
+      "step": 17669
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70551752419092e-06,
+      "loss": 0.2335,
+      "step": 17670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704438217754979e-06,
+      "loss": 0.2333,
+      "step": 17671
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.703358972676967e-06,
+      "loss": 0.2026,
+      "step": 17672
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7022797889723044e-06,
+      "loss": 0.2266,
+      "step": 17673
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.701200666656403e-06,
+      "loss": 0.2305,
+      "step": 17674
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.700121605744676e-06,
+      "loss": 0.2125,
+      "step": 17675
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699042606252543e-06,
+      "loss": 0.2477,
+      "step": 17676
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69796366819541e-06,
+      "loss": 0.2163,
+      "step": 17677
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696884791588693e-06,
+      "loss": 0.2031,
+      "step": 17678
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695805976447794e-06,
+      "loss": 0.2414,
+      "step": 17679
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694727222788136e-06,
+      "loss": 0.2208,
+      "step": 17680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69364853062512e-06,
+      "loss": 0.2042,
+      "step": 17681
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692569899974152e-06,
+      "loss": 0.2276,
+      "step": 17682
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.691491330850645e-06,
+      "loss": 0.224,
+      "step": 17683
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.69041282327e-06,
+      "loss": 0.2148,
+      "step": 17684
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689334377247629e-06,
+      "loss": 0.2305,
+      "step": 17685
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688255992798932e-06,
+      "loss": 0.2279,
+      "step": 17686
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687177669939311e-06,
+      "loss": 0.2035,
+      "step": 17687
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.686099408684175e-06,
+      "loss": 0.1802,
+      "step": 17688
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.685021209048918e-06,
+      "loss": 0.2475,
+      "step": 17689
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683943071048952e-06,
+      "loss": 0.1956,
+      "step": 17690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682864994699664e-06,
+      "loss": 0.2094,
+      "step": 17691
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681786980016464e-06,
+      "loss": 0.2384,
+      "step": 17692
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.680709027014744e-06,
+      "loss": 0.2381,
+      "step": 17693
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679631135709901e-06,
+      "loss": 0.2062,
+      "step": 17694
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.678553306117338e-06,
+      "loss": 0.2194,
+      "step": 17695
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677475538252442e-06,
+      "loss": 0.2103,
+      "step": 17696
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.676397832130618e-06,
+      "loss": 0.2262,
+      "step": 17697
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6753201877672545e-06,
+      "loss": 0.1913,
+      "step": 17698
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674242605177742e-06,
+      "loss": 0.233,
+      "step": 17699
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673165084377479e-06,
+      "loss": 0.2131,
+      "step": 17700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.672087625381852e-06,
+      "loss": 0.1957,
+      "step": 17701
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.671010228206251e-06,
+      "loss": 0.2441,
+      "step": 17702
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669932892866074e-06,
+      "loss": 0.2422,
+      "step": 17703
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.668855619376696e-06,
+      "loss": 0.225,
+      "step": 17704
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667778407753515e-06,
+      "loss": 0.2074,
+      "step": 17705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666701258011913e-06,
+      "loss": 0.2118,
+      "step": 17706
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6656241701672795e-06,
+      "loss": 0.2095,
+      "step": 17707
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664547144235e-06,
+      "loss": 0.2246,
+      "step": 17708
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.663470180230451e-06,
+      "loss": 0.2271,
+      "step": 17709
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662393278169027e-06,
+      "loss": 0.2231,
+      "step": 17710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6613164380661025e-06,
+      "loss": 0.2126,
+      "step": 17711
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660239659937064e-06,
+      "loss": 0.2291,
+      "step": 17712
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.659162943797291e-06,
+      "loss": 0.1983,
+      "step": 17713
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.658086289662158e-06,
+      "loss": 0.2218,
+      "step": 17714
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657009697547052e-06,
+      "loss": 0.2334,
+      "step": 17715
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.655933167467349e-06,
+      "loss": 0.2027,
+      "step": 17716
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.654856699438422e-06,
+      "loss": 0.2111,
+      "step": 17717
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.653780293475649e-06,
+      "loss": 0.2153,
+      "step": 17718
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6527039495944095e-06,
+      "loss": 0.2355,
+      "step": 17719
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651627667810073e-06,
+      "loss": 0.228,
+      "step": 17720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6505514481380145e-06,
+      "loss": 0.2082,
+      "step": 17721
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.649475290593609e-06,
+      "loss": 0.2335,
+      "step": 17722
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648399195192222e-06,
+      "loss": 0.2372,
+      "step": 17723
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647323161949235e-06,
+      "loss": 0.2552,
+      "step": 17724
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.646247190880012e-06,
+      "loss": 0.2055,
+      "step": 17725
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645171281999918e-06,
+      "loss": 0.2096,
+      "step": 17726
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.64409543532433e-06,
+      "loss": 0.2225,
+      "step": 17727
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643019650868606e-06,
+      "loss": 0.2278,
+      "step": 17728
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.641943928648124e-06,
+      "loss": 0.2274,
+      "step": 17729
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6408682686782415e-06,
+      "loss": 0.2244,
+      "step": 17730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639792670974326e-06,
+      "loss": 0.2189,
+      "step": 17731
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638717135551741e-06,
+      "loss": 0.211,
+      "step": 17732
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.637641662425843e-06,
+      "loss": 0.2286,
+      "step": 17733
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.636566251612006e-06,
+      "loss": 0.2016,
+      "step": 17734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635490903125585e-06,
+      "loss": 0.2222,
+      "step": 17735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.634415616981936e-06,
+      "loss": 0.2169,
+      "step": 17736
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.633340393196427e-06,
+      "loss": 0.2459,
+      "step": 17737
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632265231784409e-06,
+      "loss": 0.2224,
+      "step": 17738
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.631190132761247e-06,
+      "loss": 0.2056,
+      "step": 17739
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.630115096142295e-06,
+      "loss": 0.2277,
+      "step": 17740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629040121942902e-06,
+      "loss": 0.2194,
+      "step": 17741
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627965210178433e-06,
+      "loss": 0.2175,
+      "step": 17742
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626890360864239e-06,
+      "loss": 0.2043,
+      "step": 17743
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.625815574015671e-06,
+      "loss": 0.2088,
+      "step": 17744
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624740849648078e-06,
+      "loss": 0.2235,
+      "step": 17745
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623666187776821e-06,
+      "loss": 0.2384,
+      "step": 17746
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622591588417243e-06,
+      "loss": 0.2123,
+      "step": 17747
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621517051584694e-06,
+      "loss": 0.2225,
+      "step": 17748
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620442577294527e-06,
+      "loss": 0.2112,
+      "step": 17749
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.619368165562085e-06,
+      "loss": 0.216,
+      "step": 17750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61829381640272e-06,
+      "loss": 0.2225,
+      "step": 17751
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6172195298317755e-06,
+      "loss": 0.2136,
+      "step": 17752
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6161453058645934e-06,
+      "loss": 0.221,
+      "step": 17753
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6150711445165226e-06,
+      "loss": 0.2289,
+      "step": 17754
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.613997045802903e-06,
+      "loss": 0.2109,
+      "step": 17755
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612923009739086e-06,
+      "loss": 0.2342,
+      "step": 17756
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611849036340398e-06,
+      "loss": 0.2264,
+      "step": 17757
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.610775125622192e-06,
+      "loss": 0.2212,
+      "step": 17758
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.609701277599804e-06,
+      "loss": 0.2179,
+      "step": 17759
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.608627492288567e-06,
+      "loss": 0.2201,
+      "step": 17760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.607553769703828e-06,
+      "loss": 0.2015,
+      "step": 17761
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6064801098609185e-06,
+      "loss": 0.2303,
+      "step": 17762
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.60540651277518e-06,
+      "loss": 0.2182,
+      "step": 17763
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604332978461944e-06,
+      "loss": 0.2249,
+      "step": 17764
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.603259506936541e-06,
+      "loss": 0.21,
+      "step": 17765
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6021860982143126e-06,
+      "loss": 0.2022,
+      "step": 17766
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601112752310584e-06,
+      "loss": 0.2267,
+      "step": 17767
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.600039469240696e-06,
+      "loss": 0.2154,
+      "step": 17768
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598966249019972e-06,
+      "loss": 0.2169,
+      "step": 17769
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.597893091663744e-06,
+      "loss": 0.1983,
+      "step": 17770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596819997187342e-06,
+      "loss": 0.2252,
+      "step": 17771
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.595746965606088e-06,
+      "loss": 0.1942,
+      "step": 17772
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5946739969353184e-06,
+      "loss": 0.231,
+      "step": 17773
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5936010911903545e-06,
+      "loss": 0.2252,
+      "step": 17774
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592528248386519e-06,
+      "loss": 0.2235,
+      "step": 17775
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.591455468539145e-06,
+      "loss": 0.2129,
+      "step": 17776
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.590382751663547e-06,
+      "loss": 0.2349,
+      "step": 17777
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.589310097775055e-06,
+      "loss": 0.2046,
+      "step": 17778
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588237506888987e-06,
+      "loss": 0.2112,
+      "step": 17779
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.587164979020661e-06,
+      "loss": 0.2343,
+      "step": 17780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586092514185405e-06,
+      "loss": 0.2213,
+      "step": 17781
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.585020112398534e-06,
+      "loss": 0.2402,
+      "step": 17782
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.583947773675364e-06,
+      "loss": 0.2169,
+      "step": 17783
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.582875498031212e-06,
+      "loss": 0.2211,
+      "step": 17784
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5818032854814e-06,
+      "loss": 0.2261,
+      "step": 17785
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5807311360412396e-06,
+      "loss": 0.1951,
+      "step": 17786
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579659049726042e-06,
+      "loss": 0.2155,
+      "step": 17787
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.578587026551129e-06,
+      "loss": 0.2468,
+      "step": 17788
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.577515066531804e-06,
+      "loss": 0.2148,
+      "step": 17789
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576443169683391e-06,
+      "loss": 0.1911,
+      "step": 17790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.575371336021192e-06,
+      "loss": 0.1963,
+      "step": 17791
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574299565560517e-06,
+      "loss": 0.2299,
+      "step": 17792
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.57322785831668e-06,
+      "loss": 0.2329,
+      "step": 17793
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572156214304982e-06,
+      "loss": 0.2491,
+      "step": 17794
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.571084633540745e-06,
+      "loss": 0.2311,
+      "step": 17795
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570013116039256e-06,
+      "loss": 0.2173,
+      "step": 17796
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568941661815836e-06,
+      "loss": 0.2082,
+      "step": 17797
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5678702708857835e-06,
+      "loss": 0.2198,
+      "step": 17798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566798943264398e-06,
+      "loss": 0.2255,
+      "step": 17799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565727678966991e-06,
+      "loss": 0.2402,
+      "step": 17800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564656478008858e-06,
+      "loss": 0.225,
+      "step": 17801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.563585340405305e-06,
+      "loss": 0.2207,
+      "step": 17802
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5625142661716304e-06,
+      "loss": 0.2136,
+      "step": 17803
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.561443255323129e-06,
+      "loss": 0.1955,
+      "step": 17804
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560372307875106e-06,
+      "loss": 0.1916,
+      "step": 17805
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.559301423842856e-06,
+      "loss": 0.2191,
+      "step": 17806
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558230603241672e-06,
+      "loss": 0.1985,
+      "step": 17807
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.557159846086857e-06,
+      "loss": 0.2369,
+      "step": 17808
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.556089152393702e-06,
+      "loss": 0.2246,
+      "step": 17809
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5550185221775e-06,
+      "loss": 0.2018,
+      "step": 17810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55394795545354e-06,
+      "loss": 0.2294,
+      "step": 17811
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552877452237124e-06,
+      "loss": 0.2308,
+      "step": 17812
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.551807012543536e-06,
+      "loss": 0.2212,
+      "step": 17813
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.550736636388064e-06,
+      "loss": 0.229,
+      "step": 17814
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549666323786006e-06,
+      "loss": 0.1992,
+      "step": 17815
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548596074752641e-06,
+      "loss": 0.2285,
+      "step": 17816
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547525889303265e-06,
+      "loss": 0.2252,
+      "step": 17817
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5464557674531596e-06,
+      "loss": 0.2186,
+      "step": 17818
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545385709217608e-06,
+      "loss": 0.2158,
+      "step": 17819
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5443157146119005e-06,
+      "loss": 0.2175,
+      "step": 17820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543245783651318e-06,
+      "loss": 0.2187,
+      "step": 17821
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542175916351149e-06,
+      "loss": 0.211,
+      "step": 17822
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541106112726664e-06,
+      "loss": 0.1942,
+      "step": 17823
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.540036372793156e-06,
+      "loss": 0.2167,
+      "step": 17824
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538966696565897e-06,
+      "loss": 0.2152,
+      "step": 17825
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.537897084060167e-06,
+      "loss": 0.2445,
+      "step": 17826
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.536827535291251e-06,
+      "loss": 0.2065,
+      "step": 17827
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.535758050274418e-06,
+      "loss": 0.2173,
+      "step": 17828
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.534688629024951e-06,
+      "loss": 0.2405,
+      "step": 17829
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5336192715581235e-06,
+      "loss": 0.2182,
+      "step": 17830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532549977889208e-06,
+      "loss": 0.2417,
+      "step": 17831
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.531480748033483e-06,
+      "loss": 0.2303,
+      "step": 17832
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.530411582006214e-06,
+      "loss": 0.2224,
+      "step": 17833
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5293424798226835e-06,
+      "loss": 0.2203,
+      "step": 17834
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528273441498155e-06,
+      "loss": 0.1925,
+      "step": 17835
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527204467047902e-06,
+      "loss": 0.2089,
+      "step": 17836
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526135556487191e-06,
+      "loss": 0.1841,
+      "step": 17837
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525066709831288e-06,
+      "loss": 0.2075,
+      "step": 17838
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.523997927095468e-06,
+      "loss": 0.1921,
+      "step": 17839
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.522929208294994e-06,
+      "loss": 0.2305,
+      "step": 17840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.521860553445127e-06,
+      "loss": 0.2007,
+      "step": 17841
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.52079196256114e-06,
+      "loss": 0.1899,
+      "step": 17842
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.519723435658289e-06,
+      "loss": 0.2103,
+      "step": 17843
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5186549727518425e-06,
+      "loss": 0.2194,
+      "step": 17844
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517586573857061e-06,
+      "loss": 0.2157,
+      "step": 17845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.516518238989202e-06,
+      "loss": 0.2085,
+      "step": 17846
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.515449968163533e-06,
+      "loss": 0.2037,
+      "step": 17847
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514381761395309e-06,
+      "loss": 0.2195,
+      "step": 17848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.513313618699787e-06,
+      "loss": 0.2066,
+      "step": 17849
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5122455400922215e-06,
+      "loss": 0.1979,
+      "step": 17850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511177525587878e-06,
+      "loss": 0.2199,
+      "step": 17851
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510109575202005e-06,
+      "loss": 0.2156,
+      "step": 17852
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.509041688949857e-06,
+      "loss": 0.2309,
+      "step": 17853
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5079738668466944e-06,
+      "loss": 0.2206,
+      "step": 17854
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5069061089077604e-06,
+      "loss": 0.2078,
+      "step": 17855
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505838415148317e-06,
+      "loss": 0.2353,
+      "step": 17856
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.504770785583609e-06,
+      "loss": 0.2053,
+      "step": 17857
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503703220228884e-06,
+      "loss": 0.2236,
+      "step": 17858
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502635719099399e-06,
+      "loss": 0.2193,
+      "step": 17859
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501568282210392e-06,
+      "loss": 0.1944,
+      "step": 17860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500500909577127e-06,
+      "loss": 0.2094,
+      "step": 17861
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49943360121483e-06,
+      "loss": 0.2498,
+      "step": 17862
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4983663571387605e-06,
+      "loss": 0.2231,
+      "step": 17863
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4972991773641595e-06,
+      "loss": 0.2354,
+      "step": 17864
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.496232061906264e-06,
+      "loss": 0.2109,
+      "step": 17865
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495165010780328e-06,
+      "loss": 0.2031,
+      "step": 17866
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494098024001584e-06,
+      "loss": 0.2234,
+      "step": 17867
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49303110158528e-06,
+      "loss": 0.2232,
+      "step": 17868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.491964243546652e-06,
+      "loss": 0.2195,
+      "step": 17869
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490897449900936e-06,
+      "loss": 0.1942,
+      "step": 17870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489830720663378e-06,
+      "loss": 0.2214,
+      "step": 17871
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.488764055849207e-06,
+      "loss": 0.2347,
+      "step": 17872
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.487697455473668e-06,
+      "loss": 0.221,
+      "step": 17873
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4866309195519895e-06,
+      "loss": 0.2253,
+      "step": 17874
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48556444809941e-06,
+      "loss": 0.2235,
+      "step": 17875
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48449804113116e-06,
+      "loss": 0.2184,
+      "step": 17876
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483431698662469e-06,
+      "loss": 0.2028,
+      "step": 17877
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4823654207085775e-06,
+      "loss": 0.1996,
+      "step": 17878
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48129920728471e-06,
+      "loss": 0.247,
+      "step": 17879
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480233058406095e-06,
+      "loss": 0.24,
+      "step": 17880
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4791669740879684e-06,
+      "loss": 0.2332,
+      "step": 17881
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.47810095434555e-06,
+      "loss": 0.2279,
+      "step": 17882
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.477034999194074e-06,
+      "loss": 0.2123,
+      "step": 17883
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4759691086487645e-06,
+      "loss": 0.2494,
+      "step": 17884
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4749032827248405e-06,
+      "loss": 0.2085,
+      "step": 17885
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.473837521437536e-06,
+      "loss": 0.2114,
+      "step": 17886
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4727718248020655e-06,
+      "loss": 0.2194,
+      "step": 17887
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4717061928336635e-06,
+      "loss": 0.2283,
+      "step": 17888
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4706406255475366e-06,
+      "loss": 0.2021,
+      "step": 17889
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.469575122958917e-06,
+      "loss": 0.238,
+      "step": 17890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468509685083017e-06,
+      "loss": 0.2082,
+      "step": 17891
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.467444311935057e-06,
+      "loss": 0.2136,
+      "step": 17892
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466379003530259e-06,
+      "loss": 0.2305,
+      "step": 17893
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.465313759883833e-06,
+      "loss": 0.2134,
+      "step": 17894
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464248581011002e-06,
+      "loss": 0.2162,
+      "step": 17895
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.46318346692698e-06,
+      "loss": 0.2321,
+      "step": 17896
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.462118417646975e-06,
+      "loss": 0.249,
+      "step": 17897
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461053433186207e-06,
+      "loss": 0.2095,
+      "step": 17898
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4599885135598815e-06,
+      "loss": 0.2202,
+      "step": 17899
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458923658783219e-06,
+      "loss": 0.235,
+      "step": 17900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457858868871424e-06,
+      "loss": 0.2095,
+      "step": 17901
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4567941438397066e-06,
+      "loss": 0.2301,
+      "step": 17902
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455729483703276e-06,
+      "loss": 0.2198,
+      "step": 17903
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4546648884773346e-06,
+      "loss": 0.2068,
+      "step": 17904
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453600358177099e-06,
+      "loss": 0.2485,
+      "step": 17905
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4525358928177655e-06,
+      "loss": 0.2318,
+      "step": 17906
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451471492414549e-06,
+      "loss": 0.2188,
+      "step": 17907
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.450407156982644e-06,
+      "loss": 0.2189,
+      "step": 17908
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4493428865372565e-06,
+      "loss": 0.208,
+      "step": 17909
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448278681093593e-06,
+      "loss": 0.2109,
+      "step": 17910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4472145406668495e-06,
+      "loss": 0.2169,
+      "step": 17911
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446150465272224e-06,
+      "loss": 0.2028,
+      "step": 17912
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4450864549249235e-06,
+      "loss": 0.2159,
+      "step": 17913
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4440225096401436e-06,
+      "loss": 0.2338,
+      "step": 17914
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4429586294330795e-06,
+      "loss": 0.2275,
+      "step": 17915
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.441894814318924e-06,
+      "loss": 0.247,
+      "step": 17916
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4408310643128835e-06,
+      "loss": 0.2244,
+      "step": 17917
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.439767379430146e-06,
+      "loss": 0.2008,
+      "step": 17918
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438703759685899e-06,
+      "loss": 0.2071,
+      "step": 17919
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.437640205095349e-06,
+      "loss": 0.2299,
+      "step": 17920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436576715673676e-06,
+      "loss": 0.2081,
+      "step": 17921
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.435513291436079e-06,
+      "loss": 0.2201,
+      "step": 17922
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.434449932397745e-06,
+      "loss": 0.209,
+      "step": 17923
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4333866385738584e-06,
+      "loss": 0.2119,
+      "step": 17924
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432323409979617e-06,
+      "loss": 0.2164,
+      "step": 17925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.431260246630197e-06,
+      "loss": 0.1997,
+      "step": 17926
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.430197148540799e-06,
+      "loss": 0.2483,
+      "step": 17927
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429134115726591e-06,
+      "loss": 0.2149,
+      "step": 17928
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.42807114820277e-06,
+      "loss": 0.2066,
+      "step": 17929
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4270082459845174e-06,
+      "loss": 0.2122,
+      "step": 17930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425945409087008e-06,
+      "loss": 0.2289,
+      "step": 17931
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.424882637525434e-06,
+      "loss": 0.2154,
+      "step": 17932
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.423819931314967e-06,
+      "loss": 0.2184,
+      "step": 17933
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.422757290470795e-06,
+      "loss": 0.2519,
+      "step": 17934
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.421694715008092e-06,
+      "loss": 0.2233,
+      "step": 17935
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.420632204942033e-06,
+      "loss": 0.2256,
+      "step": 17936
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419569760287804e-06,
+      "loss": 0.1979,
+      "step": 17937
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418507381060571e-06,
+      "loss": 0.2386,
+      "step": 17938
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.417445067275516e-06,
+      "loss": 0.2191,
+      "step": 17939
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.416382818947812e-06,
+      "loss": 0.2117,
+      "step": 17940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415320636092631e-06,
+      "loss": 0.222,
+      "step": 17941
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.414258518725145e-06,
+      "loss": 0.2161,
+      "step": 17942
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413196466860522e-06,
+      "loss": 0.2001,
+      "step": 17943
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.41213448051394e-06,
+      "loss": 0.215,
+      "step": 17944
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.411072559700563e-06,
+      "loss": 0.2235,
+      "step": 17945
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410010704435558e-06,
+      "loss": 0.2036,
+      "step": 17946
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.408948914734098e-06,
+      "loss": 0.2533,
+      "step": 17947
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407887190611344e-06,
+      "loss": 0.2095,
+      "step": 17948
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406825532082469e-06,
+      "loss": 0.2367,
+      "step": 17949
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405763939162633e-06,
+      "loss": 0.2365,
+      "step": 17950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404702411866996e-06,
+      "loss": 0.2366,
+      "step": 17951
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40364095021073e-06,
+      "loss": 0.2058,
+      "step": 17952
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402579554208992e-06,
+      "loss": 0.2301,
+      "step": 17953
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401518223876943e-06,
+      "loss": 0.221,
+      "step": 17954
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400456959229739e-06,
+      "loss": 0.2168,
+      "step": 17955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.399395760282548e-06,
+      "loss": 0.2151,
+      "step": 17956
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.398334627050523e-06,
+      "loss": 0.2309,
+      "step": 17957
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3972735595488175e-06,
+      "loss": 0.2092,
+      "step": 17958
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396212557792597e-06,
+      "loss": 0.2438,
+      "step": 17959
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.395151621797008e-06,
+      "loss": 0.2151,
+      "step": 17960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.394090751577212e-06,
+      "loss": 0.2279,
+      "step": 17961
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3930299471483605e-06,
+      "loss": 0.2076,
+      "step": 17962
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3919692085256e-06,
+      "loss": 0.208,
+      "step": 17963
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.390908535724093e-06,
+      "loss": 0.215,
+      "step": 17964
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389847928758979e-06,
+      "loss": 0.197,
+      "step": 17965
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.388787387645418e-06,
+      "loss": 0.2121,
+      "step": 17966
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387726912398552e-06,
+      "loss": 0.2041,
+      "step": 17967
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386666503033533e-06,
+      "loss": 0.2382,
+      "step": 17968
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.385606159565504e-06,
+      "loss": 0.23,
+      "step": 17969
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384545882009609e-06,
+      "loss": 0.2342,
+      "step": 17970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3834856703810015e-06,
+      "loss": 0.2165,
+      "step": 17971
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382425524694815e-06,
+      "loss": 0.2122,
+      "step": 17972
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381365444966205e-06,
+      "loss": 0.2534,
+      "step": 17973
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380305431210305e-06,
+      "loss": 0.2355,
+      "step": 17974
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3792454834422546e-06,
+      "loss": 0.2319,
+      "step": 17975
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378185601677203e-06,
+      "loss": 0.2245,
+      "step": 17976
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37712578593028e-06,
+      "loss": 0.2267,
+      "step": 17977
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.376066036216631e-06,
+      "loss": 0.2189,
+      "step": 17978
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3750063525513926e-06,
+      "loss": 0.2221,
+      "step": 17979
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3739467349497e-06,
+      "loss": 0.2319,
+      "step": 17980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372887183426689e-06,
+      "loss": 0.2243,
+      "step": 17981
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37182769799749e-06,
+      "loss": 0.2271,
+      "step": 17982
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.370768278677243e-06,
+      "loss": 0.2052,
+      "step": 17983
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369708925481079e-06,
+      "loss": 0.1989,
+      "step": 17984
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368649638424125e-06,
+      "loss": 0.2141,
+      "step": 17985
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.36759041752152e-06,
+      "loss": 0.2018,
+      "step": 17986
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3665312627883855e-06,
+      "loss": 0.2093,
+      "step": 17987
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365472174239861e-06,
+      "loss": 0.2301,
+      "step": 17988
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.364413151891068e-06,
+      "loss": 0.2374,
+      "step": 17989
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363354195757129e-06,
+      "loss": 0.2044,
+      "step": 17990
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.36229530585318e-06,
+      "loss": 0.1932,
+      "step": 17991
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361236482194338e-06,
+      "loss": 0.2294,
+      "step": 17992
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3601777247957375e-06,
+      "loss": 0.2185,
+      "step": 17993
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.359119033672488e-06,
+      "loss": 0.202,
+      "step": 17994
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358060408839724e-06,
+      "loss": 0.2106,
+      "step": 17995
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35700185031256e-06,
+      "loss": 0.2323,
+      "step": 17996
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.355943358106116e-06,
+      "loss": 0.2261,
+      "step": 17997
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354884932235518e-06,
+      "loss": 0.2467,
+      "step": 17998
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353826572715875e-06,
+      "loss": 0.2236,
+      "step": 17999
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 0.2288,
+      "step": 18000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35171005278995e-06,
+      "loss": 0.2165,
+      "step": 18001
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3506518924138915e-06,
+      "loss": 0.2437,
+      "step": 18002
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.349593798449263e-06,
+      "loss": 0.2092,
+      "step": 18003
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34853577091117e-06,
+      "loss": 0.2402,
+      "step": 18004
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347477809814732e-06,
+      "loss": 0.2248,
+      "step": 18005
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3464199151750585e-06,
+      "loss": 0.1989,
+      "step": 18006
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34536208700726e-06,
+      "loss": 0.2079,
+      "step": 18007
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344304325326447e-06,
+      "loss": 0.2348,
+      "step": 18008
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.343246630147725e-06,
+      "loss": 0.2218,
+      "step": 18009
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.342189001486208e-06,
+      "loss": 0.2088,
+      "step": 18010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341131439356997e-06,
+      "loss": 0.2357,
+      "step": 18011
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.340073943775206e-06,
+      "loss": 0.237,
+      "step": 18012
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339016514755936e-06,
+      "loss": 0.2162,
+      "step": 18013
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337959152314287e-06,
+      "loss": 0.2289,
+      "step": 18014
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.336901856465372e-06,
+      "loss": 0.2036,
+      "step": 18015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.335844627224287e-06,
+      "loss": 0.203,
+      "step": 18016
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.334787464606133e-06,
+      "loss": 0.2599,
+      "step": 18017
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3337303686260144e-06,
+      "loss": 0.2213,
+      "step": 18018
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3326733392990295e-06,
+      "loss": 0.2218,
+      "step": 18019
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3316163766402765e-06,
+      "loss": 0.2162,
+      "step": 18020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.330559480664848e-06,
+      "loss": 0.2348,
+      "step": 18021
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32950265138785e-06,
+      "loss": 0.2358,
+      "step": 18022
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.328445888824374e-06,
+      "loss": 0.2211,
+      "step": 18023
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32738919298951e-06,
+      "loss": 0.2071,
+      "step": 18024
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326332563898361e-06,
+      "loss": 0.2395,
+      "step": 18025
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.325276001566011e-06,
+      "loss": 0.1819,
+      "step": 18026
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.324219506007561e-06,
+      "loss": 0.201,
+      "step": 18027
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.323163077238096e-06,
+      "loss": 0.2276,
+      "step": 18028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.322106715272706e-06,
+      "loss": 0.2409,
+      "step": 18029
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321050420126483e-06,
+      "loss": 0.2153,
+      "step": 18030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.319994191814511e-06,
+      "loss": 0.2025,
+      "step": 18031
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.318938030351888e-06,
+      "loss": 0.2221,
+      "step": 18032
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.317881935753684e-06,
+      "loss": 0.2107,
+      "step": 18033
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.316825908034996e-06,
+      "loss": 0.2315,
+      "step": 18034
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.315769947210905e-06,
+      "loss": 0.2573,
+      "step": 18035
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3147140532964905e-06,
+      "loss": 0.2093,
+      "step": 18036
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.313658226306841e-06,
+      "loss": 0.2325,
+      "step": 18037
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312602466257034e-06,
+      "loss": 0.1995,
+      "step": 18038
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311546773162155e-06,
+      "loss": 0.2364,
+      "step": 18039
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310491147037279e-06,
+      "loss": 0.2102,
+      "step": 18040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.309435587897482e-06,
+      "loss": 0.232,
+      "step": 18041
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30838009575785e-06,
+      "loss": 0.2103,
+      "step": 18042
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30732467063345e-06,
+      "loss": 0.2262,
+      "step": 18043
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.306269312539367e-06,
+      "loss": 0.22,
+      "step": 18044
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305214021490672e-06,
+      "loss": 0.2288,
+      "step": 18045
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304158797502439e-06,
+      "loss": 0.2267,
+      "step": 18046
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.303103640589739e-06,
+      "loss": 0.2083,
+      "step": 18047
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30204855076764e-06,
+      "loss": 0.2246,
+      "step": 18048
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3009935280512234e-06,
+      "loss": 0.2306,
+      "step": 18049
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.29993857245555e-06,
+      "loss": 0.2032,
+      "step": 18050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298883683995697e-06,
+      "loss": 0.2095,
+      "step": 18051
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.297828862686727e-06,
+      "loss": 0.2059,
+      "step": 18052
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296774108543703e-06,
+      "loss": 0.2413,
+      "step": 18053
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2957194215817024e-06,
+      "loss": 0.2284,
+      "step": 18054
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.294664801815783e-06,
+      "loss": 0.2073,
+      "step": 18055
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.293610249261008e-06,
+      "loss": 0.198,
+      "step": 18056
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.292555763932446e-06,
+      "loss": 0.2067,
+      "step": 18057
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.291501345845151e-06,
+      "loss": 0.2277,
+      "step": 18058
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290446995014198e-06,
+      "loss": 0.2156,
+      "step": 18059
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2893927114546314e-06,
+      "loss": 0.2089,
+      "step": 18060
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.288338495181522e-06,
+      "loss": 0.2293,
+      "step": 18061
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287284346209924e-06,
+      "loss": 0.2155,
+      "step": 18062
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286230264554892e-06,
+      "loss": 0.2281,
+      "step": 18063
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2851762502314906e-06,
+      "loss": 0.2439,
+      "step": 18064
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284122303254766e-06,
+      "loss": 0.221,
+      "step": 18065
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283068423639781e-06,
+      "loss": 0.2301,
+      "step": 18066
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.282014611401586e-06,
+      "loss": 0.2051,
+      "step": 18067
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.280960866555229e-06,
+      "loss": 0.1801,
+      "step": 18068
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2799071891157705e-06,
+      "loss": 0.2053,
+      "step": 18069
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278853579098253e-06,
+      "loss": 0.1858,
+      "step": 18070
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277800036517733e-06,
+      "loss": 0.2035,
+      "step": 18071
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.276746561389258e-06,
+      "loss": 0.2253,
+      "step": 18072
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.275693153727873e-06,
+      "loss": 0.2182,
+      "step": 18073
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274639813548627e-06,
+      "loss": 0.1908,
+      "step": 18074
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.273586540866561e-06,
+      "loss": 0.1766,
+      "step": 18075
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.272533335696728e-06,
+      "loss": 0.1979,
+      "step": 18076
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.271480198054164e-06,
+      "loss": 0.1948,
+      "step": 18077
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.270427127953921e-06,
+      "loss": 0.1732,
+      "step": 18078
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.269374125411036e-06,
+      "loss": 0.1877,
+      "step": 18079
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.268321190440546e-06,
+      "loss": 0.201,
+      "step": 18080
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.267268323057502e-06,
+      "loss": 0.1973,
+      "step": 18081
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.26621552327693e-06,
+      "loss": 0.2001,
+      "step": 18082
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2651627911138804e-06,
+      "loss": 0.2049,
+      "step": 18083
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.264110126583384e-06,
+      "loss": 0.1671,
+      "step": 18084
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2630575297004785e-06,
+      "loss": 0.2041,
+      "step": 18085
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.262005000480199e-06,
+      "loss": 0.1811,
+      "step": 18086
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.260952538937576e-06,
+      "loss": 0.1932,
+      "step": 18087
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2599001450876495e-06,
+      "loss": 0.1895,
+      "step": 18088
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25884781894545e-06,
+      "loss": 0.1869,
+      "step": 18089
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257795560526005e-06,
+      "loss": 0.1726,
+      "step": 18090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2567433698443505e-06,
+      "loss": 0.1825,
+      "step": 18091
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25569124691551e-06,
+      "loss": 0.1956,
+      "step": 18092
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.254639191754518e-06,
+      "loss": 0.1775,
+      "step": 18093
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2535872043764025e-06,
+      "loss": 0.1957,
+      "step": 18094
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.252535284796181e-06,
+      "loss": 0.1831,
+      "step": 18095
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.251483433028891e-06,
+      "loss": 0.1852,
+      "step": 18096
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.250431649089547e-06,
+      "loss": 0.1849,
+      "step": 18097
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249379932993185e-06,
+      "loss": 0.1898,
+      "step": 18098
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.248328284754813e-06,
+      "loss": 0.1809,
+      "step": 18099
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.247276704389463e-06,
+      "loss": 0.1757,
+      "step": 18100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246225191912153e-06,
+      "loss": 0.1796,
+      "step": 18101
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.245173747337898e-06,
+      "loss": 0.1829,
+      "step": 18102
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.244122370681726e-06,
+      "loss": 0.1878,
+      "step": 18103
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.243071061958647e-06,
+      "loss": 0.1813,
+      "step": 18104
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.242019821183684e-06,
+      "loss": 0.1864,
+      "step": 18105
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240968648371849e-06,
+      "loss": 0.1839,
+      "step": 18106
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.239917543538156e-06,
+      "loss": 0.1789,
+      "step": 18107
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.238866506697624e-06,
+      "loss": 0.1815,
+      "step": 18108
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.23781553786526e-06,
+      "loss": 0.1987,
+      "step": 18109
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.236764637056082e-06,
+      "loss": 0.1917,
+      "step": 18110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2357138042851005e-06,
+      "loss": 0.1832,
+      "step": 18111
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.234663039567322e-06,
+      "loss": 0.1759,
+      "step": 18112
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.233612342917758e-06,
+      "loss": 0.1807,
+      "step": 18113
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.23256171435141e-06,
+      "loss": 0.1777,
+      "step": 18114
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231511153883296e-06,
+      "loss": 0.1847,
+      "step": 18115
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2304606615284135e-06,
+      "loss": 0.1839,
+      "step": 18116
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.229410237301775e-06,
+      "loss": 0.1781,
+      "step": 18117
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.228359881218382e-06,
+      "loss": 0.1724,
+      "step": 18118
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227309593293233e-06,
+      "loss": 0.1948,
+      "step": 18119
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.226259373541338e-06,
+      "loss": 0.1908,
+      "step": 18120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.225209221977695e-06,
+      "loss": 0.1954,
+      "step": 18121
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2241591386172995e-06,
+      "loss": 0.1899,
+      "step": 18122
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.223109123475161e-06,
+      "loss": 0.1768,
+      "step": 18123
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2220591765662674e-06,
+      "loss": 0.1916,
+      "step": 18124
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2210092979056305e-06,
+      "loss": 0.1799,
+      "step": 18125
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.219959487508229e-06,
+      "loss": 0.1993,
+      "step": 18126
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.218909745389071e-06,
+      "loss": 0.1849,
+      "step": 18127
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2178600715631475e-06,
+      "loss": 0.1823,
+      "step": 18128
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216810466045448e-06,
+      "loss": 0.1875,
+      "step": 18129
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.215760928850972e-06,
+      "loss": 0.1835,
+      "step": 18130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214711459994703e-06,
+      "loss": 0.1788,
+      "step": 18131
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.213662059491643e-06,
+      "loss": 0.1827,
+      "step": 18132
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212612727356774e-06,
+      "loss": 0.2042,
+      "step": 18133
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2115634636050826e-06,
+      "loss": 0.1602,
+      "step": 18134
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.210514268251563e-06,
+      "loss": 0.192,
+      "step": 18135
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209465141311194e-06,
+      "loss": 0.1848,
+      "step": 18136
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.208416082798972e-06,
+      "loss": 0.2084,
+      "step": 18137
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.207367092729875e-06,
+      "loss": 0.1729,
+      "step": 18138
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206318171118887e-06,
+      "loss": 0.1935,
+      "step": 18139
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.205269317980992e-06,
+      "loss": 0.1689,
+      "step": 18140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.204220533331168e-06,
+      "loss": 0.1949,
+      "step": 18141
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.203171817184403e-06,
+      "loss": 0.1903,
+      "step": 18142
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.202123169555668e-06,
+      "loss": 0.185,
+      "step": 18143
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.201074590459954e-06,
+      "loss": 0.1801,
+      "step": 18144
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2000260799122295e-06,
+      "loss": 0.2063,
+      "step": 18145
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.198977637927471e-06,
+      "loss": 0.1754,
+      "step": 18146
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.19792926452066e-06,
+      "loss": 0.1756,
+      "step": 18147
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.196880959706767e-06,
+      "loss": 0.1939,
+      "step": 18148
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1958327235007715e-06,
+      "loss": 0.1965,
+      "step": 18149
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.194784555917641e-06,
+      "loss": 0.1804,
+      "step": 18150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1937364569723515e-06,
+      "loss": 0.2,
+      "step": 18151
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1926884266798714e-06,
+      "loss": 0.1883,
+      "step": 18152
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191640465055167e-06,
+      "loss": 0.2038,
+      "step": 18153
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1905925721132175e-06,
+      "loss": 0.1738,
+      "step": 18154
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.189544747868981e-06,
+      "loss": 0.1816,
+      "step": 18155
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.188496992337433e-06,
+      "loss": 0.1848,
+      "step": 18156
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.187449305533535e-06,
+      "loss": 0.1976,
+      "step": 18157
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.186401687472249e-06,
+      "loss": 0.1786,
+      "step": 18158
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185354138168548e-06,
+      "loss": 0.1895,
+      "step": 18159
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18430665763739e-06,
+      "loss": 0.1832,
+      "step": 18160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.183259245893734e-06,
+      "loss": 0.1694,
+      "step": 18161
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.182211902952549e-06,
+      "loss": 0.185,
+      "step": 18162
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1811646288287875e-06,
+      "loss": 0.1885,
+      "step": 18163
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.180117423537419e-06,
+      "loss": 0.196,
+      "step": 18164
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179070287093388e-06,
+      "loss": 0.1925,
+      "step": 18165
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.178023219511663e-06,
+      "loss": 0.1833,
+      "step": 18166
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.176976220807197e-06,
+      "loss": 0.187,
+      "step": 18167
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.175929290994941e-06,
+      "loss": 0.1932,
+      "step": 18168
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.174882430089856e-06,
+      "loss": 0.1898,
+      "step": 18169
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.17383563810689e-06,
+      "loss": 0.1854,
+      "step": 18170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172788915061002e-06,
+      "loss": 0.1795,
+      "step": 18171
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.171742260967138e-06,
+      "loss": 0.1939,
+      "step": 18172
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170695675840247e-06,
+      "loss": 0.19,
+      "step": 18173
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169649159695285e-06,
+      "loss": 0.1899,
+      "step": 18174
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.168602712547193e-06,
+      "loss": 0.1837,
+      "step": 18175
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167556334410926e-06,
+      "loss": 0.1831,
+      "step": 18176
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166510025301427e-06,
+      "loss": 0.2026,
+      "step": 18177
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.165463785233641e-06,
+      "loss": 0.2,
+      "step": 18178
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.164417614222512e-06,
+      "loss": 0.1718,
+      "step": 18179
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163371512282982e-06,
+      "loss": 0.1922,
+      "step": 18180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.162325479429999e-06,
+      "loss": 0.1844,
+      "step": 18181
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1612795156784975e-06,
+      "loss": 0.1896,
+      "step": 18182
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1602336210434266e-06,
+      "loss": 0.1853,
+      "step": 18183
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1591877955397205e-06,
+      "loss": 0.1862,
+      "step": 18184
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158142039182315e-06,
+      "loss": 0.1745,
+      "step": 18185
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.157096351986155e-06,
+      "loss": 0.2095,
+      "step": 18186
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.15605073396617e-06,
+      "loss": 0.2285,
+      "step": 18187
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.155005185137303e-06,
+      "loss": 0.1809,
+      "step": 18188
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.153959705514484e-06,
+      "loss": 0.1792,
+      "step": 18189
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.152914295112649e-06,
+      "loss": 0.1778,
+      "step": 18190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.151868953946728e-06,
+      "loss": 0.1872,
+      "step": 18191
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150823682031649e-06,
+      "loss": 0.1843,
+      "step": 18192
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149778479382352e-06,
+      "loss": 0.2111,
+      "step": 18193
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.148733346013763e-06,
+      "loss": 0.1803,
+      "step": 18194
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1476882819408054e-06,
+      "loss": 0.1867,
+      "step": 18195
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1466432871784165e-06,
+      "loss": 0.1964,
+      "step": 18196
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.145598361741513e-06,
+      "loss": 0.1688,
+      "step": 18197
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.144553505645029e-06,
+      "loss": 0.1853,
+      "step": 18198
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143508718903887e-06,
+      "loss": 0.1747,
+      "step": 18199
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1424640015330044e-06,
+      "loss": 0.1842,
+      "step": 18200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1414193535473135e-06,
+      "loss": 0.1842,
+      "step": 18201
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.140374774961727e-06,
+      "loss": 0.1871,
+      "step": 18202
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.139330265791176e-06,
+      "loss": 0.1887,
+      "step": 18203
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.138285826050575e-06,
+      "loss": 0.174,
+      "step": 18204
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.137241455754841e-06,
+      "loss": 0.2048,
+      "step": 18205
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136197154918894e-06,
+      "loss": 0.1886,
+      "step": 18206
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.135152923557647e-06,
+      "loss": 0.1795,
+      "step": 18207
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.134108761686023e-06,
+      "loss": 0.1927,
+      "step": 18208
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.133064669318928e-06,
+      "loss": 0.192,
+      "step": 18209
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132020646471285e-06,
+      "loss": 0.1994,
+      "step": 18210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.130976693158003e-06,
+      "loss": 0.1895,
+      "step": 18211
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.129932809393989e-06,
+      "loss": 0.1654,
+      "step": 18212
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.128888995194161e-06,
+      "loss": 0.1899,
+      "step": 18213
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127845250573424e-06,
+      "loss": 0.1805,
+      "step": 18214
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.126801575546693e-06,
+      "loss": 0.1812,
+      "step": 18215
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.125757970128871e-06,
+      "loss": 0.1759,
+      "step": 18216
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.124714434334867e-06,
+      "loss": 0.198,
+      "step": 18217
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123670968179585e-06,
+      "loss": 0.1718,
+      "step": 18218
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.122627571677927e-06,
+      "loss": 0.185,
+      "step": 18219
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121584244844804e-06,
+      "loss": 0.1734,
+      "step": 18220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.120540987695111e-06,
+      "loss": 0.2041,
+      "step": 18221
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.11949780024376e-06,
+      "loss": 0.1759,
+      "step": 18222
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.118454682505645e-06,
+      "loss": 0.2095,
+      "step": 18223
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.117411634495663e-06,
+      "loss": 0.1857,
+      "step": 18224
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.116368656228721e-06,
+      "loss": 0.1843,
+      "step": 18225
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115325747719715e-06,
+      "loss": 0.1735,
+      "step": 18226
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1142829089835346e-06,
+      "loss": 0.1973,
+      "step": 18227
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113240140035085e-06,
+      "loss": 0.1767,
+      "step": 18228
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.112197440889254e-06,
+      "loss": 0.1827,
+      "step": 18229
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.111154811560946e-06,
+      "loss": 0.1809,
+      "step": 18230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.110112252065041e-06,
+      "loss": 0.1846,
+      "step": 18231
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109069762416438e-06,
+      "loss": 0.1941,
+      "step": 18232
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108027342630027e-06,
+      "loss": 0.1941,
+      "step": 18233
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106984992720694e-06,
+      "loss": 0.1933,
+      "step": 18234
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.105942712703336e-06,
+      "loss": 0.1849,
+      "step": 18235
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.104900502592834e-06,
+      "loss": 0.1908,
+      "step": 18236
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1038583624040795e-06,
+      "loss": 0.1779,
+      "step": 18237
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102816292151955e-06,
+      "loss": 0.1912,
+      "step": 18238
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.101774291851346e-06,
+      "loss": 0.1894,
+      "step": 18239
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.10073236151714e-06,
+      "loss": 0.2105,
+      "step": 18240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099690501164214e-06,
+      "loss": 0.1873,
+      "step": 18241
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.098648710807457e-06,
+      "loss": 0.1842,
+      "step": 18242
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.097606990461745e-06,
+      "loss": 0.1813,
+      "step": 18243
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0965653401419616e-06,
+      "loss": 0.1838,
+      "step": 18244
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.095523759862982e-06,
+      "loss": 0.1925,
+      "step": 18245
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094482249639683e-06,
+      "loss": 0.1731,
+      "step": 18246
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.093440809486946e-06,
+      "loss": 0.1721,
+      "step": 18247
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092399439419642e-06,
+      "loss": 0.1823,
+      "step": 18248
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0913581394526535e-06,
+      "loss": 0.1818,
+      "step": 18249
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.09031690960085e-06,
+      "loss": 0.1692,
+      "step": 18250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0892757498791005e-06,
+      "loss": 0.1839,
+      "step": 18251
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.088234660302285e-06,
+      "loss": 0.17,
+      "step": 18252
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0871936408852655e-06,
+      "loss": 0.1947,
+      "step": 18253
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08615269164292e-06,
+      "loss": 0.1616,
+      "step": 18254
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.085111812590117e-06,
+      "loss": 0.1788,
+      "step": 18255
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08407100374172e-06,
+      "loss": 0.205,
+      "step": 18256
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.083030265112598e-06,
+      "loss": 0.1804,
+      "step": 18257
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081989596717614e-06,
+      "loss": 0.1971,
+      "step": 18258
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080948998571638e-06,
+      "loss": 0.196,
+      "step": 18259
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079908470689527e-06,
+      "loss": 0.2002,
+      "step": 18260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.078868013086153e-06,
+      "loss": 0.1957,
+      "step": 18261
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077827625776375e-06,
+      "loss": 0.1861,
+      "step": 18262
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0767873087750465e-06,
+      "loss": 0.1744,
+      "step": 18263
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.075747062097037e-06,
+      "loss": 0.1816,
+      "step": 18264
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074706885757203e-06,
+      "loss": 0.1738,
+      "step": 18265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073666779770396e-06,
+      "loss": 0.1984,
+      "step": 18266
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.072626744151482e-06,
+      "loss": 0.1794,
+      "step": 18267
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.07158677891531e-06,
+      "loss": 0.1827,
+      "step": 18268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.070546884076744e-06,
+      "loss": 0.1719,
+      "step": 18269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0695070596506244e-06,
+      "loss": 0.1913,
+      "step": 18270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068467305651815e-06,
+      "loss": 0.1795,
+      "step": 18271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.067427622095165e-06,
+      "loss": 0.1685,
+      "step": 18272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0663880089955195e-06,
+      "loss": 0.1782,
+      "step": 18273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065348466367737e-06,
+      "loss": 0.1932,
+      "step": 18274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0643089942266586e-06,
+      "loss": 0.1859,
+      "step": 18275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063269592587139e-06,
+      "loss": 0.1905,
+      "step": 18276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.062230261464023e-06,
+      "loss": 0.1848,
+      "step": 18277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.061191000872152e-06,
+      "loss": 0.1888,
+      "step": 18278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.060151810826378e-06,
+      "loss": 0.1875,
+      "step": 18279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.059112691341536e-06,
+      "loss": 0.1839,
+      "step": 18280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0580736424324796e-06,
+      "loss": 0.1625,
+      "step": 18281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.057034664114044e-06,
+      "loss": 0.2171,
+      "step": 18282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.055995756401072e-06,
+      "loss": 0.2042,
+      "step": 18283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.054956919308401e-06,
+      "loss": 0.1759,
+      "step": 18284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053918152850868e-06,
+      "loss": 0.178,
+      "step": 18285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.052879457043318e-06,
+      "loss": 0.1919,
+      "step": 18286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.05184083190058e-06,
+      "loss": 0.178,
+      "step": 18287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0508022774374964e-06,
+      "loss": 0.1774,
+      "step": 18288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.049763793668899e-06,
+      "loss": 0.1779,
+      "step": 18289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.048725380609618e-06,
+      "loss": 0.1741,
+      "step": 18290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0476870382744935e-06,
+      "loss": 0.1712,
+      "step": 18291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.046648766678349e-06,
+      "loss": 0.1846,
+      "step": 18292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.045610565836023e-06,
+      "loss": 0.1767,
+      "step": 18293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044572435762342e-06,
+      "loss": 0.1847,
+      "step": 18294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.043534376472132e-06,
+      "loss": 0.1777,
+      "step": 18295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.04249638798023e-06,
+      "loss": 0.1837,
+      "step": 18296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.041458470301448e-06,
+      "loss": 0.1741,
+      "step": 18297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040420623450622e-06,
+      "loss": 0.1861,
+      "step": 18298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.039382847442575e-06,
+      "loss": 0.1976,
+      "step": 18299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.038345142292127e-06,
+      "loss": 0.1735,
+      "step": 18300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.037307508014105e-06,
+      "loss": 0.1876,
+      "step": 18301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.036269944623326e-06,
+      "loss": 0.1819,
+      "step": 18302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.035232452134617e-06,
+      "loss": 0.1768,
+      "step": 18303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034195030562794e-06,
+      "loss": 0.2002,
+      "step": 18304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.033157679922672e-06,
+      "loss": 0.1862,
+      "step": 18305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.032120400229077e-06,
+      "loss": 0.2098,
+      "step": 18306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.031083191496815e-06,
+      "loss": 0.1709,
+      "step": 18307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.030046053740711e-06,
+      "loss": 0.1778,
+      "step": 18308
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.029008986975577e-06,
+      "loss": 0.1677,
+      "step": 18309
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.027971991216225e-06,
+      "loss": 0.187,
+      "step": 18310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.026935066477467e-06,
+      "loss": 0.1753,
+      "step": 18311
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0258982127741105e-06,
+      "loss": 0.2067,
+      "step": 18312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.024861430120975e-06,
+      "loss": 0.1911,
+      "step": 18313
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.023824718532863e-06,
+      "loss": 0.1938,
+      "step": 18314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022788078024587e-06,
+      "loss": 0.1905,
+      "step": 18315
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0217515086109545e-06,
+      "loss": 0.1853,
+      "step": 18316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.020715010306766e-06,
+      "loss": 0.1827,
+      "step": 18317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.019678583126835e-06,
+      "loss": 0.1715,
+      "step": 18318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0186422270859566e-06,
+      "loss": 0.1934,
+      "step": 18319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017605942198944e-06,
+      "loss": 0.1659,
+      "step": 18320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.016569728480596e-06,
+      "loss": 0.1786,
+      "step": 18321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.015533585945713e-06,
+      "loss": 0.187,
+      "step": 18322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.014497514609095e-06,
+      "loss": 0.1908,
+      "step": 18323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013461514485536e-06,
+      "loss": 0.1859,
+      "step": 18324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.012425585589845e-06,
+      "loss": 0.1792,
+      "step": 18325
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0113897279368105e-06,
+      "loss": 0.1691,
+      "step": 18326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.010353941541235e-06,
+      "loss": 0.2097,
+      "step": 18327
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.009318226417912e-06,
+      "loss": 0.1675,
+      "step": 18328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008282582581629e-06,
+      "loss": 0.1995,
+      "step": 18329
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007247010047189e-06,
+      "loss": 0.1822,
+      "step": 18330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.006211508829382e-06,
+      "loss": 0.1795,
+      "step": 18331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0051760789429906e-06,
+      "loss": 0.1846,
+      "step": 18332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.004140720402817e-06,
+      "loss": 0.165,
+      "step": 18333
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.003105433223638e-06,
+      "loss": 0.1845,
+      "step": 18334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0020702174202586e-06,
+      "loss": 0.1876,
+      "step": 18335
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.001035073007448e-06,
+      "loss": 0.1656,
+      "step": 18336
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.1692,
+      "step": 18337
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.998964998412704e-06,
+      "loss": 0.2033,
+      "step": 18338
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.997930068260334e-06,
+      "loss": 0.1805,
+      "step": 18339
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996895209557682e-06,
+      "loss": 0.1847,
+      "step": 18340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9958604223195225e-06,
+      "loss": 0.1849,
+      "step": 18341
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994825706560644e-06,
+      "loss": 0.1853,
+      "step": 18342
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9937910622958215e-06,
+      "loss": 0.1797,
+      "step": 18343
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.992756489539833e-06,
+      "loss": 0.1747,
+      "step": 18344
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.991721988307462e-06,
+      "loss": 0.1894,
+      "step": 18345
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9906875586134775e-06,
+      "loss": 0.1641,
+      "step": 18346
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9896532004726625e-06,
+      "loss": 0.1695,
+      "step": 18347
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.98861891389979e-06,
+      "loss": 0.1776,
+      "step": 18348
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987584698909632e-06,
+      "loss": 0.1981,
+      "step": 18349
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.986550555516962e-06,
+      "loss": 0.1993,
+      "step": 18350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985516483736549e-06,
+      "loss": 0.1982,
+      "step": 18351
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984482483583168e-06,
+      "loss": 0.1903,
+      "step": 18352
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.983448555071585e-06,
+      "loss": 0.1878,
+      "step": 18353
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982414698216574e-06,
+      "loss": 0.1796,
+      "step": 18354
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9813809130329e-06,
+      "loss": 0.1918,
+      "step": 18355
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.980347199535325e-06,
+      "loss": 0.1851,
+      "step": 18356
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.979313557738621e-06,
+      "loss": 0.1812,
+      "step": 18357
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978279987657547e-06,
+      "loss": 0.1961,
+      "step": 18358
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.977246489306874e-06,
+      "loss": 0.184,
+      "step": 18359
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97621306270136e-06,
+      "loss": 0.1872,
+      "step": 18360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.975179707855763e-06,
+      "loss": 0.1841,
+      "step": 18361
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974146424784855e-06,
+      "loss": 0.1757,
+      "step": 18362
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.973113213503379e-06,
+      "loss": 0.1936,
+      "step": 18363
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.972080074026107e-06,
+      "loss": 0.2094,
+      "step": 18364
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971047006367786e-06,
+      "loss": 0.2045,
+      "step": 18365
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.970014010543182e-06,
+      "loss": 0.1925,
+      "step": 18366
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968981086567048e-06,
+      "loss": 0.2123,
+      "step": 18367
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.967948234454132e-06,
+      "loss": 0.1774,
+      "step": 18368
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9669154542191945e-06,
+      "loss": 0.1859,
+      "step": 18369
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9658827458769855e-06,
+      "loss": 0.1852,
+      "step": 18370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964850109442252e-06,
+      "loss": 0.1902,
+      "step": 18371
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.963817544929752e-06,
+      "loss": 0.202,
+      "step": 18372
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.962785052354226e-06,
+      "loss": 0.1719,
+      "step": 18373
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961752631730431e-06,
+      "loss": 0.1884,
+      "step": 18374
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.960720283073111e-06,
+      "loss": 0.1886,
+      "step": 18375
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9596880063970095e-06,
+      "loss": 0.1742,
+      "step": 18376
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.958655801716874e-06,
+      "loss": 0.1649,
+      "step": 18377
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957623669047443e-06,
+      "loss": 0.1731,
+      "step": 18378
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956591608403469e-06,
+      "loss": 0.1818,
+      "step": 18379
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9555596197996844e-06,
+      "loss": 0.1823,
+      "step": 18380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9545277032508395e-06,
+      "loss": 0.1818,
+      "step": 18381
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.953495858771668e-06,
+      "loss": 0.1727,
+      "step": 18382
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.952464086376909e-06,
+      "loss": 0.1997,
+      "step": 18383
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.951432386081304e-06,
+      "loss": 0.1775,
+      "step": 18384
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.950400757899585e-06,
+      "loss": 0.1929,
+      "step": 18385
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949369201846493e-06,
+      "loss": 0.1737,
+      "step": 18386
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.948337717936761e-06,
+      "loss": 0.1773,
+      "step": 18387
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9473063061851214e-06,
+      "loss": 0.1886,
+      "step": 18388
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.946274966606307e-06,
+      "loss": 0.1809,
+      "step": 18389
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.945243699215048e-06,
+      "loss": 0.1663,
+      "step": 18390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.944212504026079e-06,
+      "loss": 0.1788,
+      "step": 18391
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943181381054124e-06,
+      "loss": 0.1764,
+      "step": 18392
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.94215033031392e-06,
+      "loss": 0.1789,
+      "step": 18393
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941119351820188e-06,
+      "loss": 0.193,
+      "step": 18394
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.940088445587654e-06,
+      "loss": 0.1732,
+      "step": 18395
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.939057611631049e-06,
+      "loss": 0.1747,
+      "step": 18396
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938026849965092e-06,
+      "loss": 0.1832,
+      "step": 18397
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9369961606045104e-06,
+      "loss": 0.1838,
+      "step": 18398
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.935965543564026e-06,
+      "loss": 0.2016,
+      "step": 18399
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.934934998858355e-06,
+      "loss": 0.1731,
+      "step": 18400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933904526502229e-06,
+      "loss": 0.1822,
+      "step": 18401
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932874126510353e-06,
+      "loss": 0.176,
+      "step": 18402
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931843798897456e-06,
+      "loss": 0.177,
+      "step": 18403
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9308135436782514e-06,
+      "loss": 0.166,
+      "step": 18404
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.929783360867452e-06,
+      "loss": 0.1712,
+      "step": 18405
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9287532504797795e-06,
+      "loss": 0.1812,
+      "step": 18406
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927723212529942e-06,
+      "loss": 0.185,
+      "step": 18407
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92669324703266e-06,
+      "loss": 0.2162,
+      "step": 18408
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.925663354002641e-06,
+      "loss": 0.1923,
+      "step": 18409
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.924633533454591e-06,
+      "loss": 0.1787,
+      "step": 18410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92360378540323e-06,
+      "loss": 0.1786,
+      "step": 18411
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92257410986326e-06,
+      "loss": 0.1852,
+      "step": 18412
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9215445068493925e-06,
+      "loss": 0.203,
+      "step": 18413
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920514976376334e-06,
+      "loss": 0.2046,
+      "step": 18414
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919485518458789e-06,
+      "loss": 0.1695,
+      "step": 18415
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.918456133111462e-06,
+      "loss": 0.177,
+      "step": 18416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917426820349055e-06,
+      "loss": 0.1941,
+      "step": 18417
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.916397580186276e-06,
+      "loss": 0.1912,
+      "step": 18418
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915368412637821e-06,
+      "loss": 0.1864,
+      "step": 18419
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9143393177183965e-06,
+      "loss": 0.1896,
+      "step": 18420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.913310295442698e-06,
+      "loss": 0.2008,
+      "step": 18421
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9122813458254225e-06,
+      "loss": 0.1892,
+      "step": 18422
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.911252468881275e-06,
+      "loss": 0.1981,
+      "step": 18423
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.910223664624942e-06,
+      "loss": 0.1952,
+      "step": 18424
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.909194933071128e-06,
+      "loss": 0.1838,
+      "step": 18425
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908166274234525e-06,
+      "loss": 0.1644,
+      "step": 18426
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.907137688129824e-06,
+      "loss": 0.209,
+      "step": 18427
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906109174771718e-06,
+      "loss": 0.21,
+      "step": 18428
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.905080734174895e-06,
+      "loss": 0.1771,
+      "step": 18429
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904052366354054e-06,
+      "loss": 0.1911,
+      "step": 18430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.903024071323874e-06,
+      "loss": 0.1665,
+      "step": 18431
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.901995849099052e-06,
+      "loss": 0.1807,
+      "step": 18432
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.900967699694273e-06,
+      "loss": 0.179,
+      "step": 18433
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.899939623124218e-06,
+      "loss": 0.1655,
+      "step": 18434
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898911619403578e-06,
+      "loss": 0.1711,
+      "step": 18435
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.897883688547032e-06,
+      "loss": 0.2083,
+      "step": 18436
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896855830569269e-06,
+      "loss": 0.1888,
+      "step": 18437
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8958280454849685e-06,
+      "loss": 0.1871,
+      "step": 18438
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8948003333088066e-06,
+      "loss": 0.1984,
+      "step": 18439
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.893772694055471e-06,
+      "loss": 0.178,
+      "step": 18440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8927451277396365e-06,
+      "loss": 0.1801,
+      "step": 18441
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891717634375981e-06,
+      "loss": 0.1761,
+      "step": 18442
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890690213979182e-06,
+      "loss": 0.1903,
+      "step": 18443
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88966286656391e-06,
+      "loss": 0.1987,
+      "step": 18444
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.888635592144849e-06,
+      "loss": 0.1672,
+      "step": 18445
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8876083907366634e-06,
+      "loss": 0.182,
+      "step": 18446
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.886581262354034e-06,
+      "loss": 0.1808,
+      "step": 18447
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.885554207011629e-06,
+      "loss": 0.1945,
+      "step": 18448
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884527224724115e-06,
+      "loss": 0.1878,
+      "step": 18449
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.883500315506168e-06,
+      "loss": 0.1877,
+      "step": 18450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.882473479372449e-06,
+      "loss": 0.1691,
+      "step": 18451
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881446716337634e-06,
+      "loss": 0.1928,
+      "step": 18452
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.880420026416386e-06,
+      "loss": 0.1765,
+      "step": 18453
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879393409623368e-06,
+      "loss": 0.1729,
+      "step": 18454
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878366865973246e-06,
+      "loss": 0.1707,
+      "step": 18455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8773403954806785e-06,
+      "loss": 0.1801,
+      "step": 18456
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.876313998160336e-06,
+      "loss": 0.1809,
+      "step": 18457
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875287674026872e-06,
+      "loss": 0.1704,
+      "step": 18458
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.874261423094953e-06,
+      "loss": 0.1841,
+      "step": 18459
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.873235245379235e-06,
+      "loss": 0.1895,
+      "step": 18460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872209140894371e-06,
+      "loss": 0.1831,
+      "step": 18461
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.871183109655029e-06,
+      "loss": 0.1863,
+      "step": 18462
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.870157151675854e-06,
+      "loss": 0.1765,
+      "step": 18463
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869131266971509e-06,
+      "loss": 0.1924,
+      "step": 18464
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868105455556643e-06,
+      "loss": 0.2056,
+      "step": 18465
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.867079717445907e-06,
+      "loss": 0.1837,
+      "step": 18466
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866054052653963e-06,
+      "loss": 0.1907,
+      "step": 18467
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8650284611954476e-06,
+      "loss": 0.178,
+      "step": 18468
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8640029430850215e-06,
+      "loss": 0.1657,
+      "step": 18469
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8629774983373234e-06,
+      "loss": 0.1993,
+      "step": 18470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.86195212696701e-06,
+      "loss": 0.1906,
+      "step": 18471
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.860926828988724e-06,
+      "loss": 0.1805,
+      "step": 18472
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.859901604417106e-06,
+      "loss": 0.1853,
+      "step": 18473
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8588764532668105e-06,
+      "loss": 0.1869,
+      "step": 18474
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.857851375552475e-06,
+      "loss": 0.1607,
+      "step": 18475
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.856826371288737e-06,
+      "loss": 0.174,
+      "step": 18476
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.855801440490248e-06,
+      "loss": 0.1984,
+      "step": 18477
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8547765831716385e-06,
+      "loss": 0.1948,
+      "step": 18478
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853751799347556e-06,
+      "loss": 0.1735,
+      "step": 18479
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.852727089032634e-06,
+      "loss": 0.187,
+      "step": 18480
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85170245224151e-06,
+      "loss": 0.2002,
+      "step": 18481
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85067788898882e-06,
+      "loss": 0.1727,
+      "step": 18482
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.849653399289194e-06,
+      "loss": 0.1973,
+      "step": 18483
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8486289831572765e-06,
+      "loss": 0.1741,
+      "step": 18484
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847604640607689e-06,
+      "loss": 0.1797,
+      "step": 18485
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8465803716550745e-06,
+      "loss": 0.1939,
+      "step": 18486
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845556176314055e-06,
+      "loss": 0.1618,
+      "step": 18487
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844532054599261e-06,
+      "loss": 0.198,
+      "step": 18488
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.843508006525327e-06,
+      "loss": 0.1824,
+      "step": 18489
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842484032106873e-06,
+      "loss": 0.1666,
+      "step": 18490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.841460131358533e-06,
+      "loss": 0.1978,
+      "step": 18491
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.840436304294927e-06,
+      "loss": 0.1829,
+      "step": 18492
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839412550930682e-06,
+      "loss": 0.1996,
+      "step": 18493
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8383888712804205e-06,
+      "loss": 0.1667,
+      "step": 18494
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83736526535876e-06,
+      "loss": 0.1792,
+      "step": 18495
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836341733180331e-06,
+      "loss": 0.1961,
+      "step": 18496
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.835318274759744e-06,
+      "loss": 0.1994,
+      "step": 18497
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8342948901116274e-06,
+      "loss": 0.1961,
+      "step": 18498
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833271579250595e-06,
+      "loss": 0.1855,
+      "step": 18499
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8322483421912595e-06,
+      "loss": 0.167,
+      "step": 18500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.831225178948244e-06,
+      "loss": 0.1733,
+      "step": 18501
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830202089536158e-06,
+      "loss": 0.1719,
+      "step": 18502
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829179073969621e-06,
+      "loss": 0.191,
+      "step": 18503
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.828156132263243e-06,
+      "loss": 0.1762,
+      "step": 18504
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.827133264431631e-06,
+      "loss": 0.1827,
+      "step": 18505
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826110470489403e-06,
+      "loss": 0.1898,
+      "step": 18506
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8250877504511674e-06,
+      "loss": 0.1977,
+      "step": 18507
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824065104331529e-06,
+      "loss": 0.1796,
+      "step": 18508
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.823042532145099e-06,
+      "loss": 0.1829,
+      "step": 18509
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.822020033906477e-06,
+      "loss": 0.1789,
+      "step": 18510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.820997609630278e-06,
+      "loss": 0.1803,
+      "step": 18511
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8199752593310964e-06,
+      "loss": 0.1885,
+      "step": 18512
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.818952983023547e-06,
+      "loss": 0.1932,
+      "step": 18513
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.817930780722224e-06,
+      "loss": 0.2063,
+      "step": 18514
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.816908652441728e-06,
+      "loss": 0.1971,
+      "step": 18515
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.815886598196665e-06,
+      "loss": 0.1861,
+      "step": 18516
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.814864618001627e-06,
+      "loss": 0.2037,
+      "step": 18517
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813842711871218e-06,
+      "loss": 0.1926,
+      "step": 18518
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812820879820034e-06,
+      "loss": 0.1748,
+      "step": 18519
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.811799121862669e-06,
+      "loss": 0.1794,
+      "step": 18520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.810777438013719e-06,
+      "loss": 0.1757,
+      "step": 18521
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.809755828287772e-06,
+      "loss": 0.1792,
+      "step": 18522
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80873429269943e-06,
+      "loss": 0.1815,
+      "step": 18523
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.807712831263278e-06,
+      "loss": 0.1994,
+      "step": 18524
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8066914439939115e-06,
+      "loss": 0.2046,
+      "step": 18525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805670130905917e-06,
+      "loss": 0.18,
+      "step": 18526
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.804648892013881e-06,
+      "loss": 0.1889,
+      "step": 18527
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.803627727332396e-06,
+      "loss": 0.1978,
+      "step": 18528
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8026066368760425e-06,
+      "loss": 0.1744,
+      "step": 18529
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8015856206594124e-06,
+      "loss": 0.1786,
+      "step": 18530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800564678697087e-06,
+      "loss": 0.1712,
+      "step": 18531
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799543811003645e-06,
+      "loss": 0.1808,
+      "step": 18532
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.79852301759368e-06,
+      "loss": 0.1721,
+      "step": 18533
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7975022984817575e-06,
+      "loss": 0.1888,
+      "step": 18534
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.79648165368247e-06,
+      "loss": 0.1739,
+      "step": 18535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.795461083210389e-06,
+      "loss": 0.1754,
+      "step": 18536
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794440587080097e-06,
+      "loss": 0.1863,
+      "step": 18537
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.793420165306171e-06,
+      "loss": 0.2128,
+      "step": 18538
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.792399817903179e-06,
+      "loss": 0.1804,
+      "step": 18539
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.791379544885707e-06,
+      "loss": 0.1847,
+      "step": 18540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.790359346268319e-06,
+      "loss": 0.1785,
+      "step": 18541
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789339222065595e-06,
+      "loss": 0.1902,
+      "step": 18542
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7883191722921044e-06,
+      "loss": 0.1799,
+      "step": 18543
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.787299196962411e-06,
+      "loss": 0.1767,
+      "step": 18544
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.786279296091094e-06,
+      "loss": 0.1825,
+      "step": 18545
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.785259469692717e-06,
+      "loss": 0.1794,
+      "step": 18546
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.784239717781848e-06,
+      "loss": 0.1787,
+      "step": 18547
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.783220040373052e-06,
+      "loss": 0.1819,
+      "step": 18548
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.782200437480891e-06,
+      "loss": 0.1965,
+      "step": 18549
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.781180909119937e-06,
+      "loss": 0.1874,
+      "step": 18550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.780161455304745e-06,
+      "loss": 0.1788,
+      "step": 18551
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779142076049883e-06,
+      "loss": 0.1931,
+      "step": 18552
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.778122771369911e-06,
+      "loss": 0.1913,
+      "step": 18553
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777103541279383e-06,
+      "loss": 0.1945,
+      "step": 18554
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.776084385792866e-06,
+      "loss": 0.1728,
+      "step": 18555
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77506530492491e-06,
+      "loss": 0.1795,
+      "step": 18556
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774046298690078e-06,
+      "loss": 0.2005,
+      "step": 18557
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7730273671029235e-06,
+      "loss": 0.1923,
+      "step": 18558
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.772008510178e-06,
+      "loss": 0.1977,
+      "step": 18559
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.770989727929861e-06,
+      "loss": 0.1925,
+      "step": 18560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.769971020373056e-06,
+      "loss": 0.2158,
+      "step": 18561
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.768952387522141e-06,
+      "loss": 0.1819,
+      "step": 18562
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.767933829391661e-06,
+      "loss": 0.1924,
+      "step": 18563
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.766915345996173e-06,
+      "loss": 0.1778,
+      "step": 18564
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.765896937350221e-06,
+      "loss": 0.1731,
+      "step": 18565
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.764878603468346e-06,
+      "loss": 0.183,
+      "step": 18566
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.763860344365105e-06,
+      "loss": 0.1812,
+      "step": 18567
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762842160055031e-06,
+      "loss": 0.1789,
+      "step": 18568
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.761824050552679e-06,
+      "loss": 0.19,
+      "step": 18569
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.760806015872587e-06,
+      "loss": 0.1705,
+      "step": 18570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759788056029292e-06,
+      "loss": 0.1909,
+      "step": 18571
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758770171037347e-06,
+      "loss": 0.174,
+      "step": 18572
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.757752360911275e-06,
+      "loss": 0.1878,
+      "step": 18573
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7567346256656276e-06,
+      "loss": 0.1723,
+      "step": 18574
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7557169653149324e-06,
+      "loss": 0.2124,
+      "step": 18575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754699379873735e-06,
+      "loss": 0.1837,
+      "step": 18576
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753681869356567e-06,
+      "loss": 0.1937,
+      "step": 18577
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.752664433777957e-06,
+      "loss": 0.1936,
+      "step": 18578
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751647073152448e-06,
+      "loss": 0.1934,
+      "step": 18579
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.750629787494565e-06,
+      "loss": 0.1911,
+      "step": 18580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.749612576818839e-06,
+      "loss": 0.2059,
+      "step": 18581
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.748595441139804e-06,
+      "loss": 0.2014,
+      "step": 18582
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.747578380471983e-06,
+      "loss": 0.1732,
+      "step": 18583
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.746561394829912e-06,
+      "loss": 0.1969,
+      "step": 18584
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745544484228112e-06,
+      "loss": 0.1757,
+      "step": 18585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.74452764868111e-06,
+      "loss": 0.1859,
+      "step": 18586
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7435108882034295e-06,
+      "loss": 0.2015,
+      "step": 18587
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742494202809591e-06,
+      "loss": 0.1866,
+      "step": 18588
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741477592514123e-06,
+      "loss": 0.1792,
+      "step": 18589
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7404610573315425e-06,
+      "loss": 0.1641,
+      "step": 18590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.739444597276374e-06,
+      "loss": 0.184,
+      "step": 18591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.738428212363133e-06,
+      "loss": 0.1805,
+      "step": 18592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737411902606335e-06,
+      "loss": 0.1876,
+      "step": 18593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.736395668020505e-06,
+      "loss": 0.1725,
+      "step": 18594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.73537950862015e-06,
+      "loss": 0.1856,
+      "step": 18595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.734363424419793e-06,
+      "loss": 0.2061,
+      "step": 18596
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733347415433946e-06,
+      "loss": 0.1888,
+      "step": 18597
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732331481677114e-06,
+      "loss": 0.1668,
+      "step": 18598
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7313156231638225e-06,
+      "loss": 0.195,
+      "step": 18599
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.730299839908567e-06,
+      "loss": 0.1755,
+      "step": 18600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.729284131925866e-06,
+      "loss": 0.2052,
+      "step": 18601
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728268499230223e-06,
+      "loss": 0.1627,
+      "step": 18602
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.727252941836153e-06,
+      "loss": 0.1816,
+      "step": 18603
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726237459758155e-06,
+      "loss": 0.1822,
+      "step": 18604
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.725222053010735e-06,
+      "loss": 0.1741,
+      "step": 18605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.724206721608401e-06,
+      "loss": 0.1812,
+      "step": 18606
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723191465565651e-06,
+      "loss": 0.1767,
+      "step": 18607
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722176284896993e-06,
+      "loss": 0.189,
+      "step": 18608
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.721161179616924e-06,
+      "loss": 0.1912,
+      "step": 18609
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720146149739941e-06,
+      "loss": 0.1759,
+      "step": 18610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.719131195280551e-06,
+      "loss": 0.1851,
+      "step": 18611
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.718116316253245e-06,
+      "loss": 0.1907,
+      "step": 18612
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.717101512672522e-06,
+      "loss": 0.1941,
+      "step": 18613
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.716086784552875e-06,
+      "loss": 0.1846,
+      "step": 18614
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715072131908799e-06,
+      "loss": 0.1861,
+      "step": 18615
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7140575547547905e-06,
+      "loss": 0.185,
+      "step": 18616
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.713043053105338e-06,
+      "loss": 0.2089,
+      "step": 18617
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.712028626974936e-06,
+      "loss": 0.1961,
+      "step": 18618
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711014276378073e-06,
+      "loss": 0.1755,
+      "step": 18619
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.710000001329235e-06,
+      "loss": 0.2036,
+      "step": 18620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708985801842917e-06,
+      "loss": 0.1701,
+      "step": 18621
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.707971677933597e-06,
+      "loss": 0.188,
+      "step": 18622
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7069576296157704e-06,
+      "loss": 0.1808,
+      "step": 18623
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.705943656903916e-06,
+      "loss": 0.19,
+      "step": 18624
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.704929759812519e-06,
+      "loss": 0.1948,
+      "step": 18625
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703915938356062e-06,
+      "loss": 0.1767,
+      "step": 18626
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.70290219254902e-06,
+      "loss": 0.2033,
+      "step": 18627
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.701888522405885e-06,
+      "loss": 0.1765,
+      "step": 18628
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700874927941126e-06,
+      "loss": 0.1854,
+      "step": 18629
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.699861409169229e-06,
+      "loss": 0.1723,
+      "step": 18630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.698847966104669e-06,
+      "loss": 0.1935,
+      "step": 18631
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697834598761917e-06,
+      "loss": 0.1734,
+      "step": 18632
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696821307155455e-06,
+      "loss": 0.1985,
+      "step": 18633
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.69580809129975e-06,
+      "loss": 0.1838,
+      "step": 18634
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.694794951209283e-06,
+      "loss": 0.189,
+      "step": 18635
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.693781886898521e-06,
+      "loss": 0.1945,
+      "step": 18636
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.692768898381931e-06,
+      "loss": 0.1908,
+      "step": 18637
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.691755985673995e-06,
+      "loss": 0.1683,
+      "step": 18638
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.690743148789164e-06,
+      "loss": 0.1794,
+      "step": 18639
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.689730387741919e-06,
+      "loss": 0.1641,
+      "step": 18640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688717702546717e-06,
+      "loss": 0.1804,
+      "step": 18641
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687705093218032e-06,
+      "loss": 0.1831,
+      "step": 18642
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.686692559770325e-06,
+      "loss": 0.1735,
+      "step": 18643
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685680102218053e-06,
+      "loss": 0.1525,
+      "step": 18644
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6846677205756885e-06,
+      "loss": 0.1759,
+      "step": 18645
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.683655414857682e-06,
+      "loss": 0.2005,
+      "step": 18646
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6826431850785014e-06,
+      "loss": 0.1769,
+      "step": 18647
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681631031252604e-06,
+      "loss": 0.1828,
+      "step": 18648
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6806189533944405e-06,
+      "loss": 0.1966,
+      "step": 18649
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.679606951518477e-06,
+      "loss": 0.1899,
+      "step": 18650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.678595025639165e-06,
+      "loss": 0.1775,
+      "step": 18651
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.677583175770958e-06,
+      "loss": 0.1927,
+      "step": 18652
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.67657140192831e-06,
+      "loss": 0.1796,
+      "step": 18653
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675559704125669e-06,
+      "loss": 0.1889,
+      "step": 18654
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.674548082377495e-06,
+      "loss": 0.1838,
+      "step": 18655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.673536536698228e-06,
+      "loss": 0.1669,
+      "step": 18656
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.672525067102326e-06,
+      "loss": 0.1843,
+      "step": 18657
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6715136736042335e-06,
+      "loss": 0.1883,
+      "step": 18658
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670502356218392e-06,
+      "loss": 0.2004,
+      "step": 18659
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669491114959257e-06,
+      "loss": 0.1608,
+      "step": 18660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.668479949841263e-06,
+      "loss": 0.1811,
+      "step": 18661
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.667468860878864e-06,
+      "loss": 0.1677,
+      "step": 18662
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.666457848086496e-06,
+      "loss": 0.1505,
+      "step": 18663
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.665446911478598e-06,
+      "loss": 0.18,
+      "step": 18664
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.66443605106962e-06,
+      "loss": 0.1716,
+      "step": 18665
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663425266873989e-06,
+      "loss": 0.2065,
+      "step": 18666
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.662414558906151e-06,
+      "loss": 0.1684,
+      "step": 18667
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.661403927180538e-06,
+      "loss": 0.1897,
+      "step": 18668
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6603933717115925e-06,
+      "loss": 0.1676,
+      "step": 18669
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659382892513744e-06,
+      "loss": 0.2044,
+      "step": 18670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6583724896014245e-06,
+      "loss": 0.1743,
+      "step": 18671
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.657362162989074e-06,
+      "loss": 0.1953,
+      "step": 18672
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.656351912691116e-06,
+      "loss": 0.1871,
+      "step": 18673
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655341738721989e-06,
+      "loss": 0.1818,
+      "step": 18674
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.654331641096118e-06,
+      "loss": 0.1846,
+      "step": 18675
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653321619827927e-06,
+      "loss": 0.177,
+      "step": 18676
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.652311674931852e-06,
+      "loss": 0.1935,
+      "step": 18677
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651301806422314e-06,
+      "loss": 0.1969,
+      "step": 18678
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6502920143137395e-06,
+      "loss": 0.1897,
+      "step": 18679
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.649282298620548e-06,
+      "loss": 0.1827,
+      "step": 18680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6482726593571684e-06,
+      "loss": 0.1745,
+      "step": 18681
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6472630965380204e-06,
+      "loss": 0.184,
+      "step": 18682
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64625361017752e-06,
+      "loss": 0.1666,
+      "step": 18683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6452442002900955e-06,
+      "loss": 0.1632,
+      "step": 18684
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64423486689016e-06,
+      "loss": 0.1683,
+      "step": 18685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643225609992128e-06,
+      "loss": 0.1711,
+      "step": 18686
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642216429610423e-06,
+      "loss": 0.1909,
+      "step": 18687
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641207325759454e-06,
+      "loss": 0.1691,
+      "step": 18688
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.640198298453638e-06,
+      "loss": 0.1725,
+      "step": 18689
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.63918934770739e-06,
+      "loss": 0.1911,
+      "step": 18690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6381804735351195e-06,
+      "loss": 0.187,
+      "step": 18691
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.637171675951237e-06,
+      "loss": 0.196,
+      "step": 18692
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636162954970148e-06,
+      "loss": 0.1644,
+      "step": 18693
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.635154310606269e-06,
+      "loss": 0.1871,
+      "step": 18694
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.634145742874001e-06,
+      "loss": 0.1914,
+      "step": 18695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.633137251787756e-06,
+      "loss": 0.1811,
+      "step": 18696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632128837361938e-06,
+      "loss": 0.1863,
+      "step": 18697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.631120499610946e-06,
+      "loss": 0.1778,
+      "step": 18698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.63011223854919e-06,
+      "loss": 0.1905,
+      "step": 18699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629104054191066e-06,
+      "loss": 0.1818,
+      "step": 18700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6280959465509815e-06,
+      "loss": 0.1867,
+      "step": 18701
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.627087915643332e-06,
+      "loss": 0.1716,
+      "step": 18702
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626079961482515e-06,
+      "loss": 0.1933,
+      "step": 18703
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.625072084082936e-06,
+      "loss": 0.1869,
+      "step": 18704
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.62406428345898e-06,
+      "loss": 0.1916,
+      "step": 18705
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623056559625051e-06,
+      "loss": 0.2029,
+      "step": 18706
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.622048912595537e-06,
+      "loss": 0.1812,
+      "step": 18707
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621041342384839e-06,
+      "loss": 0.1885,
+      "step": 18708
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620033849007344e-06,
+      "loss": 0.1898,
+      "step": 18709
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6190264324774415e-06,
+      "loss": 0.1632,
+      "step": 18710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618019092809527e-06,
+      "loss": 0.1883,
+      "step": 18711
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6170118300179824e-06,
+      "loss": 0.1927,
+      "step": 18712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.616004644117205e-06,
+      "loss": 0.194,
+      "step": 18713
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614997535121574e-06,
+      "loss": 0.1804,
+      "step": 18714
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613990503045473e-06,
+      "loss": 0.1756,
+      "step": 18715
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612983547903296e-06,
+      "loss": 0.1726,
+      "step": 18716
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.611976669709421e-06,
+      "loss": 0.2037,
+      "step": 18717
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.610969868478228e-06,
+      "loss": 0.1877,
+      "step": 18718
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609963144224101e-06,
+      "loss": 0.171,
+      "step": 18719
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608956496961416e-06,
+      "loss": 0.1898,
+      "step": 18720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607949926704559e-06,
+      "loss": 0.1873,
+      "step": 18721
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6069434334678995e-06,
+      "loss": 0.183,
+      "step": 18722
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6059370172658245e-06,
+      "loss": 0.223,
+      "step": 18723
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.604930678112703e-06,
+      "loss": 0.1729,
+      "step": 18724
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.603924416022907e-06,
+      "loss": 0.1797,
+      "step": 18725
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602918231010818e-06,
+      "loss": 0.1817,
+      "step": 18726
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.601912123090799e-06,
+      "loss": 0.1801,
+      "step": 18727
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.600906092277232e-06,
+      "loss": 0.1904,
+      "step": 18728
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.59990013858448e-06,
+      "loss": 0.1931,
+      "step": 18729
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598894262026914e-06,
+      "loss": 0.1675,
+      "step": 18730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597888462618903e-06,
+      "loss": 0.183,
+      "step": 18731
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.596882740374806e-06,
+      "loss": 0.1958,
+      "step": 18732
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.595877095309002e-06,
+      "loss": 0.1854,
+      "step": 18733
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594871527435842e-06,
+      "loss": 0.1664,
+      "step": 18734
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.593866036769702e-06,
+      "loss": 0.1805,
+      "step": 18735
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5928606233249385e-06,
+      "loss": 0.1869,
+      "step": 18736
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.59185528711591e-06,
+      "loss": 0.1861,
+      "step": 18737
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590850028156983e-06,
+      "loss": 0.1867,
+      "step": 18738
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.58984484646251e-06,
+      "loss": 0.173,
+      "step": 18739
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5888397420468566e-06,
+      "loss": 0.1909,
+      "step": 18740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.587834714924376e-06,
+      "loss": 0.17,
+      "step": 18741
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.586829765109421e-06,
+      "loss": 0.1801,
+      "step": 18742
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5858248926163514e-06,
+      "loss": 0.1881,
+      "step": 18743
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.584820097459519e-06,
+      "loss": 0.1708,
+      "step": 18744
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.583815379653276e-06,
+      "loss": 0.1879,
+      "step": 18745
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5828107392119705e-06,
+      "loss": 0.1979,
+      "step": 18746
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.581806176149959e-06,
+      "loss": 0.1704,
+      "step": 18747
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5808016904815885e-06,
+      "loss": 0.1857,
+      "step": 18748
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5797972822212025e-06,
+      "loss": 0.1952,
+      "step": 18749
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.578792951383155e-06,
+      "loss": 0.1816,
+      "step": 18750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577788697981786e-06,
+      "loss": 0.1743,
+      "step": 18751
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5767845220314464e-06,
+      "loss": 0.1699,
+      "step": 18752
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.575780423546476e-06,
+      "loss": 0.1754,
+      "step": 18753
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5747764025412155e-06,
+      "loss": 0.1921,
+      "step": 18754
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5737724590300115e-06,
+      "loss": 0.1795,
+      "step": 18755
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.572768593027201e-06,
+      "loss": 0.1807,
+      "step": 18756
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5717648045471255e-06,
+      "loss": 0.1937,
+      "step": 18757
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.570761093604121e-06,
+      "loss": 0.186,
+      "step": 18758
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5697574602125215e-06,
+      "loss": 0.1843,
+      "step": 18759
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.56875390438667e-06,
+      "loss": 0.1897,
+      "step": 18760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.567750426140893e-06,
+      "loss": 0.1848,
+      "step": 18761
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.566747025489534e-06,
+      "loss": 0.1877,
+      "step": 18762
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5657437024469205e-06,
+      "loss": 0.1792,
+      "step": 18763
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.564740457027381e-06,
+      "loss": 0.2011,
+      "step": 18764
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.563737289245252e-06,
+      "loss": 0.1903,
+      "step": 18765
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5627341991148556e-06,
+      "loss": 0.1882,
+      "step": 18766
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.561731186650529e-06,
+      "loss": 0.179,
+      "step": 18767
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.560728251866594e-06,
+      "loss": 0.1819,
+      "step": 18768
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559725394777373e-06,
+      "loss": 0.1892,
+      "step": 18769
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.558722615397202e-06,
+      "loss": 0.2078,
+      "step": 18770
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.557719913740391e-06,
+      "loss": 0.2163,
+      "step": 18771
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5567172898212716e-06,
+      "loss": 0.193,
+      "step": 18772
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.55571474365416e-06,
+      "loss": 0.1845,
+      "step": 18773
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554712275253384e-06,
+      "loss": 0.1949,
+      "step": 18774
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.553709884633257e-06,
+      "loss": 0.1959,
+      "step": 18775
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552707571808096e-06,
+      "loss": 0.1653,
+      "step": 18776
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551705336792224e-06,
+      "loss": 0.162,
+      "step": 18777
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.550703179599949e-06,
+      "loss": 0.1651,
+      "step": 18778
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.549701100245595e-06,
+      "loss": 0.1918,
+      "step": 18779
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5486990987434705e-06,
+      "loss": 0.1747,
+      "step": 18780
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547697175107886e-06,
+      "loss": 0.1885,
+      "step": 18781
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.546695329353159e-06,
+      "loss": 0.1894,
+      "step": 18782
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545693561493596e-06,
+      "loss": 0.1812,
+      "step": 18783
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.544691871543508e-06,
+      "loss": 0.1879,
+      "step": 18784
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.543690259517197e-06,
+      "loss": 0.1932,
+      "step": 18785
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.542688725428981e-06,
+      "loss": 0.1861,
+      "step": 18786
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.541687269293158e-06,
+      "loss": 0.1805,
+      "step": 18787
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.540685891124033e-06,
+      "loss": 0.1669,
+      "step": 18788
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.539684590935915e-06,
+      "loss": 0.1815,
+      "step": 18789
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.538683368743102e-06,
+      "loss": 0.175,
+      "step": 18790
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5376822245598935e-06,
+      "loss": 0.182,
+      "step": 18791
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536681158400598e-06,
+      "loss": 0.185,
+      "step": 18792
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.535680170279504e-06,
+      "loss": 0.1932,
+      "step": 18793
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.534679260210921e-06,
+      "loss": 0.1831,
+      "step": 18794
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53367842820914e-06,
+      "loss": 0.1795,
+      "step": 18795
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.532677674288457e-06,
+      "loss": 0.1913,
+      "step": 18796
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531676998463168e-06,
+      "loss": 0.1833,
+      "step": 18797
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530676400747562e-06,
+      "loss": 0.1664,
+      "step": 18798
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.52967588115594e-06,
+      "loss": 0.1882,
+      "step": 18799
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.528675439702584e-06,
+      "loss": 0.1812,
+      "step": 18800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.527675076401793e-06,
+      "loss": 0.1828,
+      "step": 18801
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.526674791267853e-06,
+      "loss": 0.1799,
+      "step": 18802
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.525674584315049e-06,
+      "loss": 0.2046,
+      "step": 18803
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524674455557672e-06,
+      "loss": 0.2012,
+      "step": 18804
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.523674405010005e-06,
+      "loss": 0.196,
+      "step": 18805
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.522674432686337e-06,
+      "loss": 0.1711,
+      "step": 18806
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.521674538600949e-06,
+      "loss": 0.1701,
+      "step": 18807
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520674722768119e-06,
+      "loss": 0.1618,
+      "step": 18808
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5196749852021405e-06,
+      "loss": 0.1989,
+      "step": 18809
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518675325917279e-06,
+      "loss": 0.1907,
+      "step": 18810
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.517675744927824e-06,
+      "loss": 0.1997,
+      "step": 18811
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.516676242248047e-06,
+      "loss": 0.1877,
+      "step": 18812
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.515676817892231e-06,
+      "loss": 0.1692,
+      "step": 18813
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.51467747187465e-06,
+      "loss": 0.191,
+      "step": 18814
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5136782042095735e-06,
+      "loss": 0.1615,
+      "step": 18815
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.512679014911283e-06,
+      "loss": 0.1845,
+      "step": 18816
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.511679903994044e-06,
+      "loss": 0.185,
+      "step": 18817
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.510680871472134e-06,
+      "loss": 0.1777,
+      "step": 18818
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.50968191735982e-06,
+      "loss": 0.1986,
+      "step": 18819
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.508683041671369e-06,
+      "loss": 0.2012,
+      "step": 18820
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507684244421054e-06,
+      "loss": 0.2089,
+      "step": 18821
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.506685525623139e-06,
+      "loss": 0.1852,
+      "step": 18822
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5056868852918914e-06,
+      "loss": 0.1897,
+      "step": 18823
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504688323441568e-06,
+      "loss": 0.1885,
+      "step": 18824
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.503689840086445e-06,
+      "loss": 0.1723,
+      "step": 18825
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.502691435240777e-06,
+      "loss": 0.1815,
+      "step": 18826
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501693108918822e-06,
+      "loss": 0.1914,
+      "step": 18827
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.50069486113485e-06,
+      "loss": 0.1918,
+      "step": 18828
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4996966919031136e-06,
+      "loss": 0.1736,
+      "step": 18829
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498698601237869e-06,
+      "loss": 0.1897,
+      "step": 18830
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.497700589153379e-06,
+      "loss": 0.17,
+      "step": 18831
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496702655663891e-06,
+      "loss": 0.1807,
+      "step": 18832
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.49570480078367e-06,
+      "loss": 0.1684,
+      "step": 18833
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.494707024526963e-06,
+      "loss": 0.1887,
+      "step": 18834
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.49370932690802e-06,
+      "loss": 0.1838,
+      "step": 18835
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.492711707941102e-06,
+      "loss": 0.19,
+      "step": 18836
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.491714167640447e-06,
+      "loss": 0.1782,
+      "step": 18837
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4907167060203105e-06,
+      "loss": 0.1704,
+      "step": 18838
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.489719323094936e-06,
+      "loss": 0.1781,
+      "step": 18839
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.488722018878576e-06,
+      "loss": 0.191,
+      "step": 18840
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4877247933854736e-06,
+      "loss": 0.1964,
+      "step": 18841
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48672764662987e-06,
+      "loss": 0.1868,
+      "step": 18842
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.485730578626014e-06,
+      "loss": 0.1767,
+      "step": 18843
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.484733589388142e-06,
+      "loss": 0.1827,
+      "step": 18844
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4837366789305015e-06,
+      "loss": 0.1713,
+      "step": 18845
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4827398472673275e-06,
+      "loss": 0.2092,
+      "step": 18846
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481743094412858e-06,
+      "loss": 0.212,
+      "step": 18847
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.480746420381336e-06,
+      "loss": 0.1853,
+      "step": 18848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4797498251869955e-06,
+      "loss": 0.1859,
+      "step": 18849
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478753308844071e-06,
+      "loss": 0.173,
+      "step": 18850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.477756871366792e-06,
+      "loss": 0.1811,
+      "step": 18851
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4767605127694015e-06,
+      "loss": 0.1888,
+      "step": 18852
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475764233066127e-06,
+      "loss": 0.1818,
+      "step": 18853
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.474768032271194e-06,
+      "loss": 0.188,
+      "step": 18854
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.473771910398842e-06,
+      "loss": 0.1888,
+      "step": 18855
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.47277586746329e-06,
+      "loss": 0.1986,
+      "step": 18856
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.471779903478776e-06,
+      "loss": 0.1828,
+      "step": 18857
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4707840184595195e-06,
+      "loss": 0.196,
+      "step": 18858
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469788212419744e-06,
+      "loss": 0.1893,
+      "step": 18859
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46879248537368e-06,
+      "loss": 0.1959,
+      "step": 18860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.467796837335547e-06,
+      "loss": 0.1869,
+      "step": 18861
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466801268319567e-06,
+      "loss": 0.1723,
+      "step": 18862
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465805778339961e-06,
+      "loss": 0.1861,
+      "step": 18863
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.464810367410946e-06,
+      "loss": 0.2093,
+      "step": 18864
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4638150355467445e-06,
+      "loss": 0.1746,
+      "step": 18865
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46281978276157e-06,
+      "loss": 0.1774,
+      "step": 18866
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461824609069644e-06,
+      "loss": 0.18,
+      "step": 18867
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.460829514485179e-06,
+      "loss": 0.2105,
+      "step": 18868
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4598344990223845e-06,
+      "loss": 0.1828,
+      "step": 18869
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.458839562695481e-06,
+      "loss": 0.1657,
+      "step": 18870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457844705518671e-06,
+      "loss": 0.181,
+      "step": 18871
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.456849927506177e-06,
+      "loss": 0.1902,
+      "step": 18872
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455855228672201e-06,
+      "loss": 0.1986,
+      "step": 18873
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.454860609030949e-06,
+      "loss": 0.2,
+      "step": 18874
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.453866068596638e-06,
+      "loss": 0.1856,
+      "step": 18875
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452871607383461e-06,
+      "loss": 0.1924,
+      "step": 18876
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4518772254056335e-06,
+      "loss": 0.1708,
+      "step": 18877
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.450882922677351e-06,
+      "loss": 0.189,
+      "step": 18878
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.449888699212823e-06,
+      "loss": 0.1684,
+      "step": 18879
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.448894555026249e-06,
+      "loss": 0.1776,
+      "step": 18880
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.447900490131826e-06,
+      "loss": 0.1829,
+      "step": 18881
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44690650454376e-06,
+      "loss": 0.1831,
+      "step": 18882
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4459125982762406e-06,
+      "loss": 0.1852,
+      "step": 18883
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4449187713434736e-06,
+      "loss": 0.1981,
+      "step": 18884
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.443925023759651e-06,
+      "loss": 0.1861,
+      "step": 18885
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.442931355538962e-06,
+      "loss": 0.2274,
+      "step": 18886
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4419377666956095e-06,
+      "loss": 0.1734,
+      "step": 18887
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440944257243782e-06,
+      "loss": 0.1936,
+      "step": 18888
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.439950827197671e-06,
+      "loss": 0.1829,
+      "step": 18889
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.438957476571462e-06,
+      "loss": 0.1833,
+      "step": 18890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.437964205379352e-06,
+      "loss": 0.173,
+      "step": 18891
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436971013635526e-06,
+      "loss": 0.1846,
+      "step": 18892
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.435977901354166e-06,
+      "loss": 0.2015,
+      "step": 18893
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434984868549466e-06,
+      "loss": 0.176,
+      "step": 18894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4339919152356046e-06,
+      "loss": 0.2067,
+      "step": 18895
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.432999041426764e-06,
+      "loss": 0.177,
+      "step": 18896
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4320062471371325e-06,
+      "loss": 0.1773,
+      "step": 18897
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4310135323808844e-06,
+      "loss": 0.1799,
+      "step": 18898
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430020897172207e-06,
+      "loss": 0.2142,
+      "step": 18899
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429028341525275e-06,
+      "loss": 0.1653,
+      "step": 18900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4280358654542635e-06,
+      "loss": 0.1785,
+      "step": 18901
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4270434689733584e-06,
+      "loss": 0.1927,
+      "step": 18902
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426051152096722e-06,
+      "loss": 0.2026,
+      "step": 18903
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.425058914838538e-06,
+      "loss": 0.1694,
+      "step": 18904
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424066757212974e-06,
+      "loss": 0.1804,
+      "step": 18905
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423074679234208e-06,
+      "loss": 0.1854,
+      "step": 18906
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.422082680916408e-06,
+      "loss": 0.1659,
+      "step": 18907
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.42109076227374e-06,
+      "loss": 0.1733,
+      "step": 18908
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420098923320378e-06,
+      "loss": 0.1844,
+      "step": 18909
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419107164070485e-06,
+      "loss": 0.1747,
+      "step": 18910
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.418115484538234e-06,
+      "loss": 0.1697,
+      "step": 18911
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4171238847377854e-06,
+      "loss": 0.1889,
+      "step": 18912
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4161323646833e-06,
+      "loss": 0.2024,
+      "step": 18913
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.415140924388948e-06,
+      "loss": 0.1776,
+      "step": 18914
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414149563868888e-06,
+      "loss": 0.1832,
+      "step": 18915
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.413158283137281e-06,
+      "loss": 0.185,
+      "step": 18916
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.412167082208281e-06,
+      "loss": 0.1953,
+      "step": 18917
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.411175961096055e-06,
+      "loss": 0.2093,
+      "step": 18918
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4101849198147565e-06,
+      "loss": 0.1828,
+      "step": 18919
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4091939583785386e-06,
+      "loss": 0.2003,
+      "step": 18920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.40820307680156e-06,
+      "loss": 0.1899,
+      "step": 18921
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.407212275097972e-06,
+      "loss": 0.1891,
+      "step": 18922
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4062215532819315e-06,
+      "loss": 0.2049,
+      "step": 18923
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.405230911367588e-06,
+      "loss": 0.1741,
+      "step": 18924
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.404240349369088e-06,
+      "loss": 0.1804,
+      "step": 18925
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.403249867300585e-06,
+      "loss": 0.1676,
+      "step": 18926
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.402259465176224e-06,
+      "loss": 0.1869,
+      "step": 18927
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.401269143010162e-06,
+      "loss": 0.1723,
+      "step": 18928
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400278900816527e-06,
+      "loss": 0.1774,
+      "step": 18929
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.399288738609478e-06,
+      "loss": 0.1869,
+      "step": 18930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.398298656403154e-06,
+      "loss": 0.1888,
+      "step": 18931
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3973086542116935e-06,
+      "loss": 0.1846,
+      "step": 18932
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.396318732049245e-06,
+      "loss": 0.1708,
+      "step": 18933
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.395328889929944e-06,
+      "loss": 0.1871,
+      "step": 18934
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394339127867927e-06,
+      "loss": 0.1868,
+      "step": 18935
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.393349445877339e-06,
+      "loss": 0.1837,
+      "step": 18936
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.392359843972311e-06,
+      "loss": 0.1885,
+      "step": 18937
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391370322166981e-06,
+      "loss": 0.1876,
+      "step": 18938
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.390380880475483e-06,
+      "loss": 0.2112,
+      "step": 18939
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.389391518911948e-06,
+      "loss": 0.1809,
+      "step": 18940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388402237490516e-06,
+      "loss": 0.1957,
+      "step": 18941
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.387413036225306e-06,
+      "loss": 0.1903,
+      "step": 18942
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.386423915130455e-06,
+      "loss": 0.202,
+      "step": 18943
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3854348742200884e-06,
+      "loss": 0.1897,
+      "step": 18944
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.384445913508337e-06,
+      "loss": 0.1891,
+      "step": 18945
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.383457033009328e-06,
+      "loss": 0.1659,
+      "step": 18946
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3824682327371795e-06,
+      "loss": 0.1739,
+      "step": 18947
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.381479512706025e-06,
+      "loss": 0.1784,
+      "step": 18948
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3804908729299775e-06,
+      "loss": 0.1704,
+      "step": 18949
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37950231342317e-06,
+      "loss": 0.1766,
+      "step": 18950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378513834199715e-06,
+      "loss": 0.1909,
+      "step": 18951
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.377525435273731e-06,
+      "loss": 0.1877,
+      "step": 18952
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.376537116659344e-06,
+      "loss": 0.1908,
+      "step": 18953
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375548878370665e-06,
+      "loss": 0.1733,
+      "step": 18954
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374560720421813e-06,
+      "loss": 0.1947,
+      "step": 18955
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373572642826898e-06,
+      "loss": 0.1955,
+      "step": 18956
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37258464560004e-06,
+      "loss": 0.1522,
+      "step": 18957
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.371596728755348e-06,
+      "loss": 0.1716,
+      "step": 18958
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.370608892306932e-06,
+      "loss": 0.166,
+      "step": 18959
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369621136268908e-06,
+      "loss": 0.1753,
+      "step": 18960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.368633460655377e-06,
+      "loss": 0.1903,
+      "step": 18961
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.367645865480457e-06,
+      "loss": 0.183,
+      "step": 18962
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.366658350758249e-06,
+      "loss": 0.1811,
+      "step": 18963
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365670916502854e-06,
+      "loss": 0.173,
+      "step": 18964
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364683562728387e-06,
+      "loss": 0.1896,
+      "step": 18965
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363696289448945e-06,
+      "loss": 0.175,
+      "step": 18966
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.362709096678632e-06,
+      "loss": 0.1802,
+      "step": 18967
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.361721984431548e-06,
+      "loss": 0.1787,
+      "step": 18968
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.36073495272179e-06,
+      "loss": 0.189,
+      "step": 18969
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.359748001563464e-06,
+      "loss": 0.1727,
+      "step": 18970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35876113097066e-06,
+      "loss": 0.1897,
+      "step": 18971
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3577743409574815e-06,
+      "loss": 0.1873,
+      "step": 18972
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356787631538021e-06,
+      "loss": 0.2095,
+      "step": 18973
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.355801002726369e-06,
+      "loss": 0.1765,
+      "step": 18974
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354814454536625e-06,
+      "loss": 0.1896,
+      "step": 18975
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353827986982874e-06,
+      "loss": 0.1721,
+      "step": 18976
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.352841600079215e-06,
+      "loss": 0.1747,
+      "step": 18977
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3518552938397325e-06,
+      "loss": 0.1866,
+      "step": 18978
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350869068278511e-06,
+      "loss": 0.1832,
+      "step": 18979
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.349882923409648e-06,
+      "loss": 0.1886,
+      "step": 18980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348896859247222e-06,
+      "loss": 0.1847,
+      "step": 18981
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.34791087580532e-06,
+      "loss": 0.1877,
+      "step": 18982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.346924973098023e-06,
+      "loss": 0.1681,
+      "step": 18983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345939151139419e-06,
+      "loss": 0.2105,
+      "step": 18984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.344953409943588e-06,
+      "loss": 0.1761,
+      "step": 18985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343967749524605e-06,
+      "loss": 0.1864,
+      "step": 18986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342982169896555e-06,
+      "loss": 0.1765,
+      "step": 18987
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3419966710735136e-06,
+      "loss": 0.1685,
+      "step": 18988
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.341011253069561e-06,
+      "loss": 0.1959,
+      "step": 18989
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3400259158987715e-06,
+      "loss": 0.1865,
+      "step": 18990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.339040659575213e-06,
+      "loss": 0.1913,
+      "step": 18991
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3380554841129695e-06,
+      "loss": 0.1725,
+      "step": 18992
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3370703895261044e-06,
+      "loss": 0.1707,
+      "step": 18993
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3360853758287005e-06,
+      "loss": 0.1737,
+      "step": 18994
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.335100443034813e-06,
+      "loss": 0.1881,
+      "step": 18995
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.334115591158521e-06,
+      "loss": 0.1742,
+      "step": 18996
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.333130820213889e-06,
+      "loss": 0.1789,
+      "step": 18997
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.332146130214979e-06,
+      "loss": 0.1742,
+      "step": 18998
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331161521175865e-06,
+      "loss": 0.173,
+      "step": 18999
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330176993110608e-06,
+      "loss": 0.1874,
+      "step": 19000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.329192546033264e-06,
+      "loss": 0.1838,
+      "step": 19001
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.328208179957906e-06,
+      "loss": 0.1843,
+      "step": 19002
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.327223894898585e-06,
+      "loss": 0.1809,
+      "step": 19003
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326239690869369e-06,
+      "loss": 0.1761,
+      "step": 19004
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.325255567884314e-06,
+      "loss": 0.1819,
+      "step": 19005
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.324271525957471e-06,
+      "loss": 0.1747,
+      "step": 19006
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.323287565102908e-06,
+      "loss": 0.1972,
+      "step": 19007
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322303685334666e-06,
+      "loss": 0.1703,
+      "step": 19008
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.321319886666808e-06,
+      "loss": 0.1971,
+      "step": 19009
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3203361691133815e-06,
+      "loss": 0.19,
+      "step": 19010
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.2064,
+      "step": 19011
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.318368977406042e-06,
+      "loss": 0.1742,
+      "step": 19012
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.317385503280221e-06,
+      "loss": 0.1784,
+      "step": 19013
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.316402110325037e-06,
+      "loss": 0.1714,
+      "step": 19014
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3154187985545305e-06,
+      "loss": 0.1852,
+      "step": 19015
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.314435567982752e-06,
+      "loss": 0.2018,
+      "step": 19016
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3134524186237435e-06,
+      "loss": 0.1834,
+      "step": 19017
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.312469350491544e-06,
+      "loss": 0.1898,
+      "step": 19018
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.311486363600205e-06,
+      "loss": 0.169,
+      "step": 19019
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310503457963761e-06,
+      "loss": 0.1732,
+      "step": 19020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.309520633596253e-06,
+      "loss": 0.1918,
+      "step": 19021
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308537890511718e-06,
+      "loss": 0.1854,
+      "step": 19022
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.307555228724199e-06,
+      "loss": 0.1729,
+      "step": 19023
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.306572648247728e-06,
+      "loss": 0.1882,
+      "step": 19024
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.305590149096338e-06,
+      "loss": 0.187,
+      "step": 19025
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.30460773128407e-06,
+      "loss": 0.192,
+      "step": 19026
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3036253948249484e-06,
+      "loss": 0.1772,
+      "step": 19027
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.302643139733014e-06,
+      "loss": 0.1944,
+      "step": 19028
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.301660966022293e-06,
+      "loss": 0.2027,
+      "step": 19029
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300678873706811e-06,
+      "loss": 0.1786,
+      "step": 19030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299696862800603e-06,
+      "loss": 0.1833,
+      "step": 19031
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298714933317691e-06,
+      "loss": 0.176,
+      "step": 19032
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2977330852721085e-06,
+      "loss": 0.1846,
+      "step": 19033
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2967513186778685e-06,
+      "loss": 0.1868,
+      "step": 19034
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295769633549004e-06,
+      "loss": 0.1853,
+      "step": 19035
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.294788029899533e-06,
+      "loss": 0.1767,
+      "step": 19036
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293806507743477e-06,
+      "loss": 0.1914,
+      "step": 19037
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292825067094859e-06,
+      "loss": 0.1639,
+      "step": 19038
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.291843707967696e-06,
+      "loss": 0.1882,
+      "step": 19039
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2908624303760015e-06,
+      "loss": 0.1788,
+      "step": 19040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.289881234333802e-06,
+      "loss": 0.1631,
+      "step": 19041
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.288900119855102e-06,
+      "loss": 0.1782,
+      "step": 19042
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.287919086953924e-06,
+      "loss": 0.1658,
+      "step": 19043
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.28693813564428e-06,
+      "loss": 0.1886,
+      "step": 19044
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285957265940175e-06,
+      "loss": 0.191,
+      "step": 19045
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2849764778556335e-06,
+      "loss": 0.1747,
+      "step": 19046
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.283995771404648e-06,
+      "loss": 0.1798,
+      "step": 19047
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.283015146601239e-06,
+      "loss": 0.1934,
+      "step": 19048
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282034603459407e-06,
+      "loss": 0.1867,
+      "step": 19049
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281054141993165e-06,
+      "loss": 0.1762,
+      "step": 19050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2800737622165146e-06,
+      "loss": 0.1998,
+      "step": 19051
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.279093464143455e-06,
+      "loss": 0.1876,
+      "step": 19052
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2781132477879965e-06,
+      "loss": 0.1749,
+      "step": 19053
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.277133113164133e-06,
+      "loss": 0.1951,
+      "step": 19054
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.276153060285874e-06,
+      "loss": 0.1891,
+      "step": 19055
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.275173089167211e-06,
+      "loss": 0.19,
+      "step": 19056
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.274193199822143e-06,
+      "loss": 0.1884,
+      "step": 19057
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273213392264671e-06,
+      "loss": 0.176,
+      "step": 19058
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.272233666508784e-06,
+      "loss": 0.1615,
+      "step": 19059
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271254022568487e-06,
+      "loss": 0.1522,
+      "step": 19060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.270274460457759e-06,
+      "loss": 0.1702,
+      "step": 19061
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.269294980190604e-06,
+      "loss": 0.1707,
+      "step": 19062
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.268315581781008e-06,
+      "loss": 0.1798,
+      "step": 19063
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267336265242958e-06,
+      "loss": 0.1969,
+      "step": 19064
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266357030590449e-06,
+      "loss": 0.1777,
+      "step": 19065
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.265377877837462e-06,
+      "loss": 0.1906,
+      "step": 19066
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.264398806997989e-06,
+      "loss": 0.1776,
+      "step": 19067
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263419818086013e-06,
+      "loss": 0.1921,
+      "step": 19068
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.262440911115514e-06,
+      "loss": 0.1875,
+      "step": 19069
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2614620861004815e-06,
+      "loss": 0.1828,
+      "step": 19070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260483343054894e-06,
+      "loss": 0.1816,
+      "step": 19071
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2595046819927285e-06,
+      "loss": 0.18,
+      "step": 19072
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2585261029279746e-06,
+      "loss": 0.1832,
+      "step": 19073
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.257547605874595e-06,
+      "loss": 0.1877,
+      "step": 19074
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2565691908465775e-06,
+      "loss": 0.183,
+      "step": 19075
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255590857857893e-06,
+      "loss": 0.2073,
+      "step": 19076
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2546126069225215e-06,
+      "loss": 0.1891,
+      "step": 19077
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.253634438054433e-06,
+      "loss": 0.1918,
+      "step": 19078
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.252656351267594e-06,
+      "loss": 0.1737,
+      "step": 19079
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.251678346575987e-06,
+      "loss": 0.1589,
+      "step": 19080
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2507004239935715e-06,
+      "loss": 0.1932,
+      "step": 19081
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.249722583534324e-06,
+      "loss": 0.1777,
+      "step": 19082
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.248744825212209e-06,
+      "loss": 0.192,
+      "step": 19083
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.247767149041187e-06,
+      "loss": 0.1659,
+      "step": 19084
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246789555035234e-06,
+      "loss": 0.1726,
+      "step": 19085
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24581204320831e-06,
+      "loss": 0.1963,
+      "step": 19086
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2448346135743736e-06,
+      "loss": 0.1853,
+      "step": 19087
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243857266147388e-06,
+      "loss": 0.1992,
+      "step": 19088
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.242880000941316e-06,
+      "loss": 0.1855,
+      "step": 19089
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.241902817970117e-06,
+      "loss": 0.183,
+      "step": 19090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.240925717247745e-06,
+      "loss": 0.1949,
+      "step": 19091
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2399486987881626e-06,
+      "loss": 0.1839,
+      "step": 19092
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23897176260532e-06,
+      "loss": 0.1817,
+      "step": 19093
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.237994908713177e-06,
+      "loss": 0.187,
+      "step": 19094
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.237018137125685e-06,
+      "loss": 0.163,
+      "step": 19095
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.236041447856792e-06,
+      "loss": 0.1826,
+      "step": 19096
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.235064840920456e-06,
+      "loss": 0.1969,
+      "step": 19097
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234088316330621e-06,
+      "loss": 0.1821,
+      "step": 19098
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233111874101246e-06,
+      "loss": 0.1621,
+      "step": 19099
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2321355142462625e-06,
+      "loss": 0.2099,
+      "step": 19100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2311592367796305e-06,
+      "loss": 0.2136,
+      "step": 19101
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230183041715289e-06,
+      "loss": 0.1803,
+      "step": 19102
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.22920692906718e-06,
+      "loss": 0.1887,
+      "step": 19103
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.228230898849253e-06,
+      "loss": 0.1869,
+      "step": 19104
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.227254951075444e-06,
+      "loss": 0.1893,
+      "step": 19105
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2262790857596945e-06,
+      "loss": 0.2011,
+      "step": 19106
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.225303302915947e-06,
+      "loss": 0.1814,
+      "step": 19107
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.224327602558134e-06,
+      "loss": 0.1865,
+      "step": 19108
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.223351984700199e-06,
+      "loss": 0.1804,
+      "step": 19109
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.222376449356075e-06,
+      "loss": 0.1944,
+      "step": 19110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221400996539693e-06,
+      "loss": 0.1914,
+      "step": 19111
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.220425626264996e-06,
+      "loss": 0.171,
+      "step": 19112
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219450338545903e-06,
+      "loss": 0.1802,
+      "step": 19113
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2184751333963545e-06,
+      "loss": 0.1877,
+      "step": 19114
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.217500010830274e-06,
+      "loss": 0.1792,
+      "step": 19115
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2165249708615985e-06,
+      "loss": 0.1854,
+      "step": 19116
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2155500135042495e-06,
+      "loss": 0.172,
+      "step": 19117
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.214575138772152e-06,
+      "loss": 0.1788,
+      "step": 19118
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.213600346679237e-06,
+      "loss": 0.1991,
+      "step": 19119
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.212625637239421e-06,
+      "loss": 0.1703,
+      "step": 19120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.211651010466636e-06,
+      "loss": 0.1915,
+      "step": 19121
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210676466374798e-06,
+      "loss": 0.1764,
+      "step": 19122
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209702004977824e-06,
+      "loss": 0.1705,
+      "step": 19123
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.208727626289641e-06,
+      "loss": 0.1715,
+      "step": 19124
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.207753330324163e-06,
+      "loss": 0.2067,
+      "step": 19125
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.206779117095308e-06,
+      "loss": 0.1708,
+      "step": 19126
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.205804986616987e-06,
+      "loss": 0.1668,
+      "step": 19127
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204830938903122e-06,
+      "loss": 0.1793,
+      "step": 19128
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.203856973967622e-06,
+      "loss": 0.1736,
+      "step": 19129
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202883091824397e-06,
+      "loss": 0.1787,
+      "step": 19130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.201909292487364e-06,
+      "loss": 0.1889,
+      "step": 19131
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200935575970426e-06,
+      "loss": 0.1903,
+      "step": 19132
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.199961942287499e-06,
+      "loss": 0.1978,
+      "step": 19133
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1989883914524866e-06,
+      "loss": 0.1793,
+      "step": 19134
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.198014923479291e-06,
+      "loss": 0.1755,
+      "step": 19135
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197041538381825e-06,
+      "loss": 0.1676,
+      "step": 19136
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196068236173985e-06,
+      "loss": 0.1823,
+      "step": 19137
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.195095016869681e-06,
+      "loss": 0.1924,
+      "step": 19138
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.194121880482811e-06,
+      "loss": 0.1694,
+      "step": 19139
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193148827027276e-06,
+      "loss": 0.1983,
+      "step": 19140
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.192175856516974e-06,
+      "loss": 0.1893,
+      "step": 19141
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.191202968965798e-06,
+      "loss": 0.1805,
+      "step": 19142
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1902301643876555e-06,
+      "loss": 0.1851,
+      "step": 19143
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189257442796436e-06,
+      "loss": 0.2026,
+      "step": 19144
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.188284804206031e-06,
+      "loss": 0.1816,
+      "step": 19145
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.187312248630341e-06,
+      "loss": 0.1713,
+      "step": 19146
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.18633977608325e-06,
+      "loss": 0.2049,
+      "step": 19147
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185367386578657e-06,
+      "loss": 0.1782,
+      "step": 19148
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1843950801304465e-06,
+      "loss": 0.1677,
+      "step": 19149
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183422856752506e-06,
+      "loss": 0.1763,
+      "step": 19150
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.182450716458728e-06,
+      "loss": 0.1983,
+      "step": 19151
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1814786592629956e-06,
+      "loss": 0.1598,
+      "step": 19152
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.180506685179193e-06,
+      "loss": 0.1868,
+      "step": 19153
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.179534794221202e-06,
+      "loss": 0.1793,
+      "step": 19154
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17856298640291e-06,
+      "loss": 0.1937,
+      "step": 19155
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177591261738196e-06,
+      "loss": 0.1774,
+      "step": 19156
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.176619620240937e-06,
+      "loss": 0.1736,
+      "step": 19157
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1756480619250185e-06,
+      "loss": 0.1726,
+      "step": 19158
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.174676586804311e-06,
+      "loss": 0.1915,
+      "step": 19159
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173705194892698e-06,
+      "loss": 0.1988,
+      "step": 19160
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172733886204052e-06,
+      "loss": 0.1753,
+      "step": 19161
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.171762660752243e-06,
+      "loss": 0.1861,
+      "step": 19162
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170791518551152e-06,
+      "loss": 0.1974,
+      "step": 19163
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1698204596146445e-06,
+      "loss": 0.1898,
+      "step": 19164
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168849483956598e-06,
+      "loss": 0.1881,
+      "step": 19165
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.167878591590873e-06,
+      "loss": 0.1737,
+      "step": 19166
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166907782531343e-06,
+      "loss": 0.176,
+      "step": 19167
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165937056791876e-06,
+      "loss": 0.1898,
+      "step": 19168
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.164966414386332e-06,
+      "loss": 0.1647,
+      "step": 19169
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163995855328583e-06,
+      "loss": 0.1845,
+      "step": 19170
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163025379632486e-06,
+      "loss": 0.2039,
+      "step": 19171
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1620549873119095e-06,
+      "loss": 0.1888,
+      "step": 19172
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161084678380712e-06,
+      "loss": 0.1767,
+      "step": 19173
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.160114452852749e-06,
+      "loss": 0.1594,
+      "step": 19174
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.159144310741888e-06,
+      "loss": 0.191,
+      "step": 19175
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158174252061981e-06,
+      "loss": 0.1707,
+      "step": 19176
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1572042768268825e-06,
+      "loss": 0.1631,
+      "step": 19177
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1562343850504576e-06,
+      "loss": 0.1785,
+      "step": 19178
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.155264576746544e-06,
+      "loss": 0.1899,
+      "step": 19179
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154294851929008e-06,
+      "loss": 0.1981,
+      "step": 19180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153325210611694e-06,
+      "loss": 0.1693,
+      "step": 19181
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.152355652808457e-06,
+      "loss": 0.1718,
+      "step": 19182
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.151386178533146e-06,
+      "loss": 0.1879,
+      "step": 19183
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.150416787799603e-06,
+      "loss": 0.174,
+      "step": 19184
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149447480621682e-06,
+      "loss": 0.1723,
+      "step": 19185
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.148478257013222e-06,
+      "loss": 0.1735,
+      "step": 19186
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147509116988075e-06,
+      "loss": 0.1994,
+      "step": 19187
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.14654006056008e-06,
+      "loss": 0.191,
+      "step": 19188
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.145571087743076e-06,
+      "loss": 0.1843,
+      "step": 19189
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144602198550911e-06,
+      "loss": 0.1805,
+      "step": 19190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1436333929974216e-06,
+      "loss": 0.1823,
+      "step": 19191
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.142664671096444e-06,
+      "loss": 0.1777,
+      "step": 19192
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.141696032861814e-06,
+      "loss": 0.1775,
+      "step": 19193
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.140727478307374e-06,
+      "loss": 0.1852,
+      "step": 19194
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.139759007446955e-06,
+      "loss": 0.1781,
+      "step": 19195
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1387906202943875e-06,
+      "loss": 0.2032,
+      "step": 19196
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.137822316863512e-06,
+      "loss": 0.1855,
+      "step": 19197
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.13685409716815e-06,
+      "loss": 0.1797,
+      "step": 19198
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.135885961222143e-06,
+      "loss": 0.2063,
+      "step": 19199
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.134917909039312e-06,
+      "loss": 0.1986,
+      "step": 19200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.133949940633484e-06,
+      "loss": 0.1803,
+      "step": 19201
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.13298205601849e-06,
+      "loss": 0.1942,
+      "step": 19202
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132014255208151e-06,
+      "loss": 0.186,
+      "step": 19203
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1310465382163e-06,
+      "loss": 0.1898,
+      "step": 19204
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130078905056746e-06,
+      "loss": 0.1951,
+      "step": 19205
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.129111355743321e-06,
+      "loss": 0.1827,
+      "step": 19206
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.128143890289844e-06,
+      "loss": 0.1967,
+      "step": 19207
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127176508710128e-06,
+      "loss": 0.174,
+      "step": 19208
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1262092110179995e-06,
+      "loss": 0.203,
+      "step": 19209
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.125241997227273e-06,
+      "loss": 0.1647,
+      "step": 19210
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124274867351758e-06,
+      "loss": 0.1795,
+      "step": 19211
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.123307821405278e-06,
+      "loss": 0.1763,
+      "step": 19212
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.122340859401639e-06,
+      "loss": 0.1816,
+      "step": 19213
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.121373981354659e-06,
+      "loss": 0.1963,
+      "step": 19214
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.120407187278147e-06,
+      "loss": 0.221,
+      "step": 19215
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1194404771859086e-06,
+      "loss": 0.1834,
+      "step": 19216
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11847385109176e-06,
+      "loss": 0.1718,
+      "step": 19217
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117507309009503e-06,
+      "loss": 0.1598,
+      "step": 19218
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.116540850952946e-06,
+      "loss": 0.1593,
+      "step": 19219
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.115574476935889e-06,
+      "loss": 0.186,
+      "step": 19220
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114608186972143e-06,
+      "loss": 0.187,
+      "step": 19221
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113641981075507e-06,
+      "loss": 0.1745,
+      "step": 19222
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112675859259779e-06,
+      "loss": 0.1938,
+      "step": 19223
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.111709821538766e-06,
+      "loss": 0.1897,
+      "step": 19224
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1107438679262605e-06,
+      "loss": 0.1684,
+      "step": 19225
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109777998436066e-06,
+      "loss": 0.1765,
+      "step": 19226
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108812213081977e-06,
+      "loss": 0.1868,
+      "step": 19227
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.107846511877784e-06,
+      "loss": 0.2032,
+      "step": 19228
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.106880894837289e-06,
+      "loss": 0.179,
+      "step": 19229
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.105915361974278e-06,
+      "loss": 0.1942,
+      "step": 19230
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.104949913302551e-06,
+      "loss": 0.1794,
+      "step": 19231
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103984548835888e-06,
+      "loss": 0.1809,
+      "step": 19232
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103019268588085e-06,
+      "loss": 0.1922,
+      "step": 19233
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.10205407257293e-06,
+      "loss": 0.1825,
+      "step": 19234
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.101088960804205e-06,
+      "loss": 0.1957,
+      "step": 19235
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100123933295702e-06,
+      "loss": 0.1997,
+      "step": 19236
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.099158990061199e-06,
+      "loss": 0.1865,
+      "step": 19237
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098194131114488e-06,
+      "loss": 0.1842,
+      "step": 19238
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.097229356469344e-06,
+      "loss": 0.1941,
+      "step": 19239
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.096264666139547e-06,
+      "loss": 0.1899,
+      "step": 19240
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095300060138885e-06,
+      "loss": 0.1755,
+      "step": 19241
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.094335538481126e-06,
+      "loss": 0.1936,
+      "step": 19242
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093371101180054e-06,
+      "loss": 0.1742,
+      "step": 19243
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092406748249446e-06,
+      "loss": 0.1682,
+      "step": 19244
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091442479703073e-06,
+      "loss": 0.184,
+      "step": 19245
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.09047829555471e-06,
+      "loss": 0.1783,
+      "step": 19246
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.089514195818125e-06,
+      "loss": 0.17,
+      "step": 19247
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0885501805070985e-06,
+      "loss": 0.1675,
+      "step": 19248
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.087586249635393e-06,
+      "loss": 0.1793,
+      "step": 19249
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.086622403216779e-06,
+      "loss": 0.194,
+      "step": 19250
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085658641265026e-06,
+      "loss": 0.1983,
+      "step": 19251
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.084694963793897e-06,
+      "loss": 0.1868,
+      "step": 19252
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.083731370817164e-06,
+      "loss": 0.1663,
+      "step": 19253
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082767862348585e-06,
+      "loss": 0.1907,
+      "step": 19254
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.081804438401922e-06,
+      "loss": 0.189,
+      "step": 19255
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.080841098990941e-06,
+      "loss": 0.1796,
+      "step": 19256
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079877844129401e-06,
+      "loss": 0.1766,
+      "step": 19257
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.078914673831061e-06,
+      "loss": 0.1769,
+      "step": 19258
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.077951588109673e-06,
+      "loss": 0.1818,
+      "step": 19259
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076988586979004e-06,
+      "loss": 0.1762,
+      "step": 19260
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076025670452803e-06,
+      "loss": 0.1923,
+      "step": 19261
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0750628385448245e-06,
+      "loss": 0.191,
+      "step": 19262
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0741000912688255e-06,
+      "loss": 0.1992,
+      "step": 19263
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0731374286385526e-06,
+      "loss": 0.1901,
+      "step": 19264
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0721748506677624e-06,
+      "loss": 0.1973,
+      "step": 19265
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071212357370201e-06,
+      "loss": 0.1936,
+      "step": 19266
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070249948759615e-06,
+      "loss": 0.1871,
+      "step": 19267
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.069287624849756e-06,
+      "loss": 0.1799,
+      "step": 19268
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068325385654364e-06,
+      "loss": 0.1919,
+      "step": 19269
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067363231187194e-06,
+      "loss": 0.2016,
+      "step": 19270
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066401161461974e-06,
+      "loss": 0.1856,
+      "step": 19271
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065439176492461e-06,
+      "loss": 0.1935,
+      "step": 19272
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.064477276292388e-06,
+      "loss": 0.1731,
+      "step": 19273
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.063515460875493e-06,
+      "loss": 0.1793,
+      "step": 19274
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062553730255523e-06,
+      "loss": 0.172,
+      "step": 19275
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.061592084446205e-06,
+      "loss": 0.1977,
+      "step": 19276
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060630523461286e-06,
+      "loss": 0.1786,
+      "step": 19277
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059669047314495e-06,
+      "loss": 0.1747,
+      "step": 19278
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.058707656019563e-06,
+      "loss": 0.1836,
+      "step": 19279
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.057746349590229e-06,
+      "loss": 0.1943,
+      "step": 19280
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0567851280402215e-06,
+      "loss": 0.1861,
+      "step": 19281
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.055823991383268e-06,
+      "loss": 0.1756,
+      "step": 19282
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054862939633105e-06,
+      "loss": 0.1765,
+      "step": 19283
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05390197280345e-06,
+      "loss": 0.1951,
+      "step": 19284
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.052941090908037e-06,
+      "loss": 0.1949,
+      "step": 19285
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.051980293960584e-06,
+      "loss": 0.1805,
+      "step": 19286
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.051019581974825e-06,
+      "loss": 0.1847,
+      "step": 19287
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.050058954964478e-06,
+      "loss": 0.1868,
+      "step": 19288
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.04909841294326e-06,
+      "loss": 0.1787,
+      "step": 19289
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.048137955924899e-06,
+      "loss": 0.1955,
+      "step": 19290
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.047177583923107e-06,
+      "loss": 0.1878,
+      "step": 19291
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.04621729695161e-06,
+      "loss": 0.1645,
+      "step": 19292
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045257095024121e-06,
+      "loss": 0.1726,
+      "step": 19293
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.044296978154351e-06,
+      "loss": 0.1806,
+      "step": 19294
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043336946356022e-06,
+      "loss": 0.1795,
+      "step": 19295
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.042376999642841e-06,
+      "loss": 0.1879,
+      "step": 19296
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.041417138028528e-06,
+      "loss": 0.1817,
+      "step": 19297
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.040457361526783e-06,
+      "loss": 0.1914,
+      "step": 19298
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0394976701513235e-06,
+      "loss": 0.1684,
+      "step": 19299
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0385380639158544e-06,
+      "loss": 0.2085,
+      "step": 19300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.03757854283408e-06,
+      "loss": 0.185,
+      "step": 19301
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0366191069197135e-06,
+      "loss": 0.1648,
+      "step": 19302
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.035659756186452e-06,
+      "loss": 0.1821,
+      "step": 19303
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034700490648004e-06,
+      "loss": 0.1716,
+      "step": 19304
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033741310318072e-06,
+      "loss": 0.1738,
+      "step": 19305
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0327822152103514e-06,
+      "loss": 0.1965,
+      "step": 19306
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031823205338548e-06,
+      "loss": 0.1937,
+      "step": 19307
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.030864280716356e-06,
+      "loss": 0.168,
+      "step": 19308
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029905441357477e-06,
+      "loss": 0.1741,
+      "step": 19309
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028946687275605e-06,
+      "loss": 0.187,
+      "step": 19310
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.027988018484435e-06,
+      "loss": 0.1855,
+      "step": 19311
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.027029434997659e-06,
+      "loss": 0.1819,
+      "step": 19312
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026070936828967e-06,
+      "loss": 0.1746,
+      "step": 19313
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.025112523992058e-06,
+      "loss": 0.1791,
+      "step": 19314
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.024154196500615e-06,
+      "loss": 0.1783,
+      "step": 19315
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023195954368332e-06,
+      "loss": 0.174,
+      "step": 19316
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.022237797608895e-06,
+      "loss": 0.1756,
+      "step": 19317
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0212797262359845e-06,
+      "loss": 0.1826,
+      "step": 19318
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.020321740263295e-06,
+      "loss": 0.1938,
+      "step": 19319
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.019363839704508e-06,
+      "loss": 0.2043,
+      "step": 19320
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.018406024573298e-06,
+      "loss": 0.1726,
+      "step": 19321
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017448294883357e-06,
+      "loss": 0.1681,
+      "step": 19322
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0164906506483615e-06,
+      "loss": 0.1689,
+      "step": 19323
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.015533091881989e-06,
+      "loss": 0.1992,
+      "step": 19324
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0145756185979146e-06,
+      "loss": 0.1906,
+      "step": 19325
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.013618230809822e-06,
+      "loss": 0.1745,
+      "step": 19326
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012660928531385e-06,
+      "loss": 0.1803,
+      "step": 19327
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.011703711776271e-06,
+      "loss": 0.1809,
+      "step": 19328
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.01074658055816e-06,
+      "loss": 0.1736,
+      "step": 19329
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009789534890719e-06,
+      "loss": 0.1864,
+      "step": 19330
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.008832574787624e-06,
+      "loss": 0.2012,
+      "step": 19331
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007875700262543e-06,
+      "loss": 0.1798,
+      "step": 19332
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.006918911329136e-06,
+      "loss": 0.1828,
+      "step": 19333
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00596220800108e-06,
+      "loss": 0.1784,
+      "step": 19334
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.005005590292034e-06,
+      "loss": 0.1947,
+      "step": 19335
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004049058215671e-06,
+      "loss": 0.1881,
+      "step": 19336
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.003092611785642e-06,
+      "loss": 0.174,
+      "step": 19337
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.002136251015617e-06,
+      "loss": 0.1928,
+      "step": 19338
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.001179975919256e-06,
+      "loss": 0.1873,
+      "step": 19339
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000223786510212e-06,
+      "loss": 0.1919,
+      "step": 19340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999267682802154e-06,
+      "loss": 0.1774,
+      "step": 19341
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.998311664808727e-06,
+      "loss": 0.1704,
+      "step": 19342
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997355732543599e-06,
+      "loss": 0.1879,
+      "step": 19343
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.996399886020418e-06,
+      "loss": 0.1789,
+      "step": 19344
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995444125252834e-06,
+      "loss": 0.1823,
+      "step": 19345
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.994488450254507e-06,
+      "loss": 0.1849,
+      "step": 19346
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99353286103908e-06,
+      "loss": 0.1816,
+      "step": 19347
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99257735762021e-06,
+      "loss": 0.1794,
+      "step": 19348
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.991621940011543e-06,
+      "loss": 0.1774,
+      "step": 19349
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990666608226725e-06,
+      "loss": 0.1902,
+      "step": 19350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.989711362279403e-06,
+      "loss": 0.1596,
+      "step": 19351
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.988756202183216e-06,
+      "loss": 0.1995,
+      "step": 19352
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9878011279518166e-06,
+      "loss": 0.2062,
+      "step": 19353
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.986846139598843e-06,
+      "loss": 0.1841,
+      "step": 19354
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985891237137933e-06,
+      "loss": 0.1904,
+      "step": 19355
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.984936420582734e-06,
+      "loss": 0.1719,
+      "step": 19356
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983981689946877e-06,
+      "loss": 0.1924,
+      "step": 19357
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983027045244006e-06,
+      "loss": 0.1724,
+      "step": 19358
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982072486487754e-06,
+      "loss": 0.1696,
+      "step": 19359
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981118013691753e-06,
+      "loss": 0.1761,
+      "step": 19360
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.980163626869642e-06,
+      "loss": 0.1748,
+      "step": 19361
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979209326035049e-06,
+      "loss": 0.1864,
+      "step": 19362
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.978255111201613e-06,
+      "loss": 0.1722,
+      "step": 19363
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.977300982382953e-06,
+      "loss": 0.1927,
+      "step": 19364
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.976346939592706e-06,
+      "loss": 0.1808,
+      "step": 19365
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975392982844497e-06,
+      "loss": 0.1769,
+      "step": 19366
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.974439112151948e-06,
+      "loss": 0.1852,
+      "step": 19367
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973485327528693e-06,
+      "loss": 0.1903,
+      "step": 19368
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.972531628988347e-06,
+      "loss": 0.1855,
+      "step": 19369
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9715780165445415e-06,
+      "loss": 0.2052,
+      "step": 19370
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970624490210893e-06,
+      "loss": 0.1869,
+      "step": 19371
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.969671050001018e-06,
+      "loss": 0.1691,
+      "step": 19372
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.968717695928543e-06,
+      "loss": 0.1836,
+      "step": 19373
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.967764428007078e-06,
+      "loss": 0.1854,
+      "step": 19374
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9668112462502485e-06,
+      "loss": 0.1738,
+      "step": 19375
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9658581506716645e-06,
+      "loss": 0.197,
+      "step": 19376
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.96490514128494e-06,
+      "loss": 0.1678,
+      "step": 19377
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.963952218103689e-06,
+      "loss": 0.1908,
+      "step": 19378
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962999381141519e-06,
+      "loss": 0.1739,
+      "step": 19379
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962046630412047e-06,
+      "loss": 0.1911,
+      "step": 19380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.961093965928876e-06,
+      "loss": 0.1702,
+      "step": 19381
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.960141387705619e-06,
+      "loss": 0.1922,
+      "step": 19382
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959188895755882e-06,
+      "loss": 0.1723,
+      "step": 19383
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.958236490093265e-06,
+      "loss": 0.202,
+      "step": 19384
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.957284170731379e-06,
+      "loss": 0.1873,
+      "step": 19385
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.956331937683824e-06,
+      "loss": 0.1755,
+      "step": 19386
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.955379790964199e-06,
+      "loss": 0.1812,
+      "step": 19387
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.95442773058611e-06,
+      "loss": 0.1895,
+      "step": 19388
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.953475756563153e-06,
+      "loss": 0.1921,
+      "step": 19389
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.952523868908927e-06,
+      "loss": 0.1916,
+      "step": 19390
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.951572067637026e-06,
+      "loss": 0.1996,
+      "step": 19391
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.950620352761051e-06,
+      "loss": 0.1681,
+      "step": 19392
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949668724294592e-06,
+      "loss": 0.1859,
+      "step": 19393
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948717182251241e-06,
+      "loss": 0.1793,
+      "step": 19394
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947765726644595e-06,
+      "loss": 0.1757,
+      "step": 19395
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.946814357488238e-06,
+      "loss": 0.1639,
+      "step": 19396
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945863074795767e-06,
+      "loss": 0.1802,
+      "step": 19397
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944911878580766e-06,
+      "loss": 0.1832,
+      "step": 19398
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.943960768856819e-06,
+      "loss": 0.1915,
+      "step": 19399
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.943009745637518e-06,
+      "loss": 0.1991,
+      "step": 19400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.942058808936442e-06,
+      "loss": 0.181,
+      "step": 19401
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.941107958767183e-06,
+      "loss": 0.1633,
+      "step": 19402
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940157195143309e-06,
+      "loss": 0.1724,
+      "step": 19403
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.939206518078412e-06,
+      "loss": 0.1951,
+      "step": 19404
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.938255927586068e-06,
+      "loss": 0.2028,
+      "step": 19405
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937305423679851e-06,
+      "loss": 0.1746,
+      "step": 19406
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.936355006373346e-06,
+      "loss": 0.1672,
+      "step": 19407
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9354046756801215e-06,
+      "loss": 0.1548,
+      "step": 19408
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934454431613758e-06,
+      "loss": 0.1925,
+      "step": 19409
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933504274187827e-06,
+      "loss": 0.1944,
+      "step": 19410
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932554203415896e-06,
+      "loss": 0.2079,
+      "step": 19411
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931604219311544e-06,
+      "loss": 0.185,
+      "step": 19412
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930654321888331e-06,
+      "loss": 0.167,
+      "step": 19413
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.929704511159835e-06,
+      "loss": 0.1664,
+      "step": 19414
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928754787139618e-06,
+      "loss": 0.195,
+      "step": 19415
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9278051498412475e-06,
+      "loss": 0.2157,
+      "step": 19416
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926855599278288e-06,
+      "loss": 0.1924,
+      "step": 19417
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925906135464297e-06,
+      "loss": 0.1669,
+      "step": 19418
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924956758412845e-06,
+      "loss": 0.1805,
+      "step": 19419
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9240074681374874e-06,
+      "loss": 0.2052,
+      "step": 19420
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92305826465179e-06,
+      "loss": 0.1661,
+      "step": 19421
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922109147969306e-06,
+      "loss": 0.1738,
+      "step": 19422
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9211601181035915e-06,
+      "loss": 0.2029,
+      "step": 19423
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92021117506821e-06,
+      "loss": 0.172,
+      "step": 19424
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9192623188767085e-06,
+      "loss": 0.1725,
+      "step": 19425
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.918313549542642e-06,
+      "loss": 0.192,
+      "step": 19426
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917364867079566e-06,
+      "loss": 0.1858,
+      "step": 19427
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.91641627150103e-06,
+      "loss": 0.1934,
+      "step": 19428
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915467762820584e-06,
+      "loss": 0.1853,
+      "step": 19429
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9145193410517725e-06,
+      "loss": 0.1684,
+      "step": 19430
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913571006208149e-06,
+      "loss": 0.1886,
+      "step": 19431
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912622758303256e-06,
+      "loss": 0.1661,
+      "step": 19432
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911674597350636e-06,
+      "loss": 0.189,
+      "step": 19433
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.910726523363839e-06,
+      "loss": 0.1936,
+      "step": 19434
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9097785363564004e-06,
+      "loss": 0.1859,
+      "step": 19435
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.908830636341868e-06,
+      "loss": 0.1727,
+      "step": 19436
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907882823333779e-06,
+      "loss": 0.1929,
+      "step": 19437
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.906935097345668e-06,
+      "loss": 0.17,
+      "step": 19438
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.905987458391078e-06,
+      "loss": 0.1719,
+      "step": 19439
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90503990648354e-06,
+      "loss": 0.189,
+      "step": 19440
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.904092441636594e-06,
+      "loss": 0.189,
+      "step": 19441
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9031450638637725e-06,
+      "loss": 0.198,
+      "step": 19442
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902197773178606e-06,
+      "loss": 0.1929,
+      "step": 19443
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901250569594628e-06,
+      "loss": 0.1658,
+      "step": 19444
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.900303453125361e-06,
+      "loss": 0.194,
+      "step": 19445
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.899356423784344e-06,
+      "loss": 0.1931,
+      "step": 19446
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8984094815850946e-06,
+      "loss": 0.1949,
+      "step": 19447
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89746262654115e-06,
+      "loss": 0.1895,
+      "step": 19448
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.896515858666027e-06,
+      "loss": 0.1816,
+      "step": 19449
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.895569177973248e-06,
+      "loss": 0.1943,
+      "step": 19450
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.894622584476344e-06,
+      "loss": 0.1901,
+      "step": 19451
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893676078188826e-06,
+      "loss": 0.1847,
+      "step": 19452
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892729659124223e-06,
+      "loss": 0.1948,
+      "step": 19453
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891783327296049e-06,
+      "loss": 0.1846,
+      "step": 19454
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.890837082717822e-06,
+      "loss": 0.1739,
+      "step": 19455
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88989092540306e-06,
+      "loss": 0.1804,
+      "step": 19456
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888944855365271e-06,
+      "loss": 0.1721,
+      "step": 19457
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8879988726179775e-06,
+      "loss": 0.1832,
+      "step": 19458
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.887052977174688e-06,
+      "loss": 0.175,
+      "step": 19459
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88610716904891e-06,
+      "loss": 0.1741,
+      "step": 19460
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.885161448254162e-06,
+      "loss": 0.1874,
+      "step": 19461
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.884215814803944e-06,
+      "loss": 0.1888,
+      "step": 19462
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883270268711772e-06,
+      "loss": 0.1773,
+      "step": 19463
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882324809991147e-06,
+      "loss": 0.1788,
+      "step": 19464
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.881379438655571e-06,
+      "loss": 0.1741,
+      "step": 19465
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.880434154718555e-06,
+      "loss": 0.1976,
+      "step": 19466
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.879488958193596e-06,
+      "loss": 0.1724,
+      "step": 19467
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.878543849094203e-06,
+      "loss": 0.205,
+      "step": 19468
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.877598827433863e-06,
+      "loss": 0.1926,
+      "step": 19469
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.876653893226084e-06,
+      "loss": 0.192,
+      "step": 19470
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.875709046484364e-06,
+      "loss": 0.1608,
+      "step": 19471
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8747642872221914e-06,
+      "loss": 0.1619,
+      "step": 19472
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873819615453069e-06,
+      "loss": 0.1825,
+      "step": 19473
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.872875031190486e-06,
+      "loss": 0.1746,
+      "step": 19474
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87193053444794e-06,
+      "loss": 0.1951,
+      "step": 19475
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870986125238918e-06,
+      "loss": 0.2056,
+      "step": 19476
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870041803576907e-06,
+      "loss": 0.19,
+      "step": 19477
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.869097569475404e-06,
+      "loss": 0.1767,
+      "step": 19478
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.868153422947889e-06,
+      "loss": 0.2144,
+      "step": 19479
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.867209364007853e-06,
+      "loss": 0.1735,
+      "step": 19480
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.866265392668781e-06,
+      "loss": 0.1659,
+      "step": 19481
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.865321508944154e-06,
+      "loss": 0.1744,
+      "step": 19482
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864377712847456e-06,
+      "loss": 0.1651,
+      "step": 19483
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863434004392164e-06,
+      "loss": 0.1746,
+      "step": 19484
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.862490383591764e-06,
+      "loss": 0.1838,
+      "step": 19485
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.861546850459729e-06,
+      "loss": 0.192,
+      "step": 19486
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860603405009544e-06,
+      "loss": 0.1815,
+      "step": 19487
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859660047254681e-06,
+      "loss": 0.1921,
+      "step": 19488
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.858716777208611e-06,
+      "loss": 0.1943,
+      "step": 19489
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.857773594884816e-06,
+      "loss": 0.1721,
+      "step": 19490
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.856830500296763e-06,
+      "loss": 0.1578,
+      "step": 19491
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8558874934579205e-06,
+      "loss": 0.1744,
+      "step": 19492
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854944574381766e-06,
+      "loss": 0.1737,
+      "step": 19493
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854001743081764e-06,
+      "loss": 0.1746,
+      "step": 19494
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853058999571383e-06,
+      "loss": 0.1958,
+      "step": 19495
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.852116343864085e-06,
+      "loss": 0.1709,
+      "step": 19496
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8511737759733416e-06,
+      "loss": 0.2024,
+      "step": 19497
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850231295912613e-06,
+      "loss": 0.1842,
+      "step": 19498
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.849288903695359e-06,
+      "loss": 0.186,
+      "step": 19499
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848346599335047e-06,
+      "loss": 0.1882,
+      "step": 19500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8474043828451295e-06,
+      "loss": 0.1912,
+      "step": 19501
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.846462254239072e-06,
+      "loss": 0.1924,
+      "step": 19502
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.84552021353033e-06,
+      "loss": 0.1976,
+      "step": 19503
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.844578260732355e-06,
+      "loss": 0.1835,
+      "step": 19504
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.843636395858608e-06,
+      "loss": 0.1909,
+      "step": 19505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8426946189225386e-06,
+      "loss": 0.1784,
+      "step": 19506
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.841752929937606e-06,
+      "loss": 0.181,
+      "step": 19507
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840811328917251e-06,
+      "loss": 0.1838,
+      "step": 19508
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839869815874931e-06,
+      "loss": 0.1845,
+      "step": 19509
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838928390824092e-06,
+      "loss": 0.1997,
+      "step": 19510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.837987053778178e-06,
+      "loss": 0.1844,
+      "step": 19511
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.837045804750642e-06,
+      "loss": 0.1828,
+      "step": 19512
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.836104643754923e-06,
+      "loss": 0.1681,
+      "step": 19513
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835163570804469e-06,
+      "loss": 0.1758,
+      "step": 19514
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.834222585912721e-06,
+      "loss": 0.1888,
+      "step": 19515
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833281689093115e-06,
+      "loss": 0.2005,
+      "step": 19516
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8323408803591e-06,
+      "loss": 0.1768,
+      "step": 19517
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831400159724106e-06,
+      "loss": 0.1785,
+      "step": 19518
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830459527201576e-06,
+      "loss": 0.1632,
+      "step": 19519
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.829518982804945e-06,
+      "loss": 0.1703,
+      "step": 19520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828578526547647e-06,
+      "loss": 0.1612,
+      "step": 19521
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827638158443114e-06,
+      "loss": 0.1797,
+      "step": 19522
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.826697878504777e-06,
+      "loss": 0.1822,
+      "step": 19523
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.825757686746073e-06,
+      "loss": 0.1703,
+      "step": 19524
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.824817583180425e-06,
+      "loss": 0.1891,
+      "step": 19525
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.823877567821268e-06,
+      "loss": 0.1975,
+      "step": 19526
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8229376406820254e-06,
+      "loss": 0.1927,
+      "step": 19527
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821997801776119e-06,
+      "loss": 0.1844,
+      "step": 19528
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821058051116984e-06,
+      "loss": 0.1755,
+      "step": 19529
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820118388718037e-06,
+      "loss": 0.1766,
+      "step": 19530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819178814592696e-06,
+      "loss": 0.174,
+      "step": 19531
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.818239328754393e-06,
+      "loss": 0.1601,
+      "step": 19532
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817299931216537e-06,
+      "loss": 0.1891,
+      "step": 19533
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816360621992559e-06,
+      "loss": 0.1962,
+      "step": 19534
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.815421401095859e-06,
+      "loss": 0.1981,
+      "step": 19535
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8144822685398676e-06,
+      "loss": 0.1834,
+      "step": 19536
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.813543224337992e-06,
+      "loss": 0.1674,
+      "step": 19537
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8126042685036457e-06,
+      "loss": 0.1833,
+      "step": 19538
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.811665401050245e-06,
+      "loss": 0.1731,
+      "step": 19539
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8107266219911943e-06,
+      "loss": 0.1762,
+      "step": 19540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8097879313399112e-06,
+      "loss": 0.1727,
+      "step": 19541
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8088493291097996e-06,
+      "loss": 0.184,
+      "step": 19542
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.807910815314263e-06,
+      "loss": 0.1872,
+      "step": 19543
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8069723899667143e-06,
+      "loss": 0.1925,
+      "step": 19544
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8060340530805505e-06,
+      "loss": 0.1728,
+      "step": 19545
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8050958046691834e-06,
+      "loss": 0.1681,
+      "step": 19546
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8041576447460095e-06,
+      "loss": 0.1872,
+      "step": 19547
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8032195733244314e-06,
+      "loss": 0.1695,
+      "step": 19548
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8022815904178468e-06,
+      "loss": 0.1684,
+      "step": 19549
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.801343696039651e-06,
+      "loss": 0.1849,
+      "step": 19550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8004058902032494e-06,
+      "loss": 0.1871,
+      "step": 19551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7994681729220275e-06,
+      "loss": 0.2117,
+      "step": 19552
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.79853054420939e-06,
+      "loss": 0.19,
+      "step": 19553
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7975930040787236e-06,
+      "loss": 0.1808,
+      "step": 19554
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.796655552543419e-06,
+      "loss": 0.1651,
+      "step": 19555
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7957181896168727e-06,
+      "loss": 0.1612,
+      "step": 19556
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.794780915312467e-06,
+      "loss": 0.1738,
+      "step": 19557
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.793843729643598e-06,
+      "loss": 0.1849,
+      "step": 19558
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7929066326236475e-06,
+      "loss": 0.2046,
+      "step": 19559
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7919696242660007e-06,
+      "loss": 0.1657,
+      "step": 19560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7910327045840435e-06,
+      "loss": 0.1868,
+      "step": 19561
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7900958735911554e-06,
+      "loss": 0.1896,
+      "step": 19562
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7891591313007235e-06,
+      "loss": 0.1961,
+      "step": 19563
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7882224777261255e-06,
+      "loss": 0.1776,
+      "step": 19564
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7872859128807384e-06,
+      "loss": 0.1949,
+      "step": 19565
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7863494367779454e-06,
+      "loss": 0.1743,
+      "step": 19566
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.785413049431116e-06,
+      "loss": 0.2013,
+      "step": 19567
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.784476750853634e-06,
+      "loss": 0.1926,
+      "step": 19568
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7835405410588686e-06,
+      "loss": 0.1834,
+      "step": 19569
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.782604420060191e-06,
+      "loss": 0.1847,
+      "step": 19570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.781668387870977e-06,
+      "loss": 0.1837,
+      "step": 19571
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7807324445045924e-06,
+      "loss": 0.1803,
+      "step": 19572
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.779796589974416e-06,
+      "loss": 0.1592,
+      "step": 19573
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.778860824293802e-06,
+      "loss": 0.1865,
+      "step": 19574
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.777925147476126e-06,
+      "loss": 0.1904,
+      "step": 19575
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7769895595347517e-06,
+      "loss": 0.1824,
+      "step": 19576
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7760540604830374e-06,
+      "loss": 0.182,
+      "step": 19577
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.775118650334354e-06,
+      "loss": 0.1705,
+      "step": 19578
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.774183329102056e-06,
+      "loss": 0.1831,
+      "step": 19579
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.77324809679951e-06,
+      "loss": 0.172,
+      "step": 19580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7723129534400727e-06,
+      "loss": 0.17,
+      "step": 19581
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7713778990370962e-06,
+      "loss": 0.1957,
+      "step": 19582
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.770442933603945e-06,
+      "loss": 0.1737,
+      "step": 19583
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7695080571539677e-06,
+      "loss": 0.1655,
+      "step": 19584
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.768573269700524e-06,
+      "loss": 0.1945,
+      "step": 19585
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.767638571256963e-06,
+      "loss": 0.1775,
+      "step": 19586
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7667039618366375e-06,
+      "loss": 0.1805,
+      "step": 19587
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7657694414528954e-06,
+      "loss": 0.181,
+      "step": 19588
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.764835010119083e-06,
+      "loss": 0.1947,
+      "step": 19589
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7639006678485547e-06,
+      "loss": 0.1946,
+      "step": 19590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7629664146546495e-06,
+      "loss": 0.1812,
+      "step": 19591
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.76203225055072e-06,
+      "loss": 0.2059,
+      "step": 19592
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7610981755501053e-06,
+      "loss": 0.1776,
+      "step": 19593
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7601641896661444e-06,
+      "loss": 0.1922,
+      "step": 19594
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.759230292912186e-06,
+      "loss": 0.174,
+      "step": 19595
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7582964853015657e-06,
+      "loss": 0.1662,
+      "step": 19596
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.75736276684762e-06,
+      "loss": 0.1746,
+      "step": 19597
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.756429137563691e-06,
+      "loss": 0.1953,
+      "step": 19598
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.755495597463108e-06,
+      "loss": 0.1577,
+      "step": 19599
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7545621465592185e-06,
+      "loss": 0.1814,
+      "step": 19600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7536287848653396e-06,
+      "loss": 0.188,
+      "step": 19601
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.752695512394815e-06,
+      "loss": 0.1855,
+      "step": 19602
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7517623291609717e-06,
+      "loss": 0.1793,
+      "step": 19603
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7508292351771358e-06,
+      "loss": 0.1634,
+      "step": 19604
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7498962304566435e-06,
+      "loss": 0.1767,
+      "step": 19605
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7489633150128135e-06,
+      "loss": 0.1929,
+      "step": 19606
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.748030488858979e-06,
+      "loss": 0.1839,
+      "step": 19607
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.747097752008462e-06,
+      "loss": 0.1586,
+      "step": 19608
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.746165104474582e-06,
+      "loss": 0.1902,
+      "step": 19609
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7452325462706663e-06,
+      "loss": 0.1996,
+      "step": 19610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.74430007741003e-06,
+      "loss": 0.1937,
+      "step": 19611
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7433676979059996e-06,
+      "loss": 0.1797,
+      "step": 19612
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7424354077718894e-06,
+      "loss": 0.1793,
+      "step": 19613
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7415032070210166e-06,
+      "loss": 0.1796,
+      "step": 19614
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7405710956666963e-06,
+      "loss": 0.1542,
+      "step": 19615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7396390737222386e-06,
+      "loss": 0.1583,
+      "step": 19616
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7387071412009656e-06,
+      "loss": 0.1807,
+      "step": 19617
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7377752981161807e-06,
+      "loss": 0.2152,
+      "step": 19618
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.736843544481201e-06,
+      "loss": 0.1719,
+      "step": 19619
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.735911880309333e-06,
+      "loss": 0.1722,
+      "step": 19620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.734980305613881e-06,
+      "loss": 0.1703,
+      "step": 19621
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7340488204081583e-06,
+      "loss": 0.182,
+      "step": 19622
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7331174247054637e-06,
+      "loss": 0.1834,
+      "step": 19623
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.732186118519108e-06,
+      "loss": 0.1844,
+      "step": 19624
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.73125490186239e-06,
+      "loss": 0.1618,
+      "step": 19625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.730323774748612e-06,
+      "loss": 0.1798,
+      "step": 19626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729392737191074e-06,
+      "loss": 0.1592,
+      "step": 19627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7284617892030716e-06,
+      "loss": 0.1767,
+      "step": 19628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.727530930797909e-06,
+      "loss": 0.1876,
+      "step": 19629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.726600161988877e-06,
+      "loss": 0.1678,
+      "step": 19630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7256694827892757e-06,
+      "loss": 0.1829,
+      "step": 19631
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7247388932123974e-06,
+      "loss": 0.1878,
+      "step": 19632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7238083932715296e-06,
+      "loss": 0.1807,
+      "step": 19633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.722877982979971e-06,
+      "loss": 0.1847,
+      "step": 19634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7219476623510085e-06,
+      "loss": 0.1706,
+      "step": 19635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7210174313979275e-06,
+      "loss": 0.174,
+      "step": 19636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7200872901340214e-06,
+      "loss": 0.1742,
+      "step": 19637
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7191572385725706e-06,
+      "loss": 0.1789,
+      "step": 19638
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.71822727672687e-06,
+      "loss": 0.1653,
+      "step": 19639
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7172974046101897e-06,
+      "loss": 0.1899,
+      "step": 19640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.71636762223582e-06,
+      "loss": 0.1742,
+      "step": 19641
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7154379296170418e-06,
+      "loss": 0.1849,
+      "step": 19642
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.71450832676713e-06,
+      "loss": 0.1751,
+      "step": 19643
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.71357881369937e-06,
+      "loss": 0.1655,
+      "step": 19644
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.712649390427032e-06,
+      "loss": 0.2277,
+      "step": 19645
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7117200569633995e-06,
+      "loss": 0.2082,
+      "step": 19646
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.710790813321742e-06,
+      "loss": 0.1712,
+      "step": 19647
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7098616595153304e-06,
+      "loss": 0.1972,
+      "step": 19648
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.708932595557444e-06,
+      "loss": 0.156,
+      "step": 19649
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.708003621461347e-06,
+      "loss": 0.1896,
+      "step": 19650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.707074737240315e-06,
+      "loss": 0.1838,
+      "step": 19651
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.706145942907612e-06,
+      "loss": 0.1723,
+      "step": 19652
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7052172384765075e-06,
+      "loss": 0.1847,
+      "step": 19653
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7042886239602636e-06,
+      "loss": 0.1986,
+      "step": 19654
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.703360099372144e-06,
+      "loss": 0.1652,
+      "step": 19655
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702431664725419e-06,
+      "loss": 0.1799,
+      "step": 19656
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7015033200333407e-06,
+      "loss": 0.1509,
+      "step": 19657
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7005750653091787e-06,
+      "loss": 0.1946,
+      "step": 19658
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6996469005661873e-06,
+      "loss": 0.2044,
+      "step": 19659
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6987188258176234e-06,
+      "loss": 0.1778,
+      "step": 19660
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.697790841076748e-06,
+      "loss": 0.2015,
+      "step": 19661
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6968629463568107e-06,
+      "loss": 0.2037,
+      "step": 19662
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.695935141671072e-06,
+      "loss": 0.1863,
+      "step": 19663
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.695007427032782e-06,
+      "loss": 0.181,
+      "step": 19664
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6940798024551914e-06,
+      "loss": 0.1845,
+      "step": 19665
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6931522679515507e-06,
+      "loss": 0.1663,
+      "step": 19666
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6922248235351066e-06,
+      "loss": 0.1762,
+      "step": 19667
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6912974692191115e-06,
+      "loss": 0.1848,
+      "step": 19668
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69037020501681e-06,
+      "loss": 0.1988,
+      "step": 19669
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6894430309414418e-06,
+      "loss": 0.1671,
+      "step": 19670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6885159470062594e-06,
+      "loss": 0.1715,
+      "step": 19671
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.687588953224498e-06,
+      "loss": 0.1853,
+      "step": 19672
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6866620496094064e-06,
+      "loss": 0.1796,
+      "step": 19673
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.685735236174218e-06,
+      "loss": 0.1903,
+      "step": 19674
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.684808512932172e-06,
+      "loss": 0.1948,
+      "step": 19675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6838818798965103e-06,
+      "loss": 0.1615,
+      "step": 19676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.682955337080463e-06,
+      "loss": 0.1672,
+      "step": 19677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.682028884497271e-06,
+      "loss": 0.1801,
+      "step": 19678
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6811025221601636e-06,
+      "loss": 0.1946,
+      "step": 19679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6801762500823756e-06,
+      "loss": 0.1801,
+      "step": 19680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.679250068277135e-06,
+      "loss": 0.1671,
+      "step": 19681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6783239767576707e-06,
+      "loss": 0.188,
+      "step": 19682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6773979755372147e-06,
+      "loss": 0.1912,
+      "step": 19683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6764720646289907e-06,
+      "loss": 0.1906,
+      "step": 19684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6755462440462288e-06,
+      "loss": 0.1847,
+      "step": 19685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6746205138021505e-06,
+      "loss": 0.1824,
+      "step": 19686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6736948739099764e-06,
+      "loss": 0.1768,
+      "step": 19687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.672769324382934e-06,
+      "loss": 0.1831,
+      "step": 19688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.671843865234238e-06,
+      "loss": 0.1841,
+      "step": 19689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.670918496477114e-06,
+      "loss": 0.1795,
+      "step": 19690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.669993218124778e-06,
+      "loss": 0.1786,
+      "step": 19691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6690680301904435e-06,
+      "loss": 0.1728,
+      "step": 19692
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6681429326873287e-06,
+      "loss": 0.182,
+      "step": 19693
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6672179256286432e-06,
+      "loss": 0.1813,
+      "step": 19694
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6662930090276085e-06,
+      "loss": 0.1765,
+      "step": 19695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6653681828974264e-06,
+      "loss": 0.1708,
+      "step": 19696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6644434472513144e-06,
+      "loss": 0.17,
+      "step": 19697
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6635188021024804e-06,
+      "loss": 0.1715,
+      "step": 19698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6625942474641273e-06,
+      "loss": 0.1922,
+      "step": 19699
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6616697833494674e-06,
+      "loss": 0.1801,
+      "step": 19700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6607454097716998e-06,
+      "loss": 0.18,
+      "step": 19701
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6598211267440353e-06,
+      "loss": 0.1835,
+      "step": 19702
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6588969342796733e-06,
+      "loss": 0.1602,
+      "step": 19703
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6579728323918096e-06,
+      "loss": 0.1774,
+      "step": 19704
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.657048821093657e-06,
+      "loss": 0.1733,
+      "step": 19705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.656124900398399e-06,
+      "loss": 0.1644,
+      "step": 19706
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.655201070319243e-06,
+      "loss": 0.1695,
+      "step": 19707
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.654277330869381e-06,
+      "loss": 0.1831,
+      "step": 19708
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6533536820620074e-06,
+      "loss": 0.1769,
+      "step": 19709
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6524301239103187e-06,
+      "loss": 0.19,
+      "step": 19710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6515066564275027e-06,
+      "loss": 0.197,
+      "step": 19711
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6505832796267547e-06,
+      "loss": 0.1732,
+      "step": 19712
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.649659993521264e-06,
+      "loss": 0.1903,
+      "step": 19713
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.648736798124213e-06,
+      "loss": 0.187,
+      "step": 19714
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6478136934487962e-06,
+      "loss": 0.1768,
+      "step": 19715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6468906795081926e-06,
+      "loss": 0.1654,
+      "step": 19716
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6459677563155924e-06,
+      "loss": 0.1726,
+      "step": 19717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.645044923884177e-06,
+      "loss": 0.1587,
+      "step": 19718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6441221822271266e-06,
+      "loss": 0.2115,
+      "step": 19719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.643199531357623e-06,
+      "loss": 0.1871,
+      "step": 19720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6422769712888406e-06,
+      "loss": 0.1898,
+      "step": 19721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641354502033966e-06,
+      "loss": 0.1554,
+      "step": 19722
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640432123606167e-06,
+      "loss": 0.1755,
+      "step": 19723
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6395098360186256e-06,
+      "loss": 0.1901,
+      "step": 19724
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6385876392845145e-06,
+      "loss": 0.1813,
+      "step": 19725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.637665533417001e-06,
+      "loss": 0.1906,
+      "step": 19726
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.636743518429263e-06,
+      "loss": 0.186,
+      "step": 19727
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6358215943344664e-06,
+      "loss": 0.1595,
+      "step": 19728
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6348997611457837e-06,
+      "loss": 0.1701,
+      "step": 19729
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6339780188763817e-06,
+      "loss": 0.186,
+      "step": 19730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6330563675394247e-06,
+      "loss": 0.1786,
+      "step": 19731
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.632134807148079e-06,
+      "loss": 0.1784,
+      "step": 19732
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6312133377155023e-06,
+      "loss": 0.1867,
+      "step": 19733
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.630291959254867e-06,
+      "loss": 0.1682,
+      "step": 19734
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6293706717793253e-06,
+      "loss": 0.1685,
+      "step": 19735
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6284494753020438e-06,
+      "loss": 0.1773,
+      "step": 19736
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6275283698361785e-06,
+      "loss": 0.1709,
+      "step": 19737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.626607355394881e-06,
+      "loss": 0.1684,
+      "step": 19738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6256864319913164e-06,
+      "loss": 0.1858,
+      "step": 19739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6247655996386343e-06,
+      "loss": 0.185,
+      "step": 19740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6238448583499854e-06,
+      "loss": 0.1889,
+      "step": 19741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6229242081385286e-06,
+      "loss": 0.1847,
+      "step": 19742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.622003649017405e-06,
+      "loss": 0.2004,
+      "step": 19743
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.621083180999777e-06,
+      "loss": 0.1936,
+      "step": 19744
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.620162804098779e-06,
+      "loss": 0.1821,
+      "step": 19745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6192425183275657e-06,
+      "loss": 0.1737,
+      "step": 19746
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6183223236992803e-06,
+      "loss": 0.1829,
+      "step": 19747
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6174022202270644e-06,
+      "loss": 0.1708,
+      "step": 19748
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6164822079240667e-06,
+      "loss": 0.1705,
+      "step": 19749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6155622868034213e-06,
+      "loss": 0.1813,
+      "step": 19750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6146424568782765e-06,
+      "loss": 0.1689,
+      "step": 19751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.613722718161766e-06,
+      "loss": 0.1662,
+      "step": 19752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.612803070667026e-06,
+      "loss": 0.1699,
+      "step": 19753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6118835144071986e-06,
+      "loss": 0.2031,
+      "step": 19754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6109640493954112e-06,
+      "loss": 0.1837,
+      "step": 19755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.610044675644806e-06,
+      "loss": 0.2051,
+      "step": 19756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.60912539316851e-06,
+      "loss": 0.1831,
+      "step": 19757
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.608206201979656e-06,
+      "loss": 0.195,
+      "step": 19758
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6072871020913726e-06,
+      "loss": 0.1808,
+      "step": 19759
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6063680935167855e-06,
+      "loss": 0.1765,
+      "step": 19760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.605449176269029e-06,
+      "loss": 0.1667,
+      "step": 19761
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6045303503612207e-06,
+      "loss": 0.1854,
+      "step": 19762
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6036116158064927e-06,
+      "loss": 0.1836,
+      "step": 19763
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.602692972617966e-06,
+      "loss": 0.1834,
+      "step": 19764
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6017744208087567e-06,
+      "loss": 0.1571,
+      "step": 19765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6008559603919946e-06,
+      "loss": 0.2083,
+      "step": 19766
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.599937591380791e-06,
+      "loss": 0.1766,
+      "step": 19767
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59901931378827e-06,
+      "loss": 0.1821,
+      "step": 19768
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5981011276275467e-06,
+      "loss": 0.1766,
+      "step": 19769
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5971830329117318e-06,
+      "loss": 0.189,
+      "step": 19770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5962650296539504e-06,
+      "loss": 0.1783,
+      "step": 19771
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5953471178673014e-06,
+      "loss": 0.1734,
+      "step": 19772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5944292975649055e-06,
+      "loss": 0.1852,
+      "step": 19773
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.593511568759871e-06,
+      "loss": 0.1726,
+      "step": 19774
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5925939314653025e-06,
+      "loss": 0.173,
+      "step": 19775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.591676385694315e-06,
+      "loss": 0.2058,
+      "step": 19776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.590758931460007e-06,
+      "loss": 0.1747,
+      "step": 19777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5898415687754908e-06,
+      "loss": 0.1889,
+      "step": 19778
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5889242976538662e-06,
+      "loss": 0.181,
+      "step": 19779
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.588007118108233e-06,
+      "loss": 0.1728,
+      "step": 19780
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5870900301516985e-06,
+      "loss": 0.1745,
+      "step": 19781
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5861730337973566e-06,
+      "loss": 0.1951,
+      "step": 19782
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.58525612905831e-06,
+      "loss": 0.1721,
+      "step": 19783
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5843393159476547e-06,
+      "loss": 0.1705,
+      "step": 19784
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.583422594478485e-06,
+      "loss": 0.1668,
+      "step": 19785
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5825059646638973e-06,
+      "loss": 0.186,
+      "step": 19786
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5815894265169795e-06,
+      "loss": 0.1816,
+      "step": 19787
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.580672980050831e-06,
+      "loss": 0.1909,
+      "step": 19788
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5797566252785355e-06,
+      "loss": 0.1815,
+      "step": 19789
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5788403622131885e-06,
+      "loss": 0.1851,
+      "step": 19790
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5779241908678754e-06,
+      "loss": 0.1686,
+      "step": 19791
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.57700811125568e-06,
+      "loss": 0.1788,
+      "step": 19792
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5760921233896918e-06,
+      "loss": 0.1814,
+      "step": 19793
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5751762272829903e-06,
+      "loss": 0.1733,
+      "step": 19794
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5742604229486655e-06,
+      "loss": 0.1727,
+      "step": 19795
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.573344710399793e-06,
+      "loss": 0.1684,
+      "step": 19796
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.572429089649454e-06,
+      "loss": 0.1586,
+      "step": 19797
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5715135607107276e-06,
+      "loss": 0.189,
+      "step": 19798
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.570598123596688e-06,
+      "loss": 0.183,
+      "step": 19799
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5696827783204168e-06,
+      "loss": 0.1781,
+      "step": 19800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.568767524894984e-06,
+      "loss": 0.1825,
+      "step": 19801
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5678523633334683e-06,
+      "loss": 0.1894,
+      "step": 19802
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.566937293648939e-06,
+      "loss": 0.1883,
+      "step": 19803
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.566022315854464e-06,
+      "loss": 0.1769,
+      "step": 19804
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5651074299631194e-06,
+      "loss": 0.2049,
+      "step": 19805
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5641926359879663e-06,
+      "loss": 0.1812,
+      "step": 19806
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5632779339420787e-06,
+      "loss": 0.1874,
+      "step": 19807
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.562363323838518e-06,
+      "loss": 0.1617,
+      "step": 19808
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5614488056903473e-06,
+      "loss": 0.1723,
+      "step": 19809
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.560534379510637e-06,
+      "loss": 0.1771,
+      "step": 19810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5596200453124386e-06,
+      "loss": 0.1806,
+      "step": 19811
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5587058031088195e-06,
+      "loss": 0.1715,
+      "step": 19812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5577916529128367e-06,
+      "loss": 0.1926,
+      "step": 19813
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5568775947375446e-06,
+      "loss": 0.1802,
+      "step": 19814
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.555963628596005e-06,
+      "loss": 0.1858,
+      "step": 19815
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5550497545012696e-06,
+      "loss": 0.1815,
+      "step": 19816
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.554135972466395e-06,
+      "loss": 0.1638,
+      "step": 19817
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.553222282504434e-06,
+      "loss": 0.1915,
+      "step": 19818
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5523086846284303e-06,
+      "loss": 0.1905,
+      "step": 19819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5513951788514444e-06,
+      "loss": 0.1729,
+      "step": 19820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5504817651865166e-06,
+      "loss": 0.1684,
+      "step": 19821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5495684436467e-06,
+      "loss": 0.1724,
+      "step": 19822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.548655214245038e-06,
+      "loss": 0.1833,
+      "step": 19823
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.547742076994576e-06,
+      "loss": 0.1625,
+      "step": 19824
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.546829031908355e-06,
+      "loss": 0.1889,
+      "step": 19825
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.545916078999416e-06,
+      "loss": 0.1796,
+      "step": 19826
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.545003218280805e-06,
+      "loss": 0.1699,
+      "step": 19827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.544090449765555e-06,
+      "loss": 0.1973,
+      "step": 19828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5431777734667105e-06,
+      "loss": 0.1884,
+      "step": 19829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.542265189397305e-06,
+      "loss": 0.188,
+      "step": 19830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5413526975703703e-06,
+      "loss": 0.1859,
+      "step": 19831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.540440297998947e-06,
+      "loss": 0.1867,
+      "step": 19832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.539527990696063e-06,
+      "loss": 0.1889,
+      "step": 19833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.538615775674754e-06,
+      "loss": 0.2029,
+      "step": 19834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.537703652948048e-06,
+      "loss": 0.1832,
+      "step": 19835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.53679162252897e-06,
+      "loss": 0.1691,
+      "step": 19836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5358796844305577e-06,
+      "loss": 0.1962,
+      "step": 19837
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534967838665824e-06,
+      "loss": 0.1762,
+      "step": 19838
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5340560852478033e-06,
+      "loss": 0.1773,
+      "step": 19839
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.533144424189513e-06,
+      "loss": 0.1841,
+      "step": 19840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.532232855503982e-06,
+      "loss": 0.1658,
+      "step": 19841
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5313213792042265e-06,
+      "loss": 0.1981,
+      "step": 19842
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5304099953032654e-06,
+      "loss": 0.1768,
+      "step": 19843
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5294987038141205e-06,
+      "loss": 0.1643,
+      "step": 19844
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5285875047498075e-06,
+      "loss": 0.1982,
+      "step": 19845
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5276763981233376e-06,
+      "loss": 0.2002,
+      "step": 19846
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5267653839477335e-06,
+      "loss": 0.2128,
+      "step": 19847
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.525854462235998e-06,
+      "loss": 0.1759,
+      "step": 19848
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5249436330011533e-06,
+      "loss": 0.1733,
+      "step": 19849
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5240328962562033e-06,
+      "loss": 0.1801,
+      "step": 19850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5231222520141595e-06,
+      "loss": 0.1719,
+      "step": 19851
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5222117002880272e-06,
+      "loss": 0.1707,
+      "step": 19852
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.521301241090811e-06,
+      "loss": 0.1763,
+      "step": 19853
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.520390874435523e-06,
+      "loss": 0.1702,
+      "step": 19854
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.519480600335159e-06,
+      "loss": 0.1795,
+      "step": 19855
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.518570418802728e-06,
+      "loss": 0.1845,
+      "step": 19856
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.517660329851228e-06,
+      "loss": 0.1762,
+      "step": 19857
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.516750333493656e-06,
+      "loss": 0.1839,
+      "step": 19858
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.515840429743017e-06,
+      "loss": 0.1907,
+      "step": 19859
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.514930618612301e-06,
+      "loss": 0.1997,
+      "step": 19860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5140209001145097e-06,
+      "loss": 0.1844,
+      "step": 19861
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.513111274262636e-06,
+      "loss": 0.1645,
+      "step": 19862
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5122017410696717e-06,
+      "loss": 0.1775,
+      "step": 19863
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.511292300548609e-06,
+      "loss": 0.1778,
+      "step": 19864
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5103829527124355e-06,
+      "loss": 0.1923,
+      "step": 19865
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5094736975741463e-06,
+      "loss": 0.1802,
+      "step": 19866
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.508564535146722e-06,
+      "loss": 0.1884,
+      "step": 19867
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.507655465443157e-06,
+      "loss": 0.1768,
+      "step": 19868
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.506746488476432e-06,
+      "loss": 0.1852,
+      "step": 19869
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.50583760425953e-06,
+      "loss": 0.1818,
+      "step": 19870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504928812805437e-06,
+      "loss": 0.1942,
+      "step": 19871
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.50402011412713e-06,
+      "loss": 0.1888,
+      "step": 19872
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5031115082375934e-06,
+      "loss": 0.1765,
+      "step": 19873
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.502202995149805e-06,
+      "loss": 0.1903,
+      "step": 19874
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5012945748767357e-06,
+      "loss": 0.18,
+      "step": 19875
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5003862474313733e-06,
+      "loss": 0.1962,
+      "step": 19876
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.499478012826679e-06,
+      "loss": 0.1759,
+      "step": 19877
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4985698710756353e-06,
+      "loss": 0.1814,
+      "step": 19878
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.497661822191212e-06,
+      "loss": 0.1584,
+      "step": 19879
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4967538661863755e-06,
+      "loss": 0.1836,
+      "step": 19880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4958460030741003e-06,
+      "loss": 0.166,
+      "step": 19881
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4949382328673507e-06,
+      "loss": 0.1757,
+      "step": 19882
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4940305555790977e-06,
+      "loss": 0.1709,
+      "step": 19883
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4931229712223047e-06,
+      "loss": 0.1919,
+      "step": 19884
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4922154798099318e-06,
+      "loss": 0.1809,
+      "step": 19885
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.491308081354947e-06,
+      "loss": 0.1656,
+      "step": 19886
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4904007758703073e-06,
+      "loss": 0.1703,
+      "step": 19887
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4894935633689776e-06,
+      "loss": 0.1898,
+      "step": 19888
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4885864438639126e-06,
+      "loss": 0.2032,
+      "step": 19889
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.487679417368073e-06,
+      "loss": 0.1763,
+      "step": 19890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4867724838944105e-06,
+      "loss": 0.2072,
+      "step": 19891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4858656434558803e-06,
+      "loss": 0.1807,
+      "step": 19892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.484958896065439e-06,
+      "loss": 0.1844,
+      "step": 19893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4840522417360345e-06,
+      "loss": 0.2012,
+      "step": 19894
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.483145680480623e-06,
+      "loss": 0.1646,
+      "step": 19895
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4822392123121505e-06,
+      "loss": 0.1843,
+      "step": 19896
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4813328372435627e-06,
+      "loss": 0.1908,
+      "step": 19897
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4804265552878112e-06,
+      "loss": 0.174,
+      "step": 19898
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479520366457836e-06,
+      "loss": 0.1718,
+      "step": 19899
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.478614270766587e-06,
+      "loss": 0.1701,
+      "step": 19900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4777082682270045e-06,
+      "loss": 0.2001,
+      "step": 19901
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4768023588520294e-06,
+      "loss": 0.1899,
+      "step": 19902
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.475896542654601e-06,
+      "loss": 0.1652,
+      "step": 19903
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.474990819647656e-06,
+      "loss": 0.1552,
+      "step": 19904
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4740851898441377e-06,
+      "loss": 0.1945,
+      "step": 19905
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4731796532569752e-06,
+      "loss": 0.186,
+      "step": 19906
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.472274209899111e-06,
+      "loss": 0.1739,
+      "step": 19907
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.471368859783474e-06,
+      "loss": 0.2002,
+      "step": 19908
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.470463602922993e-06,
+      "loss": 0.1706,
+      "step": 19909
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4695584393306057e-06,
+      "loss": 0.184,
+      "step": 19910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4686533690192358e-06,
+      "loss": 0.191,
+      "step": 19911
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.467748392001816e-06,
+      "loss": 0.166,
+      "step": 19912
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.466843508291271e-06,
+      "loss": 0.1623,
+      "step": 19913
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4659387179005234e-06,
+      "loss": 0.1861,
+      "step": 19914
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4650340208425013e-06,
+      "loss": 0.1921,
+      "step": 19915
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4641294171301274e-06,
+      "loss": 0.1683,
+      "step": 19916
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4632249067763214e-06,
+      "loss": 0.1692,
+      "step": 19917
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.462320489794003e-06,
+      "loss": 0.18,
+      "step": 19918
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4614161661960877e-06,
+      "loss": 0.1603,
+      "step": 19919
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.460511935995501e-06,
+      "loss": 0.1771,
+      "step": 19920
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4596077992051513e-06,
+      "loss": 0.1742,
+      "step": 19921
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.458703755837959e-06,
+      "loss": 0.1744,
+      "step": 19922
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4577998059068354e-06,
+      "loss": 0.2089,
+      "step": 19923
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.456895949424689e-06,
+      "loss": 0.1779,
+      "step": 19924
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4559921864044365e-06,
+      "loss": 0.1963,
+      "step": 19925
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.455088516858982e-06,
+      "loss": 0.167,
+      "step": 19926
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4541849408012397e-06,
+      "loss": 0.1712,
+      "step": 19927
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4532814582441123e-06,
+      "loss": 0.168,
+      "step": 19928
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4523780692005047e-06,
+      "loss": 0.1689,
+      "step": 19929
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.451474773683322e-06,
+      "loss": 0.1644,
+      "step": 19930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.450571571705463e-06,
+      "loss": 0.1685,
+      "step": 19931
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.449668463279837e-06,
+      "loss": 0.1767,
+      "step": 19932
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4487654484193367e-06,
+      "loss": 0.1694,
+      "step": 19933
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.447862527136867e-06,
+      "loss": 0.1803,
+      "step": 19934
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4469596994453214e-06,
+      "loss": 0.1803,
+      "step": 19935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4460569653575936e-06,
+      "loss": 0.1925,
+      "step": 19936
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4451543248865847e-06,
+      "loss": 0.1899,
+      "step": 19937
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.444251778045181e-06,
+      "loss": 0.1709,
+      "step": 19938
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4433493248462824e-06,
+      "loss": 0.1743,
+      "step": 19939
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4424469653027735e-06,
+      "loss": 0.1652,
+      "step": 19940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4415446994275437e-06,
+      "loss": 0.1798,
+      "step": 19941
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.440642527233489e-06,
+      "loss": 0.2033,
+      "step": 19942
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4397404487334827e-06,
+      "loss": 0.1734,
+      "step": 19943
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4388384639404216e-06,
+      "loss": 0.1534,
+      "step": 19944
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.437936572867181e-06,
+      "loss": 0.1947,
+      "step": 19945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.437034775526652e-06,
+      "loss": 0.1821,
+      "step": 19946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.436133071931711e-06,
+      "loss": 0.1769,
+      "step": 19947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4352314620952355e-06,
+      "loss": 0.1907,
+      "step": 19948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4343299460301106e-06,
+      "loss": 0.1832,
+      "step": 19949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4334285237492102e-06,
+      "loss": 0.1713,
+      "step": 19950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.432527195265407e-06,
+      "loss": 0.1818,
+      "step": 19951
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4316259605915827e-06,
+      "loss": 0.1829,
+      "step": 19952
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.430724819740603e-06,
+      "loss": 0.18,
+      "step": 19953
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.429823772725347e-06,
+      "loss": 0.1905,
+      "step": 19954
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428922819558682e-06,
+      "loss": 0.1629,
+      "step": 19955
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428021960253477e-06,
+      "loss": 0.1915,
+      "step": 19956
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4271211948226e-06,
+      "loss": 0.1872,
+      "step": 19957
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4262205232789147e-06,
+      "loss": 0.1749,
+      "step": 19958
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4253199456352924e-06,
+      "loss": 0.1786,
+      "step": 19959
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.42441946190459e-06,
+      "loss": 0.1952,
+      "step": 19960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.423519072099678e-06,
+      "loss": 0.1979,
+      "step": 19961
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.422618776233413e-06,
+      "loss": 0.1959,
+      "step": 19962
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.421718574318652e-06,
+      "loss": 0.1768,
+      "step": 19963
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4208184663682596e-06,
+      "loss": 0.1837,
+      "step": 19964
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419918452395088e-06,
+      "loss": 0.1875,
+      "step": 19965
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4190185324119973e-06,
+      "loss": 0.2051,
+      "step": 19966
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.41811870643184e-06,
+      "loss": 0.1713,
+      "step": 19967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.41721897446747e-06,
+      "loss": 0.1994,
+      "step": 19968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4163193365317373e-06,
+      "loss": 0.1933,
+      "step": 19969
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.41541979263749e-06,
+      "loss": 0.176,
+      "step": 19970
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414520342797585e-06,
+      "loss": 0.1926,
+      "step": 19971
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4136209870248607e-06,
+      "loss": 0.1827,
+      "step": 19972
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.412721725332172e-06,
+      "loss": 0.1834,
+      "step": 19973
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4118225577323617e-06,
+      "loss": 0.1881,
+      "step": 19974
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4109234842382677e-06,
+      "loss": 0.1945,
+      "step": 19975
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4100245048627402e-06,
+      "loss": 0.1602,
+      "step": 19976
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.409125619618614e-06,
+      "loss": 0.1836,
+      "step": 19977
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4082268285187346e-06,
+      "loss": 0.2106,
+      "step": 19978
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4073281315759376e-06,
+      "loss": 0.1744,
+      "step": 19979
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4064295288030568e-06,
+      "loss": 0.1803,
+      "step": 19980
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.405531020212938e-06,
+      "loss": 0.1806,
+      "step": 19981
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.404632605818401e-06,
+      "loss": 0.179,
+      "step": 19982
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4037342856322896e-06,
+      "loss": 0.1894,
+      "step": 19983
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4028360596674326e-06,
+      "loss": 0.1696,
+      "step": 19984
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.401937927936656e-06,
+      "loss": 0.1975,
+      "step": 19985
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4010398904527965e-06,
+      "loss": 0.1761,
+      "step": 19986
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4001419472286736e-06,
+      "loss": 0.1882,
+      "step": 19987
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3992440982771213e-06,
+      "loss": 0.1669,
+      "step": 19988
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.398346343610961e-06,
+      "loss": 0.1741,
+      "step": 19989
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3974486832430133e-06,
+      "loss": 0.1767,
+      "step": 19990
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3965511171861052e-06,
+      "loss": 0.1717,
+      "step": 19991
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395653645453054e-06,
+      "loss": 0.1909,
+      "step": 19992
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3947562680566824e-06,
+      "loss": 0.1851,
+      "step": 19993
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3938589850098082e-06,
+      "loss": 0.1903,
+      "step": 19994
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.392961796325247e-06,
+      "loss": 0.1934,
+      "step": 19995
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.392064702015814e-06,
+      "loss": 0.1741,
+      "step": 19996
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3911677020943203e-06,
+      "loss": 0.1756,
+      "step": 19997
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3902707965735848e-06,
+      "loss": 0.1833,
+      "step": 19998
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3893739854664133e-06,
+      "loss": 0.1823,
+      "step": 19999
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3884772687856217e-06,
+      "loss": 0.1846,
+      "step": 20000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3875806465440152e-06,
+      "loss": 0.1679,
+      "step": 20001
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3866841187543988e-06,
+      "loss": 0.1938,
+      "step": 20002
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3857876854295846e-06,
+      "loss": 0.1864,
+      "step": 20003
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3848913465823717e-06,
+      "loss": 0.1832,
+      "step": 20004
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3839951022255678e-06,
+      "loss": 0.1882,
+      "step": 20005
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3830989523719735e-06,
+      "loss": 0.1763,
+      "step": 20006
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3822028970343855e-06,
+      "loss": 0.185,
+      "step": 20007
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3813069362256135e-06,
+      "loss": 0.1792,
+      "step": 20008
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3804110699584403e-06,
+      "loss": 0.1609,
+      "step": 20009
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3795152982456745e-06,
+      "loss": 0.203,
+      "step": 20010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3786196211001045e-06,
+      "loss": 0.178,
+      "step": 20011
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.37772403853453e-06,
+      "loss": 0.1884,
+      "step": 20012
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3768285505617404e-06,
+      "loss": 0.2039,
+      "step": 20013
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.375933157194524e-06,
+      "loss": 0.171,
+      "step": 20014
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3750378584456767e-06,
+      "loss": 0.1856,
+      "step": 20015
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.374142654327981e-06,
+      "loss": 0.1691,
+      "step": 20016
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.37324754485423e-06,
+      "loss": 0.2016,
+      "step": 20017
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.372352530037206e-06,
+      "loss": 0.1956,
+      "step": 20018
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3714576098896913e-06,
+      "loss": 0.1825,
+      "step": 20019
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3705627844244737e-06,
+      "loss": 0.199,
+      "step": 20020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3696680536543335e-06,
+      "loss": 0.1771,
+      "step": 20021
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 0.1928,
+      "step": 20022
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3678788762504033e-06,
+      "loss": 0.1793,
+      "step": 20023
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.366984429642166e-06,
+      "loss": 0.1968,
+      "step": 20024
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.366090077780122e-06,
+      "loss": 0.1654,
+      "step": 20025
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.36519582067704e-06,
+      "loss": 0.1829,
+      "step": 20026
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3643016583456987e-06,
+      "loss": 0.1762,
+      "step": 20027
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3634075907988694e-06,
+      "loss": 0.1853,
+      "step": 20028
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3625136180493177e-06,
+      "loss": 0.1916,
+      "step": 20029
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3616197401098205e-06,
+      "loss": 0.1634,
+      "step": 20030
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3607259569931393e-06,
+      "loss": 0.1928,
+      "step": 20031
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359832268712048e-06,
+      "loss": 0.169,
+      "step": 20032
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3589386752793086e-06,
+      "loss": 0.1925,
+      "step": 20033
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.358045176707684e-06,
+      "loss": 0.1668,
+      "step": 20034
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.357151773009938e-06,
+      "loss": 0.1881,
+      "step": 20035
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3562584641988283e-06,
+      "loss": 0.1775,
+      "step": 20036
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3553652502871226e-06,
+      "loss": 0.1823,
+      "step": 20037
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3544721312875715e-06,
+      "loss": 0.1768,
+      "step": 20038
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3535791072129408e-06,
+      "loss": 0.1671,
+      "step": 20039
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.352686178075981e-06,
+      "loss": 0.1741,
+      "step": 20040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3517933438894447e-06,
+      "loss": 0.2016,
+      "step": 20041
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.350900604666092e-06,
+      "loss": 0.1786,
+      "step": 20042
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3500079604186674e-06,
+      "loss": 0.1778,
+      "step": 20043
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.349115411159929e-06,
+      "loss": 0.1651,
+      "step": 20044
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3482229569026214e-06,
+      "loss": 0.1708,
+      "step": 20045
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3473305976594893e-06,
+      "loss": 0.1861,
+      "step": 20046
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3464383334432903e-06,
+      "loss": 0.171,
+      "step": 20047
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3455461642667562e-06,
+      "loss": 0.1995,
+      "step": 20048
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3446540901426395e-06,
+      "loss": 0.2019,
+      "step": 20049
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.343762111083677e-06,
+      "loss": 0.1882,
+      "step": 20050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.342870227102616e-06,
+      "loss": 0.185,
+      "step": 20051
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.341978438212192e-06,
+      "loss": 0.1839,
+      "step": 20052
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3410867444251426e-06,
+      "loss": 0.1555,
+      "step": 20053
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.340195145754209e-06,
+      "loss": 0.1909,
+      "step": 20054
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.339303642212124e-06,
+      "loss": 0.1954,
+      "step": 20055
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.338412233811619e-06,
+      "loss": 0.1888,
+      "step": 20056
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3375209205654325e-06,
+      "loss": 0.1713,
+      "step": 20057
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3366297024862913e-06,
+      "loss": 0.1655,
+      "step": 20058
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.335738579586931e-06,
+      "loss": 0.1765,
+      "step": 20059
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.334847551880077e-06,
+      "loss": 0.1878,
+      "step": 20060
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.333956619378457e-06,
+      "loss": 0.1699,
+      "step": 20061
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3330657820947975e-06,
+      "loss": 0.1805,
+      "step": 20062
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3321750400418194e-06,
+      "loss": 0.1965,
+      "step": 20063
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3312843932322524e-06,
+      "loss": 0.1769,
+      "step": 20064
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.330393841678813e-06,
+      "loss": 0.1836,
+      "step": 20065
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.329503385394228e-06,
+      "loss": 0.1913,
+      "step": 20066
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3286130243912128e-06,
+      "loss": 0.1767,
+      "step": 20067
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3277227586824825e-06,
+      "loss": 0.1921,
+      "step": 20068
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3268325882807606e-06,
+      "loss": 0.1847,
+      "step": 20069
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3259425131987556e-06,
+      "loss": 0.1875,
+      "step": 20070
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3250525334491877e-06,
+      "loss": 0.1877,
+      "step": 20071
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3241626490447655e-06,
+      "loss": 0.178,
+      "step": 20072
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3232728599981978e-06,
+      "loss": 0.1695,
+      "step": 20073
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.322383166322204e-06,
+      "loss": 0.1937,
+      "step": 20074
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3214935680294804e-06,
+      "loss": 0.1945,
+      "step": 20075
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.320604065132741e-06,
+      "loss": 0.1811,
+      "step": 20076
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.319714657644688e-06,
+      "loss": 0.1721,
+      "step": 20077
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3188253455780308e-06,
+      "loss": 0.1966,
+      "step": 20078
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3179361289454694e-06,
+      "loss": 0.1925,
+      "step": 20079
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3170470077597006e-06,
+      "loss": 0.1669,
+      "step": 20080
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3161579820334345e-06,
+      "loss": 0.1812,
+      "step": 20081
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.315269051779361e-06,
+      "loss": 0.175,
+      "step": 20082
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3143802170101836e-06,
+      "loss": 0.182,
+      "step": 20083
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3134914777385975e-06,
+      "loss": 0.1859,
+      "step": 20084
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312602833977293e-06,
+      "loss": 0.1945,
+      "step": 20085
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3117142857389694e-06,
+      "loss": 0.1703,
+      "step": 20086
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3108258330363163e-06,
+      "loss": 0.1917,
+      "step": 20087
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3099374758820234e-06,
+      "loss": 0.1972,
+      "step": 20088
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309049214288782e-06,
+      "loss": 0.1575,
+      "step": 20089
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308161048269275e-06,
+      "loss": 0.1796,
+      "step": 20090
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3072729778361966e-06,
+      "loss": 0.1735,
+      "step": 20091
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.306385003002225e-06,
+      "loss": 0.1843,
+      "step": 20092
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.30549712378005e-06,
+      "loss": 0.1852,
+      "step": 20093
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304609340182352e-06,
+      "loss": 0.1738,
+      "step": 20094
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3037216522218074e-06,
+      "loss": 0.1807,
+      "step": 20095
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.302834059911104e-06,
+      "loss": 0.1805,
+      "step": 20096
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3019465632629132e-06,
+      "loss": 0.1818,
+      "step": 20097
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301059162289918e-06,
+      "loss": 0.1936,
+      "step": 20098
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3001718570047913e-06,
+      "loss": 0.1862,
+      "step": 20099
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2992846474202078e-06,
+      "loss": 0.193,
+      "step": 20100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2983975335488395e-06,
+      "loss": 0.1959,
+      "step": 20101
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.297510515403355e-06,
+      "loss": 0.1635,
+      "step": 20102
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.296623592996431e-06,
+      "loss": 0.1787,
+      "step": 20103
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2957367663407302e-06,
+      "loss": 0.1784,
+      "step": 20104
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2948500354489255e-06,
+      "loss": 0.1755,
+      "step": 20105
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.293963400333682e-06,
+      "loss": 0.1732,
+      "step": 20106
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.293076861007658e-06,
+      "loss": 0.1774,
+      "step": 20107
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.292190417483526e-06,
+      "loss": 0.1861,
+      "step": 20108
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2913040697739408e-06,
+      "loss": 0.1771,
+      "step": 20109
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.290417817891569e-06,
+      "loss": 0.1963,
+      "step": 20110
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.289531661849067e-06,
+      "loss": 0.2022,
+      "step": 20111
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2886456016590893e-06,
+      "loss": 0.1803,
+      "step": 20112
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2877596373343024e-06,
+      "loss": 0.1794,
+      "step": 20113
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2868737688873474e-06,
+      "loss": 0.1827,
+      "step": 20114
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2859879963308892e-06,
+      "loss": 0.1817,
+      "step": 20115
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.285102319677573e-06,
+      "loss": 0.1836,
+      "step": 20116
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.284216738940057e-06,
+      "loss": 0.1826,
+      "step": 20117
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.283331254130987e-06,
+      "loss": 0.1771,
+      "step": 20118
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2824458652630075e-06,
+      "loss": 0.1642,
+      "step": 20119
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.281560572348772e-06,
+      "loss": 0.1678,
+      "step": 20120
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2806753754009214e-06,
+      "loss": 0.1889,
+      "step": 20121
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.279790274432104e-06,
+      "loss": 0.1831,
+      "step": 20122
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2789052694549605e-06,
+      "loss": 0.1779,
+      "step": 20123
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2780203604821294e-06,
+      "loss": 0.1868,
+      "step": 20124
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2771355475262567e-06,
+      "loss": 0.1686,
+      "step": 20125
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2762508305999784e-06,
+      "loss": 0.1783,
+      "step": 20126
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.275366209715931e-06,
+      "loss": 0.1819,
+      "step": 20127
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.274481684886751e-06,
+      "loss": 0.1696,
+      "step": 20128
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2735972561250696e-06,
+      "loss": 0.1922,
+      "step": 20129
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2727129234435264e-06,
+      "loss": 0.1821,
+      "step": 20130
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2718286868547467e-06,
+      "loss": 0.1765,
+      "step": 20131
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2709445463713684e-06,
+      "loss": 0.1887,
+      "step": 20132
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.270060502006016e-06,
+      "loss": 0.1714,
+      "step": 20133
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.269176553771315e-06,
+      "loss": 0.1879,
+      "step": 20134
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.268292701679897e-06,
+      "loss": 0.1755,
+      "step": 20135
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2674089457443814e-06,
+      "loss": 0.1797,
+      "step": 20136
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2665252859773978e-06,
+      "loss": 0.1971,
+      "step": 20137
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.265641722391566e-06,
+      "loss": 0.1964,
+      "step": 20138
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.264758254999504e-06,
+      "loss": 0.1792,
+      "step": 20139
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.263874883813839e-06,
+      "loss": 0.1775,
+      "step": 20140
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.262991608847177e-06,
+      "loss": 0.1887,
+      "step": 20141
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2621084301121454e-06,
+      "loss": 0.185,
+      "step": 20142
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.261225347621352e-06,
+      "loss": 0.1801,
+      "step": 20143
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2603423613874174e-06,
+      "loss": 0.1837,
+      "step": 20144
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2594594714229523e-06,
+      "loss": 0.1685,
+      "step": 20145
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2585766777405626e-06,
+      "loss": 0.1716,
+      "step": 20146
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.257693980352865e-06,
+      "loss": 0.1866,
+      "step": 20147
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.256811379272463e-06,
+      "loss": 0.2074,
+      "step": 20148
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2559288745119687e-06,
+      "loss": 0.199,
+      "step": 20149
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2550464660839854e-06,
+      "loss": 0.1771,
+      "step": 20150
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.254164154001115e-06,
+      "loss": 0.1902,
+      "step": 20151
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.253281938275964e-06,
+      "loss": 0.1743,
+      "step": 20152
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2523998189211335e-06,
+      "loss": 0.1629,
+      "step": 20153
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.251517795949224e-06,
+      "loss": 0.195,
+      "step": 20154
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250635869372829e-06,
+      "loss": 0.178,
+      "step": 20155
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2497540392045534e-06,
+      "loss": 0.1729,
+      "step": 20156
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2488723054569905e-06,
+      "loss": 0.1872,
+      "step": 20157
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24799066814273e-06,
+      "loss": 0.174,
+      "step": 20158
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2471091272743747e-06,
+      "loss": 0.1828,
+      "step": 20159
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2462276828645112e-06,
+      "loss": 0.1852,
+      "step": 20160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2453463349257264e-06,
+      "loss": 0.1682,
+      "step": 20161
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2444650834706184e-06,
+      "loss": 0.1707,
+      "step": 20162
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2435839285117653e-06,
+      "loss": 0.1579,
+      "step": 20163
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2427028700617634e-06,
+      "loss": 0.1894,
+      "step": 20164
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2418219081331924e-06,
+      "loss": 0.1959,
+      "step": 20165
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.240941042738636e-06,
+      "loss": 0.183,
+      "step": 20166
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.240060273890677e-06,
+      "loss": 0.2112,
+      "step": 20167
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239179601601894e-06,
+      "loss": 0.1751,
+      "step": 20168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2382990258848713e-06,
+      "loss": 0.1759,
+      "step": 20169
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237418546752181e-06,
+      "loss": 0.1746,
+      "step": 20170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2365381642164074e-06,
+      "loss": 0.1847,
+      "step": 20171
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.235657878290123e-06,
+      "loss": 0.1576,
+      "step": 20172
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2347776889858963e-06,
+      "loss": 0.1591,
+      "step": 20173
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2338975963163078e-06,
+      "loss": 0.1753,
+      "step": 20174
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2330176002939227e-06,
+      "loss": 0.1809,
+      "step": 20175
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.232137700931316e-06,
+      "loss": 0.181,
+      "step": 20176
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.231257898241055e-06,
+      "loss": 0.1673,
+      "step": 20177
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2303781922357014e-06,
+      "loss": 0.1923,
+      "step": 20178
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2294985829278326e-06,
+      "loss": 0.1596,
+      "step": 20179
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.228619070329999e-06,
+      "loss": 0.1801,
+      "step": 20180
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2277396544547734e-06,
+      "loss": 0.1695,
+      "step": 20181
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.226860335314711e-06,
+      "loss": 0.168,
+      "step": 20182
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2259811129223784e-06,
+      "loss": 0.1827,
+      "step": 20183
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.225101987290331e-06,
+      "loss": 0.1703,
+      "step": 20184
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2242229584311237e-06,
+      "loss": 0.1942,
+      "step": 20185
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.223344026357318e-06,
+      "loss": 0.1774,
+      "step": 20186
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.222465191081463e-06,
+      "loss": 0.1877,
+      "step": 20187
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.221586452616119e-06,
+      "loss": 0.1819,
+      "step": 20188
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.220707810973833e-06,
+      "loss": 0.1805,
+      "step": 20189
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2198292661671536e-06,
+      "loss": 0.1944,
+      "step": 20190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2189508182086358e-06,
+      "loss": 0.1777,
+      "step": 20191
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2180724671108244e-06,
+      "loss": 0.1876,
+      "step": 20192
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2171942128862667e-06,
+      "loss": 0.1928,
+      "step": 20193
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2163160555475025e-06,
+      "loss": 0.169,
+      "step": 20194
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2154379951070836e-06,
+      "loss": 0.1684,
+      "step": 20195
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214560031577548e-06,
+      "loss": 0.1998,
+      "step": 20196
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2136821649714332e-06,
+      "loss": 0.1689,
+      "step": 20197
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2128043953012866e-06,
+      "loss": 0.168,
+      "step": 20198
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2119267225796414e-06,
+      "loss": 0.171,
+      "step": 20199
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2110491468190318e-06,
+      "loss": 0.1779,
+      "step": 20200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2101716680320004e-06,
+      "loss": 0.1824,
+      "step": 20201
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.209294286231073e-06,
+      "loss": 0.1552,
+      "step": 20202
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2084170014287895e-06,
+      "loss": 0.1933,
+      "step": 20203
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.207539813637678e-06,
+      "loss": 0.1849,
+      "step": 20204
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2066627228702673e-06,
+      "loss": 0.1833,
+      "step": 20205
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.205785729139088e-06,
+      "loss": 0.1718,
+      "step": 20206
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.204908832456661e-06,
+      "loss": 0.1784,
+      "step": 20207
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.20403203283552e-06,
+      "loss": 0.1729,
+      "step": 20208
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2031553302881835e-06,
+      "loss": 0.1835,
+      "step": 20209
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.202278724827179e-06,
+      "loss": 0.2064,
+      "step": 20210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.201402216465025e-06,
+      "loss": 0.1828,
+      "step": 20211
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2005258052142396e-06,
+      "loss": 0.1841,
+      "step": 20212
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1996494910873476e-06,
+      "loss": 0.1816,
+      "step": 20213
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.198773274096859e-06,
+      "loss": 0.1955,
+      "step": 20214
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1978971542552974e-06,
+      "loss": 0.1797,
+      "step": 20215
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.197021131575172e-06,
+      "loss": 0.1857,
+      "step": 20216
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1961452060689947e-06,
+      "loss": 0.1687,
+      "step": 20217
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.195269377749284e-06,
+      "loss": 0.1794,
+      "step": 20218
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1943936466285453e-06,
+      "loss": 0.1829,
+      "step": 20219
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.193518012719288e-06,
+      "loss": 0.1981,
+      "step": 20220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.192642476034017e-06,
+      "loss": 0.1901,
+      "step": 20221
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.191767036585245e-06,
+      "loss": 0.1772,
+      "step": 20222
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.190891694385473e-06,
+      "loss": 0.2065,
+      "step": 20223
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.190016449447202e-06,
+      "loss": 0.1804,
+      "step": 20224
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.18914130178294e-06,
+      "loss": 0.1742,
+      "step": 20225
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.188266251405181e-06,
+      "loss": 0.2126,
+      "step": 20226
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.187391298326431e-06,
+      "loss": 0.1815,
+      "step": 20227
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1865164425591845e-06,
+      "loss": 0.1951,
+      "step": 20228
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1856416841159343e-06,
+      "loss": 0.1737,
+      "step": 20229
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.184767023009182e-06,
+      "loss": 0.1843,
+      "step": 20230
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1838924592514197e-06,
+      "loss": 0.1801,
+      "step": 20231
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1830179928551373e-06,
+      "loss": 0.2019,
+      "step": 20232
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1821436238328273e-06,
+      "loss": 0.1488,
+      "step": 20233
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1812693521969764e-06,
+      "loss": 0.1771,
+      "step": 20234
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1803951779600774e-06,
+      "loss": 0.1883,
+      "step": 20235
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1795211011346117e-06,
+      "loss": 0.1785,
+      "step": 20236
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1786471217330707e-06,
+      "loss": 0.1724,
+      "step": 20237
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1777732397679352e-06,
+      "loss": 0.1815,
+      "step": 20238
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1768994552516843e-06,
+      "loss": 0.192,
+      "step": 20239
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1760257681968065e-06,
+      "loss": 0.1629,
+      "step": 20240
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.175152178615775e-06,
+      "loss": 0.1896,
+      "step": 20241
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.174278686521073e-06,
+      "loss": 0.1948,
+      "step": 20242
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1734052919251758e-06,
+      "loss": 0.1859,
+      "step": 20243
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.172531994840555e-06,
+      "loss": 0.1782,
+      "step": 20244
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1716587952796964e-06,
+      "loss": 0.1804,
+      "step": 20245
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.170785693255057e-06,
+      "loss": 0.1802,
+      "step": 20246
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1699126887791197e-06,
+      "loss": 0.1735,
+      "step": 20247
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1690397818643483e-06,
+      "loss": 0.1816,
+      "step": 20248
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1681669725232165e-06,
+      "loss": 0.1862,
+      "step": 20249
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.16729426076819e-06,
+      "loss": 0.2009,
+      "step": 20250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.166421646611729e-06,
+      "loss": 0.1828,
+      "step": 20251
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1655491300663067e-06,
+      "loss": 0.1623,
+      "step": 20252
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1646767111443787e-06,
+      "loss": 0.1916,
+      "step": 20253
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1638043898584135e-06,
+      "loss": 0.1679,
+      "step": 20254
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.162932166220868e-06,
+      "loss": 0.1866,
+      "step": 20255
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.162060040244197e-06,
+      "loss": 0.1657,
+      "step": 20256
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1611880119408657e-06,
+      "loss": 0.1845,
+      "step": 20257
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1603160813233267e-06,
+      "loss": 0.1818,
+      "step": 20258
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1594442484040345e-06,
+      "loss": 0.1817,
+      "step": 20259
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1585725131954383e-06,
+      "loss": 0.1856,
+      "step": 20260
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1577008757099967e-06,
+      "loss": 0.1862,
+      "step": 20261
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1568293359601586e-06,
+      "loss": 0.1924,
+      "step": 20262
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.155957893958369e-06,
+      "loss": 0.2016,
+      "step": 20263
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1550865497170815e-06,
+      "loss": 0.1673,
+      "step": 20264
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1542153032487388e-06,
+      "loss": 0.1963,
+      "step": 20265
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.153344154565784e-06,
+      "loss": 0.1884,
+      "step": 20266
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1524731036806666e-06,
+      "loss": 0.1803,
+      "step": 20267
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1516021506058213e-06,
+      "loss": 0.1837,
+      "step": 20268
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1507312953536963e-06,
+      "loss": 0.1702,
+      "step": 20269
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1498605379367274e-06,
+      "loss": 0.2024,
+      "step": 20270
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1489898783673524e-06,
+      "loss": 0.1952,
+      "step": 20271
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.148119316658007e-06,
+      "loss": 0.1582,
+      "step": 20272
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1472488528211255e-06,
+      "loss": 0.1738,
+      "step": 20273
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.146378486869146e-06,
+      "loss": 0.1738,
+      "step": 20274
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1455082188144948e-06,
+      "loss": 0.2003,
+      "step": 20275
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1446380486696103e-06,
+      "loss": 0.1624,
+      "step": 20276
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1437679764469176e-06,
+      "loss": 0.1686,
+      "step": 20277
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.142898002158841e-06,
+      "loss": 0.185,
+      "step": 20278
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1420281258178154e-06,
+      "loss": 0.1885,
+      "step": 20279
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1411583474362595e-06,
+      "loss": 0.1735,
+      "step": 20280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.140288667026603e-06,
+      "loss": 0.1721,
+      "step": 20281
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1394190846012652e-06,
+      "loss": 0.1835,
+      "step": 20282
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1385496001726635e-06,
+      "loss": 0.1597,
+      "step": 20283
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1376802137532293e-06,
+      "loss": 0.2,
+      "step": 20284
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.136810925355366e-06,
+      "loss": 0.1931,
+      "step": 20285
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.135941734991501e-06,
+      "loss": 0.1906,
+      "step": 20286
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.135072642674043e-06,
+      "loss": 0.1621,
+      "step": 20287
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.134203648415414e-06,
+      "loss": 0.1807,
+      "step": 20288
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1333347522280213e-06,
+      "loss": 0.1872,
+      "step": 20289
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1324659541242754e-06,
+      "loss": 0.1996,
+      "step": 20290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1315972541165905e-06,
+      "loss": 0.1744,
+      "step": 20291
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1307286522173695e-06,
+      "loss": 0.1759,
+      "step": 20292
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1298601484390267e-06,
+      "loss": 0.1923,
+      "step": 20293
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.128991742793964e-06,
+      "loss": 0.1814,
+      "step": 20294
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.128123435294582e-06,
+      "loss": 0.166,
+      "step": 20295
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1272552259532918e-06,
+      "loss": 0.1743,
+      "step": 20296
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1263871147824864e-06,
+      "loss": 0.1876,
+      "step": 20297
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1255191017945776e-06,
+      "loss": 0.1898,
+      "step": 20298
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.124651187001949e-06,
+      "loss": 0.1846,
+      "step": 20299
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1237833704170084e-06,
+      "loss": 0.1786,
+      "step": 20300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.12291565205215e-06,
+      "loss": 0.1777,
+      "step": 20301
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.122048031919763e-06,
+      "loss": 0.1677,
+      "step": 20302
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.121180510032249e-06,
+      "loss": 0.1913,
+      "step": 20303
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1203130864019936e-06,
+      "loss": 0.179,
+      "step": 20304
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1194457610413873e-06,
+      "loss": 0.1916,
+      "step": 20305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.118578533962823e-06,
+      "loss": 0.1936,
+      "step": 20306
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.117711405178683e-06,
+      "loss": 0.1778,
+      "step": 20307
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1168443747013587e-06,
+      "loss": 0.1574,
+      "step": 20308
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.115977442543233e-06,
+      "loss": 0.1849,
+      "step": 20309
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1151106087166847e-06,
+      "loss": 0.1824,
+      "step": 20310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1142438732341063e-06,
+      "loss": 0.173,
+      "step": 20311
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.113377236107864e-06,
+      "loss": 0.1912,
+      "step": 20312
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.112510697350348e-06,
+      "loss": 0.1721,
+      "step": 20313
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1116442569739304e-06,
+      "loss": 0.1845,
+      "step": 20314
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110777914990991e-06,
+      "loss": 0.1699,
+      "step": 20315
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1099116714139032e-06,
+      "loss": 0.1744,
+      "step": 20316
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.109045526255037e-06,
+      "loss": 0.1793,
+      "step": 20317
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1081794795267706e-06,
+      "loss": 0.1846,
+      "step": 20318
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1073135312414683e-06,
+      "loss": 0.1884,
+      "step": 20319
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1064476814115065e-06,
+      "loss": 0.1746,
+      "step": 20320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1055819300492485e-06,
+      "loss": 0.1786,
+      "step": 20321
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1047162771670592e-06,
+      "loss": 0.1927,
+      "step": 20322
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1038507227773084e-06,
+      "loss": 0.1837,
+      "step": 20323
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1029852668923576e-06,
+      "loss": 0.1811,
+      "step": 20324
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.102119909524568e-06,
+      "loss": 0.1839,
+      "step": 20325
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1012546506862986e-06,
+      "loss": 0.1894,
+      "step": 20326
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1003894903899136e-06,
+      "loss": 0.1805,
+      "step": 20327
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099524428647768e-06,
+      "loss": 0.1662,
+      "step": 20328
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.098659465472217e-06,
+      "loss": 0.1883,
+      "step": 20329
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09779460087562e-06,
+      "loss": 0.1776,
+      "step": 20330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0969298348703246e-06,
+      "loss": 0.1945,
+      "step": 20331
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0960651674686915e-06,
+      "loss": 0.1799,
+      "step": 20332
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.095200598683066e-06,
+      "loss": 0.1843,
+      "step": 20333
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0943361285257956e-06,
+      "loss": 0.1985,
+      "step": 20334
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0934717570092352e-06,
+      "loss": 0.1684,
+      "step": 20335
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0926074841457277e-06,
+      "loss": 0.1823,
+      "step": 20336
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091743309947618e-06,
+      "loss": 0.2018,
+      "step": 20337
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.090879234427251e-06,
+      "loss": 0.1814,
+      "step": 20338
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0900152575969653e-06,
+      "loss": 0.178,
+      "step": 20339
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0891513794691088e-06,
+      "loss": 0.1801,
+      "step": 20340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0882876000560135e-06,
+      "loss": 0.1664,
+      "step": 20341
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0874239193700263e-06,
+      "loss": 0.181,
+      "step": 20342
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0865603374234786e-06,
+      "loss": 0.1568,
+      "step": 20343
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0856968542287046e-06,
+      "loss": 0.1764,
+      "step": 20344
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.084833469798043e-06,
+      "loss": 0.1753,
+      "step": 20345
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.083970184143822e-06,
+      "loss": 0.1834,
+      "step": 20346
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0831069972783766e-06,
+      "loss": 0.2095,
+      "step": 20347
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0822439092140353e-06,
+      "loss": 0.1886,
+      "step": 20348
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.081380919963123e-06,
+      "loss": 0.1871,
+      "step": 20349
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0805180295379757e-06,
+      "loss": 0.1812,
+      "step": 20350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.079655237950906e-06,
+      "loss": 0.185,
+      "step": 20351
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0787925452142477e-06,
+      "loss": 0.1964,
+      "step": 20352
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.077929951340318e-06,
+      "loss": 0.1753,
+      "step": 20353
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0770674563414447e-06,
+      "loss": 0.183,
+      "step": 20354
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076205060229943e-06,
+      "loss": 0.1872,
+      "step": 20355
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0753427630181297e-06,
+      "loss": 0.1882,
+      "step": 20356
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0744805647183286e-06,
+      "loss": 0.1773,
+      "step": 20357
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.073618465342847e-06,
+      "loss": 0.1839,
+      "step": 20358
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0727564649040066e-06,
+      "loss": 0.1795,
+      "step": 20359
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0718945634141174e-06,
+      "loss": 0.1613,
+      "step": 20360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0710327608854863e-06,
+      "loss": 0.1964,
+      "step": 20361
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.070171057330431e-06,
+      "loss": 0.1883,
+      "step": 20362
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0693094527612577e-06,
+      "loss": 0.1988,
+      "step": 20363
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0684479471902717e-06,
+      "loss": 0.1569,
+      "step": 20364
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0675865406297766e-06,
+      "loss": 0.195,
+      "step": 20365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.066725233092083e-06,
+      "loss": 0.1773,
+      "step": 20366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.06586402458949e-06,
+      "loss": 0.1893,
+      "step": 20367
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.065002915134296e-06,
+      "loss": 0.1694,
+      "step": 20368
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.064141904738809e-06,
+      "loss": 0.1794,
+      "step": 20369
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.063280993415323e-06,
+      "loss": 0.1654,
+      "step": 20370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0624201811761332e-06,
+      "loss": 0.1611,
+      "step": 20371
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0615594680335404e-06,
+      "loss": 0.1722,
+      "step": 20372
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.060698853999834e-06,
+      "loss": 0.1735,
+      "step": 20373
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.059838339087313e-06,
+      "loss": 0.1738,
+      "step": 20374
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.058977923308265e-06,
+      "loss": 0.169,
+      "step": 20375
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0581176066749787e-06,
+      "loss": 0.1861,
+      "step": 20376
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0572573891997514e-06,
+      "loss": 0.1825,
+      "step": 20377
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0563972708948574e-06,
+      "loss": 0.1846,
+      "step": 20378
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0555372517725922e-06,
+      "loss": 0.174,
+      "step": 20379
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.054677331845235e-06,
+      "loss": 0.1818,
+      "step": 20380
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0538175111250746e-06,
+      "loss": 0.1796,
+      "step": 20381
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0529577896243902e-06,
+      "loss": 0.1816,
+      "step": 20382
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052098167355457e-06,
+      "loss": 0.1692,
+      "step": 20383
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.051238644330563e-06,
+      "loss": 0.1688,
+      "step": 20384
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0503792205619776e-06,
+      "loss": 0.1783,
+      "step": 20385
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0495198960619833e-06,
+      "loss": 0.1846,
+      "step": 20386
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.048660670842851e-06,
+      "loss": 0.1823,
+      "step": 20387
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0478015449168517e-06,
+      "loss": 0.1784,
+      "step": 20388
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.046942518296263e-06,
+      "loss": 0.1733,
+      "step": 20389
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0460835909933527e-06,
+      "loss": 0.1769,
+      "step": 20390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.04522476302039e-06,
+      "loss": 0.1801,
+      "step": 20391
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0443660343896375e-06,
+      "loss": 0.1623,
+      "step": 20392
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0435074051133697e-06,
+      "loss": 0.1859,
+      "step": 20393
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0426488752038464e-06,
+      "loss": 0.1679,
+      "step": 20394
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0417904446733293e-06,
+      "loss": 0.1824,
+      "step": 20395
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.040932113534085e-06,
+      "loss": 0.1976,
+      "step": 20396
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0400738817983687e-06,
+      "loss": 0.1696,
+      "step": 20397
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0392157494784456e-06,
+      "loss": 0.1733,
+      "step": 20398
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.03835771658657e-06,
+      "loss": 0.1913,
+      "step": 20399
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.037499783134994e-06,
+      "loss": 0.2099,
+      "step": 20400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.03664194913598e-06,
+      "loss": 0.1888,
+      "step": 20401
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0357842146017745e-06,
+      "loss": 0.197,
+      "step": 20402
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.034926579544639e-06,
+      "loss": 0.1853,
+      "step": 20403
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.034069043976812e-06,
+      "loss": 0.1816,
+      "step": 20404
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.03321160791055e-06,
+      "loss": 0.1768,
+      "step": 20405
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.032354271358099e-06,
+      "loss": 0.2021,
+      "step": 20406
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0314970343317016e-06,
+      "loss": 0.1781,
+      "step": 20407
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0306398968436102e-06,
+      "loss": 0.193,
+      "step": 20408
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0297828589060627e-06,
+      "loss": 0.1856,
+      "step": 20409
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0289259205312992e-06,
+      "loss": 0.1765,
+      "step": 20410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0280690817315674e-06,
+      "loss": 0.1606,
+      "step": 20411
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.027212342519099e-06,
+      "loss": 0.203,
+      "step": 20412
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0263557029061376e-06,
+      "loss": 0.181,
+      "step": 20413
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.025499162904918e-06,
+      "loss": 0.1846,
+      "step": 20414
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0246427225276697e-06,
+      "loss": 0.17,
+      "step": 20415
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0237863817866363e-06,
+      "loss": 0.1773,
+      "step": 20416
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0229301406940393e-06,
+      "loss": 0.1907,
+      "step": 20417
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0220739992621153e-06,
+      "loss": 0.1737,
+      "step": 20418
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.02121795750309e-06,
+      "loss": 0.1826,
+      "step": 20419
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0203620154291957e-06,
+      "loss": 0.1913,
+      "step": 20420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0195061730526553e-06,
+      "loss": 0.1674,
+      "step": 20421
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.018650430385692e-06,
+      "loss": 0.1814,
+      "step": 20422
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0177947874405355e-06,
+      "loss": 0.1716,
+      "step": 20423
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.016939244229401e-06,
+      "loss": 0.1946,
+      "step": 20424
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0160838007645145e-06,
+      "loss": 0.1871,
+      "step": 20425
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0152284570580935e-06,
+      "loss": 0.1749,
+      "step": 20426
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0143732131223514e-06,
+      "loss": 0.1718,
+      "step": 20427
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.013518068969512e-06,
+      "loss": 0.1905,
+      "step": 20428
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.012663024611786e-06,
+      "loss": 0.169,
+      "step": 20429
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.011808080061387e-06,
+      "loss": 0.1886,
+      "step": 20430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.010953235330525e-06,
+      "loss": 0.1652,
+      "step": 20431
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0100984904314155e-06,
+      "loss": 0.1631,
+      "step": 20432
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.009243845376266e-06,
+      "loss": 0.1796,
+      "step": 20433
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008389300177279e-06,
+      "loss": 0.173,
+      "step": 20434
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.00753485484667e-06,
+      "loss": 0.182,
+      "step": 20435
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0066805093966357e-06,
+      "loss": 0.188,
+      "step": 20436
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0058262638393853e-06,
+      "loss": 0.1713,
+      "step": 20437
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0049721181871192e-06,
+      "loss": 0.1843,
+      "step": 20438
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.004118072452035e-06,
+      "loss": 0.1837,
+      "step": 20439
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0032641266463367e-06,
+      "loss": 0.1651,
+      "step": 20440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002410280782221e-06,
+      "loss": 0.1701,
+      "step": 20441
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0015565348718822e-06,
+      "loss": 0.1943,
+      "step": 20442
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.000702888927517e-06,
+      "loss": 0.1685,
+      "step": 20443
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999849342961314e-06,
+      "loss": 0.1907,
+      "step": 20444
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.998995896985473e-06,
+      "loss": 0.1763,
+      "step": 20445
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9981425510121774e-06,
+      "loss": 0.1814,
+      "step": 20446
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9972893050536233e-06,
+      "loss": 0.1908,
+      "step": 20447
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.996436159121996e-06,
+      "loss": 0.1711,
+      "step": 20448
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9955831132294767e-06,
+      "loss": 0.1921,
+      "step": 20449
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.994730167388259e-06,
+      "loss": 0.1573,
+      "step": 20450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9938773216105177e-06,
+      "loss": 0.1824,
+      "step": 20451
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.993024575908443e-06,
+      "loss": 0.1897,
+      "step": 20452
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.992171930294212e-06,
+      "loss": 0.1672,
+      "step": 20453
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9913193847800005e-06,
+      "loss": 0.1992,
+      "step": 20454
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9904669393779927e-06,
+      "loss": 0.1673,
+      "step": 20455
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9896145941003606e-06,
+      "loss": 0.1642,
+      "step": 20456
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.988762348959282e-06,
+      "loss": 0.2017,
+      "step": 20457
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.987910203966925e-06,
+      "loss": 0.1665,
+      "step": 20458
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9870581591354674e-06,
+      "loss": 0.1711,
+      "step": 20459
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.986206214477079e-06,
+      "loss": 0.1943,
+      "step": 20460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9853543700039244e-06,
+      "loss": 0.1793,
+      "step": 20461
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9845026257281774e-06,
+      "loss": 0.1857,
+      "step": 20462
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9836509816619997e-06,
+      "loss": 0.1825,
+      "step": 20463
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9827994378175605e-06,
+      "loss": 0.1843,
+      "step": 20464
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.981947994207021e-06,
+      "loss": 0.1934,
+      "step": 20465
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.98109665084254e-06,
+      "loss": 0.1697,
+      "step": 20466
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.980245407736284e-06,
+      "loss": 0.1735,
+      "step": 20467
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.979394264900407e-06,
+      "loss": 0.1867,
+      "step": 20468
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.978543222347076e-06,
+      "loss": 0.1891,
+      "step": 20469
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9776922800884335e-06,
+      "loss": 0.1655,
+      "step": 20470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9768414381366453e-06,
+      "loss": 0.1808,
+      "step": 20471
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9759906965038598e-06,
+      "loss": 0.1681,
+      "step": 20472
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.975140055202228e-06,
+      "loss": 0.1643,
+      "step": 20473
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9742895142439064e-06,
+      "loss": 0.1757,
+      "step": 20474
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97343907364104e-06,
+      "loss": 0.1807,
+      "step": 20475
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9725887334057735e-06,
+      "loss": 0.1815,
+      "step": 20476
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97173849355026e-06,
+      "loss": 0.1986,
+      "step": 20477
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9708883540866385e-06,
+      "loss": 0.1729,
+      "step": 20478
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9700383150270573e-06,
+      "loss": 0.1762,
+      "step": 20479
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.969188376383656e-06,
+      "loss": 0.1722,
+      "step": 20480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9683385381685725e-06,
+      "loss": 0.18,
+      "step": 20481
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.967488800393955e-06,
+      "loss": 0.1648,
+      "step": 20482
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9666391630719283e-06,
+      "loss": 0.1811,
+      "step": 20483
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.965789626214638e-06,
+      "loss": 0.1729,
+      "step": 20484
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9649401898342124e-06,
+      "loss": 0.1621,
+      "step": 20485
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9640908539427924e-06,
+      "loss": 0.1846,
+      "step": 20486
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963241618552506e-06,
+      "loss": 0.1838,
+      "step": 20487
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.96239248367548e-06,
+      "loss": 0.1995,
+      "step": 20488
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9615434493238515e-06,
+      "loss": 0.1628,
+      "step": 20489
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9606945155097398e-06,
+      "loss": 0.1868,
+      "step": 20490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9598456822452782e-06,
+      "loss": 0.1746,
+      "step": 20491
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9589969495425884e-06,
+      "loss": 0.1774,
+      "step": 20492
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9581483174137916e-06,
+      "loss": 0.1676,
+      "step": 20493
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.957299785871014e-06,
+      "loss": 0.1897,
+      "step": 20494
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.956451354926374e-06,
+      "loss": 0.1955,
+      "step": 20495
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.955603024591991e-06,
+      "loss": 0.1993,
+      "step": 20496
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.954754794879978e-06,
+      "loss": 0.1774,
+      "step": 20497
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9539066658024597e-06,
+      "loss": 0.1816,
+      "step": 20498
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9530586373715454e-06,
+      "loss": 0.1934,
+      "step": 20499
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9522107095993458e-06,
+      "loss": 0.194,
+      "step": 20500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9513628824979788e-06,
+      "loss": 0.1673,
+      "step": 20501
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9505151560795507e-06,
+      "loss": 0.176,
+      "step": 20502
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.949667530356174e-06,
+      "loss": 0.1925,
+      "step": 20503
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9488200053399542e-06,
+      "loss": 0.1651,
+      "step": 20504
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9479725810429936e-06,
+      "loss": 0.1797,
+      "step": 20505
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9471252574774044e-06,
+      "loss": 0.1824,
+      "step": 20506
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.946278034655283e-06,
+      "loss": 0.166,
+      "step": 20507
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9454309125887405e-06,
+      "loss": 0.1678,
+      "step": 20508
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.944583891289864e-06,
+      "loss": 0.1876,
+      "step": 20509
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.943736970770762e-06,
+      "loss": 0.2,
+      "step": 20510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.942890151043529e-06,
+      "loss": 0.2068,
+      "step": 20511
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9420434321202596e-06,
+      "loss": 0.1892,
+      "step": 20512
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.941196814013052e-06,
+      "loss": 0.1733,
+      "step": 20513
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9403502967339968e-06,
+      "loss": 0.169,
+      "step": 20514
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9395038802951826e-06,
+      "loss": 0.1958,
+      "step": 20515
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9386575647087067e-06,
+      "loss": 0.1938,
+      "step": 20516
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.937811349986651e-06,
+      "loss": 0.1797,
+      "step": 20517
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9369652361411107e-06,
+      "loss": 0.1868,
+      "step": 20518
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.936119223184165e-06,
+      "loss": 0.1743,
+      "step": 20519
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9352733111278985e-06,
+      "loss": 0.1793,
+      "step": 20520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9344274999844023e-06,
+      "loss": 0.1891,
+      "step": 20521
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.933581789765746e-06,
+      "loss": 0.1868,
+      "step": 20522
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9327361804840195e-06,
+      "loss": 0.1847,
+      "step": 20523
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.931890672151293e-06,
+      "loss": 0.1914,
+      "step": 20524
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9310452647796526e-06,
+      "loss": 0.173,
+      "step": 20525
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9301999583811703e-06,
+      "loss": 0.1894,
+      "step": 20526
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9293547529679156e-06,
+      "loss": 0.1832,
+      "step": 20527
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9285096485519706e-06,
+      "loss": 0.1868,
+      "step": 20528
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.927664645145398e-06,
+      "loss": 0.1732,
+      "step": 20529
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9268197427602764e-06,
+      "loss": 0.1795,
+      "step": 20530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.925974941408669e-06,
+      "loss": 0.1762,
+      "step": 20531
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.925130241102642e-06,
+      "loss": 0.1824,
+      "step": 20532
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9242856418542653e-06,
+      "loss": 0.1707,
+      "step": 20533
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9234411436755982e-06,
+      "loss": 0.1825,
+      "step": 20534
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9225967465787133e-06,
+      "loss": 0.1701,
+      "step": 20535
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.921752450575659e-06,
+      "loss": 0.1636,
+      "step": 20536
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.920908255678503e-06,
+      "loss": 0.1827,
+      "step": 20537
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.920064161899303e-06,
+      "loss": 0.1938,
+      "step": 20538
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.919220169250112e-06,
+      "loss": 0.1821,
+      "step": 20539
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9183762777429924e-06,
+      "loss": 0.1937,
+      "step": 20540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.917532487389991e-06,
+      "loss": 0.1754,
+      "step": 20541
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.916688798203168e-06,
+      "loss": 0.1835,
+      "step": 20542
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.915845210194571e-06,
+      "loss": 0.1666,
+      "step": 20543
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.915001723376246e-06,
+      "loss": 0.1686,
+      "step": 20544
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.914158337760249e-06,
+      "loss": 0.1817,
+      "step": 20545
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.913315053358623e-06,
+      "loss": 0.1694,
+      "step": 20546
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.912471870183411e-06,
+      "loss": 0.1584,
+      "step": 20547
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9116287882466653e-06,
+      "loss": 0.174,
+      "step": 20548
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.910785807560417e-06,
+      "loss": 0.1963,
+      "step": 20549
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9099429281367155e-06,
+      "loss": 0.1755,
+      "step": 20550
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9091001499875958e-06,
+      "loss": 0.1664,
+      "step": 20551
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9082574731251025e-06,
+      "loss": 0.1816,
+      "step": 20552
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.907414897561267e-06,
+      "loss": 0.1939,
+      "step": 20553
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9065724233081238e-06,
+      "loss": 0.1858,
+      "step": 20554
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.905730050377712e-06,
+      "loss": 0.1816,
+      "step": 20555
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9048877787820586e-06,
+      "loss": 0.1795,
+      "step": 20556
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9040456085332004e-06,
+      "loss": 0.172,
+      "step": 20557
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9032035396431634e-06,
+      "loss": 0.1523,
+      "step": 20558
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.902361572123974e-06,
+      "loss": 0.1784,
+      "step": 20559
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.901519705987663e-06,
+      "loss": 0.1808,
+      "step": 20560
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.900677941246255e-06,
+      "loss": 0.171,
+      "step": 20561
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8998362779117716e-06,
+      "loss": 0.1663,
+      "step": 20562
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.898994715996234e-06,
+      "loss": 0.1773,
+      "step": 20563
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8981532555116676e-06,
+      "loss": 0.1774,
+      "step": 20564
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.89731189647009e-06,
+      "loss": 0.1825,
+      "step": 20565
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.896470638883514e-06,
+      "loss": 0.1988,
+      "step": 20566
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.895629482763965e-06,
+      "loss": 0.1948,
+      "step": 20567
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8947884281234506e-06,
+      "loss": 0.1695,
+      "step": 20568
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8939474749739917e-06,
+      "loss": 0.1763,
+      "step": 20569
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8931066233275962e-06,
+      "loss": 0.1913,
+      "step": 20570
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8922658731962715e-06,
+      "loss": 0.1833,
+      "step": 20571
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8914252245920337e-06,
+      "loss": 0.1637,
+      "step": 20572
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8905846775268843e-06,
+      "loss": 0.174,
+      "step": 20573
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8897442320128397e-06,
+      "loss": 0.1856,
+      "step": 20574
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8889038880618915e-06,
+      "loss": 0.1856,
+      "step": 20575
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.888063645686053e-06,
+      "loss": 0.1902,
+      "step": 20576
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8872235048973228e-06,
+      "loss": 0.1815,
+      "step": 20577
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8863834657076995e-06,
+      "loss": 0.1902,
+      "step": 20578
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8855435281291866e-06,
+      "loss": 0.1824,
+      "step": 20579
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.884703692173777e-06,
+      "loss": 0.1746,
+      "step": 20580
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.883863957853472e-06,
+      "loss": 0.1797,
+      "step": 20581
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.883024325180265e-06,
+      "loss": 0.2015,
+      "step": 20582
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.882184794166144e-06,
+      "loss": 0.1859,
+      "step": 20583
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8813453648231094e-06,
+      "loss": 0.1885,
+      "step": 20584
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.880506037163148e-06,
+      "loss": 0.1826,
+      "step": 20585
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.879666811198244e-06,
+      "loss": 0.212,
+      "step": 20586
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.878827686940395e-06,
+      "loss": 0.1543,
+      "step": 20587
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.877988664401575e-06,
+      "loss": 0.1654,
+      "step": 20588
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.877149743593779e-06,
+      "loss": 0.1894,
+      "step": 20589
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8763109245289832e-06,
+      "loss": 0.1989,
+      "step": 20590
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8754722072191743e-06,
+      "loss": 0.1769,
+      "step": 20591
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8746335916763304e-06,
+      "loss": 0.192,
+      "step": 20592
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.873795077912428e-06,
+      "loss": 0.1772,
+      "step": 20593
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8729566659394494e-06,
+      "loss": 0.1665,
+      "step": 20594
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8721183557693645e-06,
+      "loss": 0.1988,
+      "step": 20595
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.871280147414154e-06,
+      "loss": 0.1769,
+      "step": 20596
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.870442040885789e-06,
+      "loss": 0.1674,
+      "step": 20597
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.869604036196236e-06,
+      "loss": 0.1682,
+      "step": 20598
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8687661333574723e-06,
+      "loss": 0.1749,
+      "step": 20599
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8679283323814642e-06,
+      "loss": 0.183,
+      "step": 20600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.867090633280177e-06,
+      "loss": 0.1807,
+      "step": 20601
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.866253036065575e-06,
+      "loss": 0.172,
+      "step": 20602
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8654155407496277e-06,
+      "loss": 0.1915,
+      "step": 20603
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8645781473442957e-06,
+      "loss": 0.1857,
+      "step": 20604
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8637408558615366e-06,
+      "loss": 0.181,
+      "step": 20605
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.862903666313316e-06,
+      "loss": 0.1994,
+      "step": 20606
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8620665787115866e-06,
+      "loss": 0.1719,
+      "step": 20607
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.861229593068312e-06,
+      "loss": 0.1559,
+      "step": 20608
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8603927093954444e-06,
+      "loss": 0.1836,
+      "step": 20609
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8595559277049347e-06,
+      "loss": 0.1865,
+      "step": 20610
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8587192480087424e-06,
+      "loss": 0.1893,
+      "step": 20611
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.85788267031881e-06,
+      "loss": 0.1913,
+      "step": 20612
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8570461946470963e-06,
+      "loss": 0.1744,
+      "step": 20613
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8562098210055458e-06,
+      "loss": 0.1923,
+      "step": 20614
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8553735494061043e-06,
+      "loss": 0.1662,
+      "step": 20615
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.854537379860717e-06,
+      "loss": 0.1687,
+      "step": 20616
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.853701312381326e-06,
+      "loss": 0.1677,
+      "step": 20617
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8528653469798783e-06,
+      "loss": 0.1948,
+      "step": 20618
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.852029483668314e-06,
+      "loss": 0.1695,
+      "step": 20619
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8511937224585673e-06,
+      "loss": 0.1974,
+      "step": 20620
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8503580633625838e-06,
+      "loss": 0.1725,
+      "step": 20621
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.849522506392293e-06,
+      "loss": 0.1825,
+      "step": 20622
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8486870515596364e-06,
+      "loss": 0.1803,
+      "step": 20623
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.847851698876546e-06,
+      "loss": 0.1794,
+      "step": 20624
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.847016448354948e-06,
+      "loss": 0.1796,
+      "step": 20625
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8461813000067817e-06,
+      "loss": 0.1615,
+      "step": 20626
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8453462538439723e-06,
+      "loss": 0.163,
+      "step": 20627
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8445113098784484e-06,
+      "loss": 0.1828,
+      "step": 20628
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.843676468122133e-06,
+      "loss": 0.1953,
+      "step": 20629
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.842841728586956e-06,
+      "loss": 0.1606,
+      "step": 20630
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8420070912848395e-06,
+      "loss": 0.1662,
+      "step": 20631
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8411725562277015e-06,
+      "loss": 0.1624,
+      "step": 20632
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8403381234274684e-06,
+      "loss": 0.1831,
+      "step": 20633
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8395037928960544e-06,
+      "loss": 0.177,
+      "step": 20634
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.838669564645381e-06,
+      "loss": 0.1872,
+      "step": 20635
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.837835438687364e-06,
+      "loss": 0.1861,
+      "step": 20636
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.837001415033913e-06,
+      "loss": 0.1713,
+      "step": 20637
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8361674936969474e-06,
+      "loss": 0.1772,
+      "step": 20638
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.835333674688374e-06,
+      "loss": 0.1651,
+      "step": 20639
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8344999580201126e-06,
+      "loss": 0.2054,
+      "step": 20640
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.833666343704058e-06,
+      "loss": 0.1607,
+      "step": 20641
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.832832831752128e-06,
+      "loss": 0.1908,
+      "step": 20642
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8319994221762248e-06,
+      "loss": 0.1753,
+      "step": 20643
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8311661149882496e-06,
+      "loss": 0.1964,
+      "step": 20644
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.830332910200112e-06,
+      "loss": 0.1822,
+      "step": 20645
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.829499807823708e-06,
+      "loss": 0.1624,
+      "step": 20646
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8286668078709435e-06,
+      "loss": 0.1773,
+      "step": 20647
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.827833910353713e-06,
+      "loss": 0.1752,
+      "step": 20648
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8270011152839116e-06,
+      "loss": 0.1789,
+      "step": 20649
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.826168422673441e-06,
+      "loss": 0.1881,
+      "step": 20650
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8253358325341917e-06,
+      "loss": 0.1692,
+      "step": 20651
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8245033448780535e-06,
+      "loss": 0.1744,
+      "step": 20652
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.823670959716928e-06,
+      "loss": 0.1787,
+      "step": 20653
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8228386770626914e-06,
+      "loss": 0.1591,
+      "step": 20654
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.822006496927242e-06,
+      "loss": 0.1892,
+      "step": 20655
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8211744193224598e-06,
+      "loss": 0.1684,
+      "step": 20656
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.820342444260238e-06,
+      "loss": 0.1786,
+      "step": 20657
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819510571752455e-06,
+      "loss": 0.1495,
+      "step": 20658
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8186788018109924e-06,
+      "loss": 0.1797,
+      "step": 20659
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8178471344477366e-06,
+      "loss": 0.1671,
+      "step": 20660
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8170155696745604e-06,
+      "loss": 0.1694,
+      "step": 20661
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.816184107503349e-06,
+      "loss": 0.1725,
+      "step": 20662
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8153527479459753e-06,
+      "loss": 0.182,
+      "step": 20663
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8145214910143128e-06,
+      "loss": 0.1723,
+      "step": 20664
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8136903367202384e-06,
+      "loss": 0.16,
+      "step": 20665
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.812859285075624e-06,
+      "loss": 0.1749,
+      "step": 20666
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8120283360923395e-06,
+      "loss": 0.1819,
+      "step": 20667
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8111974897822514e-06,
+      "loss": 0.1636,
+      "step": 20668
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8103667461572327e-06,
+      "loss": 0.1526,
+      "step": 20669
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.809536105229147e-06,
+      "loss": 0.1919,
+      "step": 20670
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.808705567009857e-06,
+      "loss": 0.1694,
+      "step": 20671
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8078751315112318e-06,
+      "loss": 0.1897,
+      "step": 20672
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8070447987451265e-06,
+      "loss": 0.1913,
+      "step": 20673
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8062145687234087e-06,
+      "loss": 0.1763,
+      "step": 20674
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8053844414579344e-06,
+      "loss": 0.1861,
+      "step": 20675
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.804554416960558e-06,
+      "loss": 0.1711,
+      "step": 20676
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8037244952431408e-06,
+      "loss": 0.1818,
+      "step": 20677
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8028946763175324e-06,
+      "loss": 0.195,
+      "step": 20678
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.802064960195594e-06,
+      "loss": 0.1821,
+      "step": 20679
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8012353468891674e-06,
+      "loss": 0.1753,
+      "step": 20680
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8004058364101083e-06,
+      "loss": 0.182,
+      "step": 20681
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7995764287702654e-06,
+      "loss": 0.1918,
+      "step": 20682
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7987471239814824e-06,
+      "loss": 0.1828,
+      "step": 20683
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7979179220556106e-06,
+      "loss": 0.1695,
+      "step": 20684
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.797088823004488e-06,
+      "loss": 0.1661,
+      "step": 20685
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7962598268399644e-06,
+      "loss": 0.1824,
+      "step": 20686
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7954309335738773e-06,
+      "loss": 0.1853,
+      "step": 20687
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.794602143218065e-06,
+      "loss": 0.2031,
+      "step": 20688
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.79377345578437e-06,
+      "loss": 0.1989,
+      "step": 20689
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7929448712846287e-06,
+      "loss": 0.1948,
+      "step": 20690
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7921163897306713e-06,
+      "loss": 0.1859,
+      "step": 20691
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7912880111343387e-06,
+      "loss": 0.1931,
+      "step": 20692
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.790459735507461e-06,
+      "loss": 0.1795,
+      "step": 20693
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.78963156286187e-06,
+      "loss": 0.1993,
+      "step": 20694
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.78880349320939e-06,
+      "loss": 0.1779,
+      "step": 20695
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7879755265618558e-06,
+      "loss": 0.1665,
+      "step": 20696
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.787147662931093e-06,
+      "loss": 0.1843,
+      "step": 20697
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7863199023289223e-06,
+      "loss": 0.1704,
+      "step": 20698
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785492244767174e-06,
+      "loss": 0.1723,
+      "step": 20699
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7846646902576634e-06,
+      "loss": 0.1819,
+      "step": 20700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.783837238812219e-06,
+      "loss": 0.1544,
+      "step": 20701
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.783009890442656e-06,
+      "loss": 0.1935,
+      "step": 20702
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.782182645160789e-06,
+      "loss": 0.1886,
+      "step": 20703
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.781355502978441e-06,
+      "loss": 0.1803,
+      "step": 20704
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7805284639074204e-06,
+      "loss": 0.1988,
+      "step": 20705
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7797015279595507e-06,
+      "loss": 0.171,
+      "step": 20706
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7788746951466307e-06,
+      "loss": 0.1967,
+      "step": 20707
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7780479654804813e-06,
+      "loss": 0.1808,
+      "step": 20708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7772213389729062e-06,
+      "loss": 0.1793,
+      "step": 20709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7763948156357124e-06,
+      "loss": 0.1709,
+      "step": 20710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.775568395480711e-06,
+      "loss": 0.1857,
+      "step": 20711
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.774742078519701e-06,
+      "loss": 0.1696,
+      "step": 20712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.77391586476449e-06,
+      "loss": 0.1693,
+      "step": 20713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7730897542268787e-06,
+      "loss": 0.1693,
+      "step": 20714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7722637469186632e-06,
+      "loss": 0.1621,
+      "step": 20715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7714378428516477e-06,
+      "loss": 0.1848,
+      "step": 20716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.770612042037625e-06,
+      "loss": 0.1671,
+      "step": 20717
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7697863444883955e-06,
+      "loss": 0.1622,
+      "step": 20718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7689607502157514e-06,
+      "loss": 0.174,
+      "step": 20719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7681352592314834e-06,
+      "loss": 0.1801,
+      "step": 20720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7673098715473867e-06,
+      "loss": 0.1743,
+      "step": 20721
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7664845871752454e-06,
+      "loss": 0.1872,
+      "step": 20722
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7656594061268537e-06,
+      "loss": 0.1767,
+      "step": 20723
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7648343284139977e-06,
+      "loss": 0.1743,
+      "step": 20724
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.764009354048457e-06,
+      "loss": 0.1913,
+      "step": 20725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.763184483042024e-06,
+      "loss": 0.1774,
+      "step": 20726
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.762359715406473e-06,
+      "loss": 0.1923,
+      "step": 20727
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7615350511535933e-06,
+      "loss": 0.178,
+      "step": 20728
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.76071049029516e-06,
+      "loss": 0.1493,
+      "step": 20729
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7598860328429487e-06,
+      "loss": 0.1762,
+      "step": 20730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.759061678808741e-06,
+      "loss": 0.1644,
+      "step": 20731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7582374282043102e-06,
+      "loss": 0.169,
+      "step": 20732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.757413281041429e-06,
+      "loss": 0.1802,
+      "step": 20733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.756589237331868e-06,
+      "loss": 0.1785,
+      "step": 20734
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7557652970874017e-06,
+      "loss": 0.178,
+      "step": 20735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.754941460319798e-06,
+      "loss": 0.1707,
+      "step": 20736
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7541177270408215e-06,
+      "loss": 0.191,
+      "step": 20737
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7532940972622436e-06,
+      "loss": 0.1852,
+      "step": 20738
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.752470570995822e-06,
+      "loss": 0.1747,
+      "step": 20739
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7516471482533293e-06,
+      "loss": 0.1696,
+      "step": 20740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7508238290465215e-06,
+      "loss": 0.1804,
+      "step": 20741
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.750000613387157e-06,
+      "loss": 0.1851,
+      "step": 20742
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7491775012870013e-06,
+      "loss": 0.1752,
+      "step": 20743
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.748354492757803e-06,
+      "loss": 0.1672,
+      "step": 20744
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.747531587811331e-06,
+      "loss": 0.1915,
+      "step": 20745
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7467087864593246e-06,
+      "loss": 0.2054,
+      "step": 20746
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7458860887135462e-06,
+      "loss": 0.1705,
+      "step": 20747
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.745063494585746e-06,
+      "loss": 0.1907,
+      "step": 20748
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7442410040876688e-06,
+      "loss": 0.1682,
+      "step": 20749
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7434186172310697e-06,
+      "loss": 0.1787,
+      "step": 20750
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7425963340276906e-06,
+      "loss": 0.1719,
+      "step": 20751
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7417741544892816e-06,
+      "loss": 0.1734,
+      "step": 20752
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7409520786275855e-06,
+      "loss": 0.184,
+      "step": 20753
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.740130106454341e-06,
+      "loss": 0.1784,
+      "step": 20754
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.739308237981294e-06,
+      "loss": 0.1896,
+      "step": 20755
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.738486473220182e-06,
+      "loss": 0.1868,
+      "step": 20756
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.737664812182742e-06,
+      "loss": 0.1902,
+      "step": 20757
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7368432548807167e-06,
+      "loss": 0.176,
+      "step": 20758
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7360218013258298e-06,
+      "loss": 0.1861,
+      "step": 20759
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.735200451529826e-06,
+      "loss": 0.1589,
+      "step": 20760
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.734379205504429e-06,
+      "loss": 0.1893,
+      "step": 20761
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.733558063261378e-06,
+      "loss": 0.1768,
+      "step": 20762
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7327370248123962e-06,
+      "loss": 0.1651,
+      "step": 20763
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.731916090169211e-06,
+      "loss": 0.1609,
+      "step": 20764
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7310952593435534e-06,
+      "loss": 0.1725,
+      "step": 20765
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7302745323471423e-06,
+      "loss": 0.19,
+      "step": 20766
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.729453909191707e-06,
+      "loss": 0.1884,
+      "step": 20767
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.728633389888966e-06,
+      "loss": 0.1646,
+      "step": 20768
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.727812974450638e-06,
+      "loss": 0.1858,
+      "step": 20769
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.726992662888446e-06,
+      "loss": 0.1764,
+      "step": 20770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7261724552141024e-06,
+      "loss": 0.174,
+      "step": 20771
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.725352351439332e-06,
+      "loss": 0.1808,
+      "step": 20772
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.724532351575837e-06,
+      "loss": 0.169,
+      "step": 20773
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.723712455635339e-06,
+      "loss": 0.1627,
+      "step": 20774
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.722892663629547e-06,
+      "loss": 0.1716,
+      "step": 20775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.722072975570168e-06,
+      "loss": 0.177,
+      "step": 20776
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7212533914689154e-06,
+      "loss": 0.1879,
+      "step": 20777
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720433911337491e-06,
+      "loss": 0.1842,
+      "step": 20778
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.719614535187607e-06,
+      "loss": 0.1752,
+      "step": 20779
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7187952630309633e-06,
+      "loss": 0.1944,
+      "step": 20780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.71797609487926e-06,
+      "loss": 0.1598,
+      "step": 20781
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.717157030744203e-06,
+      "loss": 0.1815,
+      "step": 20782
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7163380706374876e-06,
+      "loss": 0.1726,
+      "step": 20783
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715519214570815e-06,
+      "loss": 0.1956,
+      "step": 20784
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7147004625558825e-06,
+      "loss": 0.1425,
+      "step": 20785
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7138818146043833e-06,
+      "loss": 0.1974,
+      "step": 20786
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7130632707280103e-06,
+      "loss": 0.1779,
+      "step": 20787
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7122448309384543e-06,
+      "loss": 0.17,
+      "step": 20788
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.71142649524741e-06,
+      "loss": 0.1803,
+      "step": 20789
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.710608263666561e-06,
+      "loss": 0.1895,
+      "step": 20790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7097901362076028e-06,
+      "loss": 0.1689,
+      "step": 20791
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7089721128822164e-06,
+      "loss": 0.1769,
+      "step": 20792
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7081541937020848e-06,
+      "loss": 0.1898,
+      "step": 20793
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7073363786788964e-06,
+      "loss": 0.1912,
+      "step": 20794
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7065186678243294e-06,
+      "loss": 0.1889,
+      "step": 20795
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.705701061150062e-06,
+      "loss": 0.1849,
+      "step": 20796
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704883558667779e-06,
+      "loss": 0.1819,
+      "step": 20797
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704066160389155e-06,
+      "loss": 0.1794,
+      "step": 20798
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.703248866325864e-06,
+      "loss": 0.1705,
+      "step": 20799
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7024316764895786e-06,
+      "loss": 0.1733,
+      "step": 20800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7016145908919777e-06,
+      "loss": 0.1916,
+      "step": 20801
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7007976095447287e-06,
+      "loss": 0.201,
+      "step": 20802
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6999807324594984e-06,
+      "loss": 0.1725,
+      "step": 20803
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6991639596479626e-06,
+      "loss": 0.1786,
+      "step": 20804
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6983472911217802e-06,
+      "loss": 0.1953,
+      "step": 20805
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.697530726892623e-06,
+      "loss": 0.1728,
+      "step": 20806
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6967142669721522e-06,
+      "loss": 0.1566,
+      "step": 20807
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6958979113720275e-06,
+      "loss": 0.1977,
+      "step": 20808
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6950816601039155e-06,
+      "loss": 0.1553,
+      "step": 20809
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6942655131794693e-06,
+      "loss": 0.1824,
+      "step": 20810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6934494706103554e-06,
+      "loss": 0.1758,
+      "step": 20811
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6926335324082186e-06,
+      "loss": 0.1695,
+      "step": 20812
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.691817698584723e-06,
+      "loss": 0.1616,
+      "step": 20813
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6910019691515198e-06,
+      "loss": 0.1832,
+      "step": 20814
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.690186344120256e-06,
+      "loss": 0.181,
+      "step": 20815
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6893708235025893e-06,
+      "loss": 0.1829,
+      "step": 20816
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6885554073101615e-06,
+      "loss": 0.1943,
+      "step": 20817
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6877400955546274e-06,
+      "loss": 0.1704,
+      "step": 20818
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6869248882476294e-06,
+      "loss": 0.1751,
+      "step": 20819
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.686109785400809e-06,
+      "loss": 0.1647,
+      "step": 20820
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.685294787025814e-06,
+      "loss": 0.1693,
+      "step": 20821
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.684479893134282e-06,
+      "loss": 0.2003,
+      "step": 20822
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.683665103737857e-06,
+      "loss": 0.1911,
+      "step": 20823
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.682850418848176e-06,
+      "loss": 0.1866,
+      "step": 20824
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6820358384768754e-06,
+      "loss": 0.1779,
+      "step": 20825
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.68122136263559e-06,
+      "loss": 0.1882,
+      "step": 20826
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.680406991335952e-06,
+      "loss": 0.1817,
+      "step": 20827
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.679592724589598e-06,
+      "loss": 0.1862,
+      "step": 20828
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6787785624081585e-06,
+      "loss": 0.1815,
+      "step": 20829
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6779645048032587e-06,
+      "loss": 0.179,
+      "step": 20830
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6771505517865326e-06,
+      "loss": 0.1516,
+      "step": 20831
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6763367033696007e-06,
+      "loss": 0.1788,
+      "step": 20832
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.675522959564095e-06,
+      "loss": 0.1881,
+      "step": 20833
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6747093203816344e-06,
+      "loss": 0.195,
+      "step": 20834
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.673895785833839e-06,
+      "loss": 0.1922,
+      "step": 20835
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.673082355932335e-06,
+      "loss": 0.1845,
+      "step": 20836
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.672269030688739e-06,
+      "loss": 0.1854,
+      "step": 20837
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.671455810114668e-06,
+      "loss": 0.1745,
+      "step": 20838
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6706426942217354e-06,
+      "loss": 0.1815,
+      "step": 20839
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669829683021561e-06,
+      "loss": 0.1983,
+      "step": 20840
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669016776525756e-06,
+      "loss": 0.1812,
+      "step": 20841
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.668203974745929e-06,
+      "loss": 0.1841,
+      "step": 20842
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.667391277693694e-06,
+      "loss": 0.1705,
+      "step": 20843
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6665786853806563e-06,
+      "loss": 0.1784,
+      "step": 20844
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.665766197818428e-06,
+      "loss": 0.2011,
+      "step": 20845
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6649538150186103e-06,
+      "loss": 0.1953,
+      "step": 20846
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.664141536992807e-06,
+      "loss": 0.1853,
+      "step": 20847
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.663329363752624e-06,
+      "loss": 0.1834,
+      "step": 20848
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.662517295309658e-06,
+      "loss": 0.1769,
+      "step": 20849
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.661705331675515e-06,
+      "loss": 0.1782,
+      "step": 20850
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6608934728617885e-06,
+      "loss": 0.1718,
+      "step": 20851
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6600817188800754e-06,
+      "loss": 0.1914,
+      "step": 20852
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6592700697419728e-06,
+      "loss": 0.2005,
+      "step": 20853
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.658458525459069e-06,
+      "loss": 0.1659,
+      "step": 20854
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6576470860429637e-06,
+      "loss": 0.1832,
+      "step": 20855
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6568357515052402e-06,
+      "loss": 0.1904,
+      "step": 20856
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.656024521857494e-06,
+      "loss": 0.1875,
+      "step": 20857
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6552133971113114e-06,
+      "loss": 0.1916,
+      "step": 20858
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6544023772782736e-06,
+      "loss": 0.1915,
+      "step": 20859
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.653591462369971e-06,
+      "loss": 0.2,
+      "step": 20860
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6527806523979847e-06,
+      "loss": 0.1833,
+      "step": 20861
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6519699473738936e-06,
+      "loss": 0.1916,
+      "step": 20862
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6511593473092832e-06,
+      "loss": 0.1889,
+      "step": 20863
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6503488522157296e-06,
+      "loss": 0.1622,
+      "step": 20864
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.649538462104809e-06,
+      "loss": 0.1733,
+      "step": 20865
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6487281769880947e-06,
+      "loss": 0.2057,
+      "step": 20866
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.647917996877167e-06,
+      "loss": 0.1728,
+      "step": 20867
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.647107921783596e-06,
+      "loss": 0.1758,
+      "step": 20868
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.646297951718948e-06,
+      "loss": 0.1672,
+      "step": 20869
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.645488086694801e-06,
+      "loss": 0.1869,
+      "step": 20870
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.644678326722715e-06,
+      "loss": 0.1754,
+      "step": 20871
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.643868671814265e-06,
+      "loss": 0.1948,
+      "step": 20872
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.64305912198101e-06,
+      "loss": 0.1869,
+      "step": 20873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642249677234513e-06,
+      "loss": 0.1927,
+      "step": 20874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641440337586342e-06,
+      "loss": 0.1929,
+      "step": 20875
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.64063110304805e-06,
+      "loss": 0.187,
+      "step": 20876
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.639821973631208e-06,
+      "loss": 0.1961,
+      "step": 20877
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6390129493473584e-06,
+      "loss": 0.1919,
+      "step": 20878
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.638204030208068e-06,
+      "loss": 0.1857,
+      "step": 20879
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6373952162248893e-06,
+      "loss": 0.1727,
+      "step": 20880
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.63658650740937e-06,
+      "loss": 0.1657,
+      "step": 20881
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6357779037730703e-06,
+      "loss": 0.1735,
+      "step": 20882
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.634969405327532e-06,
+      "loss": 0.201,
+      "step": 20883
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.634161012084311e-06,
+      "loss": 0.1697,
+      "step": 20884
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6333527240549518e-06,
+      "loss": 0.1761,
+      "step": 20885
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6325445412509965e-06,
+      "loss": 0.183,
+      "step": 20886
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6317364636839947e-06,
+      "loss": 0.1857,
+      "step": 20887
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.630928491365483e-06,
+      "loss": 0.2026,
+      "step": 20888
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.63012062430701e-06,
+      "loss": 0.1846,
+      "step": 20889
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6293128625201102e-06,
+      "loss": 0.1765,
+      "step": 20890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.628505206016324e-06,
+      "loss": 0.1907,
+      "step": 20891
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.627697654807185e-06,
+      "loss": 0.1804,
+      "step": 20892
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6268902089042283e-06,
+      "loss": 0.185,
+      "step": 20893
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.626082868318992e-06,
+      "loss": 0.1748,
+      "step": 20894
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6252756330630024e-06,
+      "loss": 0.1752,
+      "step": 20895
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6244685031477956e-06,
+      "loss": 0.1976,
+      "step": 20896
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.623661478584899e-06,
+      "loss": 0.1858,
+      "step": 20897
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6228545593858357e-06,
+      "loss": 0.179,
+      "step": 20898
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6220477455621384e-06,
+      "loss": 0.1864,
+      "step": 20899
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6212410371253284e-06,
+      "loss": 0.1778,
+      "step": 20900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.620434434086926e-06,
+      "loss": 0.1789,
+      "step": 20901
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6196279364584587e-06,
+      "loss": 0.1833,
+      "step": 20902
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.618821544251443e-06,
+      "loss": 0.1878,
+      "step": 20903
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6180152574773987e-06,
+      "loss": 0.1718,
+      "step": 20904
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6172090761478386e-06,
+      "loss": 0.2087,
+      "step": 20905
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6164030002742857e-06,
+      "loss": 0.1756,
+      "step": 20906
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6155970298682487e-06,
+      "loss": 0.1807,
+      "step": 20907
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6147911649412396e-06,
+      "loss": 0.1907,
+      "step": 20908
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.613985405504773e-06,
+      "loss": 0.1745,
+      "step": 20909
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.613179751570354e-06,
+      "loss": 0.178,
+      "step": 20910
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6123742031494957e-06,
+      "loss": 0.1725,
+      "step": 20911
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.611568760253703e-06,
+      "loss": 0.1919,
+      "step": 20912
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6107634228944767e-06,
+      "loss": 0.1691,
+      "step": 20913
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.609958191083326e-06,
+      "loss": 0.1837,
+      "step": 20914
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6091530648317477e-06,
+      "loss": 0.173,
+      "step": 20915
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.608348044151251e-06,
+      "loss": 0.1748,
+      "step": 20916
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6075431290533225e-06,
+      "loss": 0.177,
+      "step": 20917
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6067383195494687e-06,
+      "loss": 0.1819,
+      "step": 20918
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.605933615651183e-06,
+      "loss": 0.1773,
+      "step": 20919
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.605129017369956e-06,
+      "loss": 0.1902,
+      "step": 20920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6043245247172876e-06,
+      "loss": 0.1964,
+      "step": 20921
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6035201377046626e-06,
+      "loss": 0.1682,
+      "step": 20922
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.602715856343576e-06,
+      "loss": 0.2007,
+      "step": 20923
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.601911680645516e-06,
+      "loss": 0.1779,
+      "step": 20924
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6011076106219634e-06,
+      "loss": 0.1701,
+      "step": 20925
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.600303646284411e-06,
+      "loss": 0.1719,
+      "step": 20926
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5994997876443363e-06,
+      "loss": 0.1829,
+      "step": 20927
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5986960347132284e-06,
+      "loss": 0.1796,
+      "step": 20928
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5978923875025643e-06,
+      "loss": 0.1905,
+      "step": 20929
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.597088846023823e-06,
+      "loss": 0.1648,
+      "step": 20930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5962854102884827e-06,
+      "loss": 0.1624,
+      "step": 20931
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5954820803080174e-06,
+      "loss": 0.1731,
+      "step": 20932
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5946788560939074e-06,
+      "loss": 0.2049,
+      "step": 20933
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.593875737657623e-06,
+      "loss": 0.1763,
+      "step": 20934
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5930727250106326e-06,
+      "loss": 0.1729,
+      "step": 20935
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.592269818164412e-06,
+      "loss": 0.193,
+      "step": 20936
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.591467017130426e-06,
+      "loss": 0.1673,
+      "step": 20937
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5906643219201455e-06,
+      "loss": 0.1769,
+      "step": 20938
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.589861732545035e-06,
+      "loss": 0.1979,
+      "step": 20939
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5890592490165546e-06,
+      "loss": 0.169,
+      "step": 20940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5882568713461733e-06,
+      "loss": 0.176,
+      "step": 20941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.587454599545347e-06,
+      "loss": 0.1762,
+      "step": 20942
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5866524336255426e-06,
+      "loss": 0.1684,
+      "step": 20943
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5858503735982077e-06,
+      "loss": 0.1917,
+      "step": 20944
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5850484194748084e-06,
+      "loss": 0.17,
+      "step": 20945
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5842465712667953e-06,
+      "loss": 0.2121,
+      "step": 20946
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5834448289856195e-06,
+      "loss": 0.1931,
+      "step": 20947
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5826431926427396e-06,
+      "loss": 0.1928,
+      "step": 20948
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5818416622496e-06,
+      "loss": 0.1909,
+      "step": 20949
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.581040237817656e-06,
+      "loss": 0.1836,
+      "step": 20950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5802389193583522e-06,
+      "loss": 0.1679,
+      "step": 20951
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.57943770688313e-06,
+      "loss": 0.1879,
+      "step": 20952
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5786366004034426e-06,
+      "loss": 0.1684,
+      "step": 20953
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.577835599930726e-06,
+      "loss": 0.1647,
+      "step": 20954
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5770347054764264e-06,
+      "loss": 0.1763,
+      "step": 20955
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5762339170519825e-06,
+      "loss": 0.1581,
+      "step": 20956
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5754332346688325e-06,
+      "loss": 0.1912,
+      "step": 20957
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.574632658338413e-06,
+      "loss": 0.1952,
+      "step": 20958
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.573832188072155e-06,
+      "loss": 0.1639,
+      "step": 20959
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.573031823881501e-06,
+      "loss": 0.1799,
+      "step": 20960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5722315657778775e-06,
+      "loss": 0.1708,
+      "step": 20961
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.571431413772718e-06,
+      "loss": 0.184,
+      "step": 20962
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5706313678774533e-06,
+      "loss": 0.1784,
+      "step": 20963
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5698314281035053e-06,
+      "loss": 0.1775,
+      "step": 20964
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5690315944623067e-06,
+      "loss": 0.1757,
+      "step": 20965
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.568231866965277e-06,
+      "loss": 0.1669,
+      "step": 20966
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5674322456238454e-06,
+      "loss": 0.1657,
+      "step": 20967
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.566632730449431e-06,
+      "loss": 0.1679,
+      "step": 20968
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5658333214534537e-06,
+      "loss": 0.1873,
+      "step": 20969
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.565034018647332e-06,
+      "loss": 0.1725,
+      "step": 20970
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.564234822042482e-06,
+      "loss": 0.1805,
+      "step": 20971
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5634357316503223e-06,
+      "loss": 0.1804,
+      "step": 20972
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5626367474822665e-06,
+      "loss": 0.2023,
+      "step": 20973
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5618378695497238e-06,
+      "loss": 0.1796,
+      "step": 20974
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.56103909786411e-06,
+      "loss": 0.1625,
+      "step": 20975
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.560240432436831e-06,
+      "loss": 0.1733,
+      "step": 20976
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5594418732793004e-06,
+      "loss": 0.188,
+      "step": 20977
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5586434204029208e-06,
+      "loss": 0.1691,
+      "step": 20978
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.557845073819095e-06,
+      "loss": 0.1792,
+      "step": 20979
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.557046833539233e-06,
+      "loss": 0.1539,
+      "step": 20980
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.556248699574729e-06,
+      "loss": 0.1872,
+      "step": 20981
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5554506719369953e-06,
+      "loss": 0.1739,
+      "step": 20982
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5546527506374175e-06,
+      "loss": 0.182,
+      "step": 20983
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.553854935687402e-06,
+      "loss": 0.1672,
+      "step": 20984
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5530572270983423e-06,
+      "loss": 0.1732,
+      "step": 20985
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.55225962488163e-06,
+      "loss": 0.1844,
+      "step": 20986
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551462129048663e-06,
+      "loss": 0.1717,
+      "step": 20987
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5506647396108273e-06,
+      "loss": 0.1962,
+      "step": 20988
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.54986745657952e-06,
+      "loss": 0.1815,
+      "step": 20989
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.549070279966125e-06,
+      "loss": 0.1668,
+      "step": 20990
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5482732097820273e-06,
+      "loss": 0.164,
+      "step": 20991
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5474762460386173e-06,
+      "loss": 0.166,
+      "step": 20992
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.546679388747273e-06,
+      "loss": 0.1879,
+      "step": 20993
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5458826379193836e-06,
+      "loss": 0.1691,
+      "step": 20994
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5450859935663266e-06,
+      "loss": 0.1818,
+      "step": 20995
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.54428945569948e-06,
+      "loss": 0.171,
+      "step": 20996
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5434930243302225e-06,
+      "loss": 0.1742,
+      "step": 20997
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5426966994699286e-06,
+      "loss": 0.1884,
+      "step": 20998
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5419004811299773e-06,
+      "loss": 0.1755,
+      "step": 20999
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5411043693217363e-06,
+      "loss": 0.1815,
+      "step": 21000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.540308364056583e-06,
+      "loss": 0.1754,
+      "step": 21001
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5395124653458857e-06,
+      "loss": 0.1809,
+      "step": 21002
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5387166732010083e-06,
+      "loss": 0.186,
+      "step": 21003
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5379209876333256e-06,
+      "loss": 0.1968,
+      "step": 21004
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5371254086541996e-06,
+      "loss": 0.1622,
+      "step": 21005
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5363299362749904e-06,
+      "loss": 0.1917,
+      "step": 21006
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.535534570507069e-06,
+      "loss": 0.1866,
+      "step": 21007
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.534739311361787e-06,
+      "loss": 0.1706,
+      "step": 21008
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5339441588505154e-06,
+      "loss": 0.1835,
+      "step": 21009
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5331491129845998e-06,
+      "loss": 0.1738,
+      "step": 21010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5323541737754055e-06,
+      "loss": 0.1825,
+      "step": 21011
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.531559341234284e-06,
+      "loss": 0.1801,
+      "step": 21012
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.530764615372586e-06,
+      "loss": 0.1664,
+      "step": 21013
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5299699962016687e-06,
+      "loss": 0.1821,
+      "step": 21014
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5291754837328787e-06,
+      "loss": 0.1543,
+      "step": 21015
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5283810779775684e-06,
+      "loss": 0.16,
+      "step": 21016
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527586778947083e-06,
+      "loss": 0.1769,
+      "step": 21017
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.526792586652765e-06,
+      "loss": 0.1869,
+      "step": 21018
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5259985011059664e-06,
+      "loss": 0.1792,
+      "step": 21019
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.525204522318021e-06,
+      "loss": 0.1733,
+      "step": 21020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5244106503002785e-06,
+      "loss": 0.1815,
+      "step": 21021
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5236168850640742e-06,
+      "loss": 0.1678,
+      "step": 21022
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5228232266207464e-06,
+      "loss": 0.1632,
+      "step": 21023
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.522029674981632e-06,
+      "loss": 0.1814,
+      "step": 21024
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.521236230158064e-06,
+      "loss": 0.1857,
+      "step": 21025
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.52044289216138e-06,
+      "loss": 0.1615,
+      "step": 21026
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5196496610029087e-06,
+      "loss": 0.1644,
+      "step": 21027
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.518856536693983e-06,
+      "loss": 0.1774,
+      "step": 21028
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5180635192459335e-06,
+      "loss": 0.1763,
+      "step": 21029
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5172706086700803e-06,
+      "loss": 0.1611,
+      "step": 21030
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5164778049777584e-06,
+      "loss": 0.1682,
+      "step": 21031
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5156851081802847e-06,
+      "loss": 0.1623,
+      "step": 21032
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.166,
+      "step": 21033
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514100035315189e-06,
+      "loss": 0.191,
+      "step": 21034
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5133076592702044e-06,
+      "loss": 0.1596,
+      "step": 21035
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5125153901653533e-06,
+      "loss": 0.1817,
+      "step": 21036
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.511723228011951e-06,
+      "loss": 0.177,
+      "step": 21037
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5109311728213183e-06,
+      "loss": 0.1906,
+      "step": 21038
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.510139224604764e-06,
+      "loss": 0.1652,
+      "step": 21039
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5093473833736003e-06,
+      "loss": 0.1711,
+      "step": 21040
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5085556491391416e-06,
+      "loss": 0.1853,
+      "step": 21041
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.507764021912692e-06,
+      "loss": 0.2012,
+      "step": 21042
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5069725017055644e-06,
+      "loss": 0.1776,
+      "step": 21043
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.506181088529064e-06,
+      "loss": 0.1735,
+      "step": 21044
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5053897823944897e-06,
+      "loss": 0.1702,
+      "step": 21045
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5045985833131512e-06,
+      "loss": 0.1908,
+      "step": 21046
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.503807491296345e-06,
+      "loss": 0.1867,
+      "step": 21047
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5030165063553803e-06,
+      "loss": 0.1747,
+      "step": 21048
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502225628501541e-06,
+      "loss": 0.1577,
+      "step": 21049
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5014348577461365e-06,
+      "loss": 0.1924,
+      "step": 21050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5006441941004567e-06,
+      "loss": 0.1688,
+      "step": 21051
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.499853637575794e-06,
+      "loss": 0.1656,
+      "step": 21052
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.499063188183446e-06,
+      "loss": 0.1804,
+      "step": 21053
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4982728459346974e-06,
+      "loss": 0.167,
+      "step": 21054
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4974826108408424e-06,
+      "loss": 0.1937,
+      "step": 21055
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496692482913168e-06,
+      "loss": 0.1907,
+      "step": 21056
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.495902462162957e-06,
+      "loss": 0.1799,
+      "step": 21057
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4951125486014983e-06,
+      "loss": 0.1715,
+      "step": 21058
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.494322742240071e-06,
+      "loss": 0.1925,
+      "step": 21059
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.493533043089961e-06,
+      "loss": 0.1963,
+      "step": 21060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.492743451162446e-06,
+      "loss": 0.1692,
+      "step": 21061
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4919539664688066e-06,
+      "loss": 0.1809,
+      "step": 21062
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4911645890203162e-06,
+      "loss": 0.1756,
+      "step": 21063
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.49037531882825e-06,
+      "loss": 0.1637,
+      "step": 21064
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4895861559038868e-06,
+      "loss": 0.1648,
+      "step": 21065
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4887971002584944e-06,
+      "loss": 0.1814,
+      "step": 21066
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4880081519033473e-06,
+      "loss": 0.1973,
+      "step": 21067
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4872193108497134e-06,
+      "loss": 0.1739,
+      "step": 21068
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.486430577108858e-06,
+      "loss": 0.1914,
+      "step": 21069
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4856419506920525e-06,
+      "loss": 0.1762,
+      "step": 21070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4848534316105556e-06,
+      "loss": 0.1891,
+      "step": 21071
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4840650198756367e-06,
+      "loss": 0.1917,
+      "step": 21072
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.483276715498555e-06,
+      "loss": 0.1744,
+      "step": 21073
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4824885184905668e-06,
+      "loss": 0.1729,
+      "step": 21074
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.481700428862941e-06,
+      "loss": 0.1918,
+      "step": 21075
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4809124466269207e-06,
+      "loss": 0.1681,
+      "step": 21076
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.480124571793773e-06,
+      "loss": 0.1785,
+      "step": 21077
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4793368043747467e-06,
+      "loss": 0.1733,
+      "step": 21078
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478549144381093e-06,
+      "loss": 0.1822,
+      "step": 21079
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.477761591824066e-06,
+      "loss": 0.1757,
+      "step": 21080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4769741467149134e-06,
+      "loss": 0.1937,
+      "step": 21081
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4761868090648857e-06,
+      "loss": 0.176,
+      "step": 21082
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.475399578885228e-06,
+      "loss": 0.1895,
+      "step": 21083
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.474612456187181e-06,
+      "loss": 0.1854,
+      "step": 21084
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4738254409819953e-06,
+      "loss": 0.1658,
+      "step": 21085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4730385332809047e-06,
+      "loss": 0.1894,
+      "step": 21086
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4722517330951567e-06,
+      "loss": 0.1891,
+      "step": 21087
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4714650404359875e-06,
+      "loss": 0.2072,
+      "step": 21088
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4706784553146336e-06,
+      "loss": 0.1826,
+      "step": 21089
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4698919777423302e-06,
+      "loss": 0.1683,
+      "step": 21090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4691056077303087e-06,
+      "loss": 0.1892,
+      "step": 21091
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.468319345289808e-06,
+      "loss": 0.1904,
+      "step": 21092
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4675331904320533e-06,
+      "loss": 0.1855,
+      "step": 21093
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.466747143168279e-06,
+      "loss": 0.1899,
+      "step": 21094
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4659612035097103e-06,
+      "loss": 0.1717,
+      "step": 21095
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.465175371467571e-06,
+      "loss": 0.1866,
+      "step": 21096
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4643896470530915e-06,
+      "loss": 0.1881,
+      "step": 21097
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.46360403027749e-06,
+      "loss": 0.2138,
+      "step": 21098
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.462818521151993e-06,
+      "loss": 0.1988,
+      "step": 21099
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.462033119687819e-06,
+      "loss": 0.1759,
+      "step": 21100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461247825896186e-06,
+      "loss": 0.2018,
+      "step": 21101
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4604626397883112e-06,
+      "loss": 0.1788,
+      "step": 21102
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4596775613754077e-06,
+      "loss": 0.2014,
+      "step": 21103
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4588925906686943e-06,
+      "loss": 0.1745,
+      "step": 21104
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4581077276793797e-06,
+      "loss": 0.1898,
+      "step": 21105
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4573229724186787e-06,
+      "loss": 0.17,
+      "step": 21106
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.456538324897798e-06,
+      "loss": 0.1914,
+      "step": 21107
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4557537851279446e-06,
+      "loss": 0.1816,
+      "step": 21108
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454969353120329e-06,
+      "loss": 0.1703,
+      "step": 21109
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4541850288861536e-06,
+      "loss": 0.1835,
+      "step": 21110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.453400812436618e-06,
+      "loss": 0.1731,
+      "step": 21111
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4526167037829318e-06,
+      "loss": 0.1759,
+      "step": 21112
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4518327029362876e-06,
+      "loss": 0.1636,
+      "step": 21113
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4510488099078934e-06,
+      "loss": 0.1533,
+      "step": 21114
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4502650247089355e-06,
+      "loss": 0.1796,
+      "step": 21115
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4494813473506164e-06,
+      "loss": 0.184,
+      "step": 21116
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.448697777844128e-06,
+      "loss": 0.1925,
+      "step": 21117
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4479143162006614e-06,
+      "loss": 0.1884,
+      "step": 21118
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.447130962431411e-06,
+      "loss": 0.1838,
+      "step": 21119
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4463477165475624e-06,
+      "loss": 0.1871,
+      "step": 21120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.445564578560309e-06,
+      "loss": 0.1618,
+      "step": 21121
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4447815484808333e-06,
+      "loss": 0.1788,
+      "step": 21122
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4439986263203175e-06,
+      "loss": 0.1623,
+      "step": 21123
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4432158120899506e-06,
+      "loss": 0.1749,
+      "step": 21124
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4424331058009097e-06,
+      "loss": 0.1765,
+      "step": 21125
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4416505074643803e-06,
+      "loss": 0.1706,
+      "step": 21126
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4408680170915364e-06,
+      "loss": 0.1926,
+      "step": 21127
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4400856346935585e-06,
+      "loss": 0.1614,
+      "step": 21128
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4393033602816185e-06,
+      "loss": 0.1621,
+      "step": 21129
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4385211938668905e-06,
+      "loss": 0.1811,
+      "step": 21130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4377391354605506e-06,
+      "loss": 0.1782,
+      "step": 21131
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.436957185073766e-06,
+      "loss": 0.1695,
+      "step": 21132
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4361753427177103e-06,
+      "loss": 0.1786,
+      "step": 21133
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.435393608403549e-06,
+      "loss": 0.1807,
+      "step": 21134
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.434611982142445e-06,
+      "loss": 0.1671,
+      "step": 21135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.43383046394557e-06,
+      "loss": 0.1788,
+      "step": 21136
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4330490538240803e-06,
+      "loss": 0.1762,
+      "step": 21137
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4322677517891433e-06,
+      "loss": 0.1874,
+      "step": 21138
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4314865578519175e-06,
+      "loss": 0.1922,
+      "step": 21139
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4307054720235613e-06,
+      "loss": 0.1696,
+      "step": 21140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4299244943152313e-06,
+      "loss": 0.1701,
+      "step": 21141
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4291436247380795e-06,
+      "loss": 0.1814,
+      "step": 21142
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.428362863303266e-06,
+      "loss": 0.1578,
+      "step": 21143
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4275822100219406e-06,
+      "loss": 0.1629,
+      "step": 21144
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4268016649052515e-06,
+      "loss": 0.1749,
+      "step": 21145
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4260212279643535e-06,
+      "loss": 0.1713,
+      "step": 21146
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4252408992103893e-06,
+      "loss": 0.172,
+      "step": 21147
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424460678654509e-06,
+      "loss": 0.2035,
+      "step": 21148
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4236805663078565e-06,
+      "loss": 0.177,
+      "step": 21149
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4229005621815715e-06,
+      "loss": 0.1705,
+      "step": 21150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4221206662868e-06,
+      "loss": 0.1677,
+      "step": 21151
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4213408786346783e-06,
+      "loss": 0.1658,
+      "step": 21152
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.42056119923635e-06,
+      "loss": 0.1742,
+      "step": 21153
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4197816281029483e-06,
+      "loss": 0.1387,
+      "step": 21154
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.419002165245609e-06,
+      "loss": 0.1622,
+      "step": 21155
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4182228106754668e-06,
+      "loss": 0.1736,
+      "step": 21156
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.417443564403651e-06,
+      "loss": 0.1836,
+      "step": 21157
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4166644264412965e-06,
+      "loss": 0.1638,
+      "step": 21158
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415885396799529e-06,
+      "loss": 0.1766,
+      "step": 21159
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4151064754894803e-06,
+      "loss": 0.1741,
+      "step": 21160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4143276625222743e-06,
+      "loss": 0.1852,
+      "step": 21161
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.413548957909033e-06,
+      "loss": 0.1846,
+      "step": 21162
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4127703616608856e-06,
+      "loss": 0.2034,
+      "step": 21163
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.411991873788946e-06,
+      "loss": 0.1817,
+      "step": 21164
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.411213494304341e-06,
+      "loss": 0.1535,
+      "step": 21165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.410435223218187e-06,
+      "loss": 0.1728,
+      "step": 21166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4096570605415993e-06,
+      "loss": 0.1994,
+      "step": 21167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.408879006285694e-06,
+      "loss": 0.1767,
+      "step": 21168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4081010604615816e-06,
+      "loss": 0.1647,
+      "step": 21169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.407323223080381e-06,
+      "loss": 0.166,
+      "step": 21170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4065454941531963e-06,
+      "loss": 0.1754,
+      "step": 21171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4057678736911418e-06,
+      "loss": 0.1905,
+      "step": 21172
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.404990361705324e-06,
+      "loss": 0.1724,
+      "step": 21173
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.404212958206844e-06,
+      "loss": 0.1843,
+      "step": 21174
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4034356632068133e-06,
+      "loss": 0.1772,
+      "step": 21175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4026584767163286e-06,
+      "loss": 0.1975,
+      "step": 21176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4018813987464964e-06,
+      "loss": 0.1895,
+      "step": 21177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4011044293084153e-06,
+      "loss": 0.1806,
+      "step": 21178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4003275684131786e-06,
+      "loss": 0.2063,
+      "step": 21179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.399550816071893e-06,
+      "loss": 0.1899,
+      "step": 21180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3987741722956413e-06,
+      "loss": 0.1926,
+      "step": 21181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.397997637095527e-06,
+      "loss": 0.1933,
+      "step": 21182
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3972212104826387e-06,
+      "loss": 0.163,
+      "step": 21183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.396444892468064e-06,
+      "loss": 0.1609,
+      "step": 21184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.395668683062897e-06,
+      "loss": 0.1737,
+      "step": 21185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.39489258227822e-06,
+      "loss": 0.1757,
+      "step": 21186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.394116590125124e-06,
+      "loss": 0.1615,
+      "step": 21187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.393340706614692e-06,
+      "loss": 0.1915,
+      "step": 21188
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3925649317580014e-06,
+      "loss": 0.1728,
+      "step": 21189
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.391789265566141e-06,
+      "loss": 0.1623,
+      "step": 21190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.391013708050184e-06,
+      "loss": 0.1838,
+      "step": 21191
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3902382592212136e-06,
+      "loss": 0.1787,
+      "step": 21192
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.389462919090305e-06,
+      "loss": 0.186,
+      "step": 21193
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3886876876685326e-06,
+      "loss": 0.1617,
+      "step": 21194
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3879125649669697e-06,
+      "loss": 0.1737,
+      "step": 21195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3871375509966855e-06,
+      "loss": 0.1827,
+      "step": 21196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.386362645768755e-06,
+      "loss": 0.1771,
+      "step": 21197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.385587849294242e-06,
+      "loss": 0.1678,
+      "step": 21198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3848131615842196e-06,
+      "loss": 0.1727,
+      "step": 21199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3840385826497504e-06,
+      "loss": 0.1673,
+      "step": 21200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.383264112501896e-06,
+      "loss": 0.1901,
+      "step": 21201
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.382489751151724e-06,
+      "loss": 0.1793,
+      "step": 21202
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3817154986102907e-06,
+      "loss": 0.1728,
+      "step": 21203
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3809413548886606e-06,
+      "loss": 0.1767,
+      "step": 21204
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.380167319997888e-06,
+      "loss": 0.1919,
+      "step": 21205
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379393393949031e-06,
+      "loss": 0.1656,
+      "step": 21206
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.378619576753143e-06,
+      "loss": 0.1646,
+      "step": 21207
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3778458684212747e-06,
+      "loss": 0.1708,
+      "step": 21208
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3770722689644832e-06,
+      "loss": 0.1603,
+      "step": 21209
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.376298778393814e-06,
+      "loss": 0.1811,
+      "step": 21210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3755253967203195e-06,
+      "loss": 0.1802,
+      "step": 21211
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3747521239550464e-06,
+      "loss": 0.1756,
+      "step": 21212
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3739789601090347e-06,
+      "loss": 0.1588,
+      "step": 21213
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3732059051933367e-06,
+      "loss": 0.1779,
+      "step": 21214
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3724329592189887e-06,
+      "loss": 0.1888,
+      "step": 21215
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3716601221970305e-06,
+      "loss": 0.1866,
+      "step": 21216
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.370887394138508e-06,
+      "loss": 0.1848,
+      "step": 21217
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3701147750544506e-06,
+      "loss": 0.1933,
+      "step": 21218
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.369342264955905e-06,
+      "loss": 0.1656,
+      "step": 21219
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3685698638538945e-06,
+      "loss": 0.1807,
+      "step": 21220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.367797571759458e-06,
+      "loss": 0.178,
+      "step": 21221
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.367025388683628e-06,
+      "loss": 0.1792,
+      "step": 21222
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366253314637428e-06,
+      "loss": 0.1814,
+      "step": 21223
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.365481349631894e-06,
+      "loss": 0.1647,
+      "step": 21224
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.364709493678047e-06,
+      "loss": 0.1746,
+      "step": 21225
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.363937746786917e-06,
+      "loss": 0.1674,
+      "step": 21226
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3631661089695257e-06,
+      "loss": 0.1751,
+      "step": 21227
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.362394580236892e-06,
+      "loss": 0.1743,
+      "step": 21228
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3616231606000417e-06,
+      "loss": 0.1606,
+      "step": 21229
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3608518500699884e-06,
+      "loss": 0.1666,
+      "step": 21230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3600806486577556e-06,
+      "loss": 0.1793,
+      "step": 21231
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3593095563743563e-06,
+      "loss": 0.18,
+      "step": 21232
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.358538573230804e-06,
+      "loss": 0.1782,
+      "step": 21233
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3577676992381115e-06,
+      "loss": 0.1796,
+      "step": 21234
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3569969344072883e-06,
+      "loss": 0.173,
+      "step": 21235
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3562262787493495e-06,
+      "loss": 0.182,
+      "step": 21236
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.355455732275296e-06,
+      "loss": 0.1943,
+      "step": 21237
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.354685294996142e-06,
+      "loss": 0.1997,
+      "step": 21238
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3539149669228876e-06,
+      "loss": 0.1735,
+      "step": 21239
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.353144748066535e-06,
+      "loss": 0.1868,
+      "step": 21240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3523746384380908e-06,
+      "loss": 0.1678,
+      "step": 21241
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3516046380485504e-06,
+      "loss": 0.1966,
+      "step": 21242
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3508347469089164e-06,
+      "loss": 0.1721,
+      "step": 21243
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3500649650301853e-06,
+      "loss": 0.1805,
+      "step": 21244
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349295292423349e-06,
+      "loss": 0.1674,
+      "step": 21245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34852572909941e-06,
+      "loss": 0.1709,
+      "step": 21246
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3477562750693484e-06,
+      "loss": 0.1912,
+      "step": 21247
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346986930344166e-06,
+      "loss": 0.1846,
+      "step": 21248
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346217694934847e-06,
+      "loss": 0.1785,
+      "step": 21249
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3454485688523777e-06,
+      "loss": 0.1818,
+      "step": 21250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3446795521077482e-06,
+      "loss": 0.1694,
+      "step": 21251
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.34391064471194e-06,
+      "loss": 0.2007,
+      "step": 21252
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3431418466759404e-06,
+      "loss": 0.1773,
+      "step": 21253
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3423731580107277e-06,
+      "loss": 0.1607,
+      "step": 21254
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.341604578727281e-06,
+      "loss": 0.1834,
+      "step": 21255
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.340836108836583e-06,
+      "loss": 0.1912,
+      "step": 21256
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3400677483496037e-06,
+      "loss": 0.1796,
+      "step": 21257
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3392994972773266e-06,
+      "loss": 0.1697,
+      "step": 21258
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.33853135563072e-06,
+      "loss": 0.1776,
+      "step": 21259
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3377633234207585e-06,
+      "loss": 0.1786,
+      "step": 21260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3369954006584106e-06,
+      "loss": 0.1868,
+      "step": 21261
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3362275873546438e-06,
+      "loss": 0.1683,
+      "step": 21262
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.335459883520431e-06,
+      "loss": 0.1805,
+      "step": 21263
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3346922891667322e-06,
+      "loss": 0.1828,
+      "step": 21264
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333924804304518e-06,
+      "loss": 0.165,
+      "step": 21265
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333157428944748e-06,
+      "loss": 0.1979,
+      "step": 21266
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3323901630983794e-06,
+      "loss": 0.1847,
+      "step": 21267
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331623006776379e-06,
+      "loss": 0.1825,
+      "step": 21268
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3308559599896996e-06,
+      "loss": 0.1874,
+      "step": 21269
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.330089022749301e-06,
+      "loss": 0.1594,
+      "step": 21270
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.329322195066138e-06,
+      "loss": 0.1853,
+      "step": 21271
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3285554769511632e-06,
+      "loss": 0.1836,
+      "step": 21272
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.327788868415327e-06,
+      "loss": 0.1849,
+      "step": 21273
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3270223694695795e-06,
+      "loss": 0.1639,
+      "step": 21274
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3262559801248718e-06,
+      "loss": 0.1859,
+      "step": 21275
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3254897003921474e-06,
+      "loss": 0.1933,
+      "step": 21276
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.324723530282358e-06,
+      "loss": 0.1666,
+      "step": 21277
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3239574698064426e-06,
+      "loss": 0.1852,
+      "step": 21278
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3231915189753428e-06,
+      "loss": 0.1582,
+      "step": 21279
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3224256778000055e-06,
+      "loss": 0.1645,
+      "step": 21280
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3216599462913623e-06,
+      "loss": 0.1746,
+      "step": 21281
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3208943244603576e-06,
+      "loss": 0.1664,
+      "step": 21282
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3201288123179244e-06,
+      "loss": 0.1775,
+      "step": 21283
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.319363409874995e-06,
+      "loss": 0.17,
+      "step": 21284
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.318598117142512e-06,
+      "loss": 0.1666,
+      "step": 21285
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3178329341313933e-06,
+      "loss": 0.1789,
+      "step": 21286
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3170678608525777e-06,
+      "loss": 0.1926,
+      "step": 21287
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.316302897316992e-06,
+      "loss": 0.1798,
+      "step": 21288
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.315538043535559e-06,
+      "loss": 0.1691,
+      "step": 21289
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3147732995192097e-06,
+      "loss": 0.1659,
+      "step": 21290
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.314008665278863e-06,
+      "loss": 0.1799,
+      "step": 21291
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.313244140825445e-06,
+      "loss": 0.1667,
+      "step": 21292
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3124797261698752e-06,
+      "loss": 0.1818,
+      "step": 21293
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3117154213230674e-06,
+      "loss": 0.1748,
+      "step": 21294
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3109512262959477e-06,
+      "loss": 0.192,
+      "step": 21295
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3101871410994236e-06,
+      "loss": 0.1682,
+      "step": 21296
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3094231657444157e-06,
+      "loss": 0.1813,
+      "step": 21297
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3086593002418344e-06,
+      "loss": 0.1982,
+      "step": 21298
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.307895544602591e-06,
+      "loss": 0.1935,
+      "step": 21299
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3071318988375934e-06,
+      "loss": 0.168,
+      "step": 21300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306368362957747e-06,
+      "loss": 0.1605,
+      "step": 21301
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3056049369739665e-06,
+      "loss": 0.1714,
+      "step": 21302
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3048416208971482e-06,
+      "loss": 0.1828,
+      "step": 21303
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.304078414738201e-06,
+      "loss": 0.1927,
+      "step": 21304
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.303315318508026e-06,
+      "loss": 0.1885,
+      "step": 21305
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.302552332217518e-06,
+      "loss": 0.1583,
+      "step": 21306
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.301789455877582e-06,
+      "loss": 0.186,
+      "step": 21307
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3010266894991106e-06,
+      "loss": 0.159,
+      "step": 21308
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3002640330930026e-06,
+      "loss": 0.166,
+      "step": 21309
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2995014866701503e-06,
+      "loss": 0.1716,
+      "step": 21310
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2987390502414443e-06,
+      "loss": 0.1826,
+      "step": 21311
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297976723817781e-06,
+      "loss": 0.1865,
+      "step": 21312
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2972145074100393e-06,
+      "loss": 0.1602,
+      "step": 21313
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2964524010291167e-06,
+      "loss": 0.1733,
+      "step": 21314
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.295690404685892e-06,
+      "loss": 0.1742,
+      "step": 21315
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2949285183912553e-06,
+      "loss": 0.1701,
+      "step": 21316
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2941667421560864e-06,
+      "loss": 0.1755,
+      "step": 21317
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2934050759912643e-06,
+      "loss": 0.1858,
+      "step": 21318
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2926435199076736e-06,
+      "loss": 0.1715,
+      "step": 21319
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2918820739161908e-06,
+      "loss": 0.187,
+      "step": 21320
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.291120738027688e-06,
+      "loss": 0.1788,
+      "step": 21321
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2903595122530475e-06,
+      "loss": 0.1713,
+      "step": 21322
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2895983966031355e-06,
+      "loss": 0.1824,
+      "step": 21323
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2888373910888307e-06,
+      "loss": 0.1752,
+      "step": 21324
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.288076495721e-06,
+      "loss": 0.1741,
+      "step": 21325
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2873157105105115e-06,
+      "loss": 0.1859,
+      "step": 21326
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.286555035468233e-06,
+      "loss": 0.1672,
+      "step": 21327
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.285794470605026e-06,
+      "loss": 0.1783,
+      "step": 21328
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.285034015931762e-06,
+      "loss": 0.1853,
+      "step": 21329
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.284273671459296e-06,
+      "loss": 0.1699,
+      "step": 21330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2835134371984956e-06,
+      "loss": 0.1989,
+      "step": 21331
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2827533131602175e-06,
+      "loss": 0.1929,
+      "step": 21332
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281993299355315e-06,
+      "loss": 0.1712,
+      "step": 21333
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2812333957946507e-06,
+      "loss": 0.1735,
+      "step": 21334
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2804736024890728e-06,
+      "loss": 0.1936,
+      "step": 21335
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.279713919449441e-06,
+      "loss": 0.1666,
+      "step": 21336
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2789543466866027e-06,
+      "loss": 0.1889,
+      "step": 21337
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2781948842114077e-06,
+      "loss": 0.1654,
+      "step": 21338
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.277435532034705e-06,
+      "loss": 0.1805,
+      "step": 21339
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2766762901673376e-06,
+      "loss": 0.1862,
+      "step": 21340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275917158620158e-06,
+      "loss": 0.1766,
+      "step": 21341
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275158137404001e-06,
+      "loss": 0.1872,
+      "step": 21342
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2743992265297168e-06,
+      "loss": 0.1865,
+      "step": 21343
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2736404260081413e-06,
+      "loss": 0.1969,
+      "step": 21344
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.27288173585011e-06,
+      "loss": 0.1725,
+      "step": 21345
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272123156066468e-06,
+      "loss": 0.1712,
+      "step": 21346
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2713646866680426e-06,
+      "loss": 0.188,
+      "step": 21347
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2706063276656752e-06,
+      "loss": 0.1689,
+      "step": 21348
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2698480790701948e-06,
+      "loss": 0.1908,
+      "step": 21349
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2690899408924305e-06,
+      "loss": 0.2028,
+      "step": 21350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.268331913143218e-06,
+      "loss": 0.1539,
+      "step": 21351
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.267573995833374e-06,
+      "loss": 0.1916,
+      "step": 21352
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.266816188973735e-06,
+      "loss": 0.1954,
+      "step": 21353
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.266058492575122e-06,
+      "loss": 0.1566,
+      "step": 21354
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.265300906648353e-06,
+      "loss": 0.1713,
+      "step": 21355
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264543431204258e-06,
+      "loss": 0.1741,
+      "step": 21356
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2637860662536494e-06,
+      "loss": 0.1786,
+      "step": 21357
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2630288118073505e-06,
+      "loss": 0.2144,
+      "step": 21358
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.262271667876177e-06,
+      "loss": 0.1752,
+      "step": 21359
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2615146344709404e-06,
+      "loss": 0.1841,
+      "step": 21360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2607577116024594e-06,
+      "loss": 0.1637,
+      "step": 21361
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26000089928154e-06,
+      "loss": 0.1822,
+      "step": 21362
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2592441975190004e-06,
+      "loss": 0.1823,
+      "step": 21363
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2584876063256433e-06,
+      "loss": 0.1934,
+      "step": 21364
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.257731125712278e-06,
+      "loss": 0.1644,
+      "step": 21365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2569747556897103e-06,
+      "loss": 0.1805,
+      "step": 21366
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2562184962687393e-06,
+      "loss": 0.1639,
+      "step": 21367
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2554623474601754e-06,
+      "loss": 0.1806,
+      "step": 21368
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2547063092748123e-06,
+      "loss": 0.1849,
+      "step": 21369
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2539503817234553e-06,
+      "loss": 0.1731,
+      "step": 21370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2531945648168994e-06,
+      "loss": 0.1614,
+      "step": 21371
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2524388585659374e-06,
+      "loss": 0.1965,
+      "step": 21372
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.251683262981371e-06,
+      "loss": 0.1975,
+      "step": 21373
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.250927778073986e-06,
+      "loss": 0.1802,
+      "step": 21374
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.250172403854579e-06,
+      "loss": 0.1709,
+      "step": 21375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2494171403339393e-06,
+      "loss": 0.159,
+      "step": 21376
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.248661987522853e-06,
+      "loss": 0.1968,
+      "step": 21377
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.247906945432107e-06,
+      "loss": 0.193,
+      "step": 21378
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2471520140724845e-06,
+      "loss": 0.1751,
+      "step": 21379
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2463971934547733e-06,
+      "loss": 0.1816,
+      "step": 21380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2456424835897507e-06,
+      "loss": 0.1749,
+      "step": 21381
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2448878844882027e-06,
+      "loss": 0.1814,
+      "step": 21382
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2441333961609048e-06,
+      "loss": 0.1822,
+      "step": 21383
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.24337901861863e-06,
+      "loss": 0.1679,
+      "step": 21384
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242624751872161e-06,
+      "loss": 0.2045,
+      "step": 21385
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.241870595932267e-06,
+      "loss": 0.1634,
+      "step": 21386
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2411165508097242e-06,
+      "loss": 0.1786,
+      "step": 21387
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2403626165153015e-06,
+      "loss": 0.1674,
+      "step": 21388
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2396087930597655e-06,
+      "loss": 0.1878,
+      "step": 21389
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2388550804538877e-06,
+      "loss": 0.1578,
+      "step": 21390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2381014787084345e-06,
+      "loss": 0.1719,
+      "step": 21391
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2373479878341676e-06,
+      "loss": 0.1669,
+      "step": 21392
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.236594607841851e-06,
+      "loss": 0.1741,
+      "step": 21393
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2358413387422428e-06,
+      "loss": 0.1581,
+      "step": 21394
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2350881805461087e-06,
+      "loss": 0.1721,
+      "step": 21395
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2343351332642017e-06,
+      "loss": 0.1604,
+      "step": 21396
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2335821969072823e-06,
+      "loss": 0.182,
+      "step": 21397
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.232829371486105e-06,
+      "loss": 0.1773,
+      "step": 21398
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.232076657011418e-06,
+      "loss": 0.1617,
+      "step": 21399
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2313240534939808e-06,
+      "loss": 0.1813,
+      "step": 21400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2305715609445354e-06,
+      "loss": 0.1885,
+      "step": 21401
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.229819179373839e-06,
+      "loss": 0.1815,
+      "step": 21402
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.229066908792634e-06,
+      "loss": 0.1858,
+      "step": 21403
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2283147492116663e-06,
+      "loss": 0.1782,
+      "step": 21404
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22756270064168e-06,
+      "loss": 0.1661,
+      "step": 21405
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2268107630934144e-06,
+      "loss": 0.1948,
+      "step": 21406
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226058936577616e-06,
+      "loss": 0.176,
+      "step": 21407
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2253072211050185e-06,
+      "loss": 0.186,
+      "step": 21408
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2245556166863637e-06,
+      "loss": 0.1832,
+      "step": 21409
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.223804123332386e-06,
+      "loss": 0.1898,
+      "step": 21410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.223052741053816e-06,
+      "loss": 0.1778,
+      "step": 21411
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2223014698613944e-06,
+      "loss": 0.1891,
+      "step": 21412
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2215503097658432e-06,
+      "loss": 0.1886,
+      "step": 21413
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2207992607779005e-06,
+      "loss": 0.202,
+      "step": 21414
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2200483229082905e-06,
+      "loss": 0.162,
+      "step": 21415
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.219297496167736e-06,
+      "loss": 0.1849,
+      "step": 21416
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2185467805669716e-06,
+      "loss": 0.1774,
+      "step": 21417
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2177961761167087e-06,
+      "loss": 0.1739,
+      "step": 21418
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2170456828276774e-06,
+      "loss": 0.1849,
+      "step": 21419
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2162953007105926e-06,
+      "loss": 0.187,
+      "step": 21420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2155450297761773e-06,
+      "loss": 0.1763,
+      "step": 21421
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.214794870035146e-06,
+      "loss": 0.1997,
+      "step": 21422
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2140448214982122e-06,
+      "loss": 0.1859,
+      "step": 21423
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.213294884176095e-06,
+      "loss": 0.1764,
+      "step": 21424
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2125450580795014e-06,
+      "loss": 0.1798,
+      "step": 21425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2117953432191417e-06,
+      "loss": 0.1677,
+      "step": 21426
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2110457396057307e-06,
+      "loss": 0.1874,
+      "step": 21427
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2102962472499678e-06,
+      "loss": 0.1753,
+      "step": 21428
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2095468661625674e-06,
+      "loss": 0.1799,
+      "step": 21429
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2087975963542276e-06,
+      "loss": 0.1767,
+      "step": 21430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2080484378356528e-06,
+      "loss": 0.1829,
+      "step": 21431
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2072993906175443e-06,
+      "loss": 0.1789,
+      "step": 21432
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2065504547105986e-06,
+      "loss": 0.1839,
+      "step": 21433
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.205801630125518e-06,
+      "loss": 0.1779,
+      "step": 21434
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.205052916872994e-06,
+      "loss": 0.1718,
+      "step": 21435
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2043043149637265e-06,
+      "loss": 0.1748,
+      "step": 21436
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2035558244084065e-06,
+      "loss": 0.173,
+      "step": 21437
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.202807445217722e-06,
+      "loss": 0.1893,
+      "step": 21438
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2020591774023683e-06,
+      "loss": 0.1895,
+      "step": 21439
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.201311020973028e-06,
+      "loss": 0.1931,
+      "step": 21440
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2005629759403944e-06,
+      "loss": 0.1675,
+      "step": 21441
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1998150423151486e-06,
+      "loss": 0.176,
+      "step": 21442
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1990672201079755e-06,
+      "loss": 0.1669,
+      "step": 21443
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.198319509329556e-06,
+      "loss": 0.1738,
+      "step": 21444
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1975719099905678e-06,
+      "loss": 0.1709,
+      "step": 21445
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1968244221016945e-06,
+      "loss": 0.1827,
+      "step": 21446
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.196077045673609e-06,
+      "loss": 0.1709,
+      "step": 21447
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1953297807169914e-06,
+      "loss": 0.2012,
+      "step": 21448
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1945826272425143e-06,
+      "loss": 0.1693,
+      "step": 21449
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.193835585260845e-06,
+      "loss": 0.1859,
+      "step": 21450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1930886547826613e-06,
+      "loss": 0.1736,
+      "step": 21451
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1923418358186264e-06,
+      "loss": 0.1601,
+      "step": 21452
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1915951283794145e-06,
+      "loss": 0.1775,
+      "step": 21453
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190848532475688e-06,
+      "loss": 0.1779,
+      "step": 21454
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190102048118108e-06,
+      "loss": 0.2053,
+      "step": 21455
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.189355675317346e-06,
+      "loss": 0.1617,
+      "step": 21456
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.188609414084052e-06,
+      "loss": 0.1704,
+      "step": 21457
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1878632644288943e-06,
+      "loss": 0.1825,
+      "step": 21458
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.187117226362526e-06,
+      "loss": 0.1679,
+      "step": 21459
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.186371299895608e-06,
+      "loss": 0.1467,
+      "step": 21460
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1856254850387927e-06,
+      "loss": 0.1799,
+      "step": 21461
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.18487978180273e-06,
+      "loss": 0.1825,
+      "step": 21462
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.184134190198077e-06,
+      "loss": 0.1817,
+      "step": 21463
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1833887102354833e-06,
+      "loss": 0.1803,
+      "step": 21464
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.182643341925592e-06,
+      "loss": 0.1975,
+      "step": 21465
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1818980852790562e-06,
+      "loss": 0.183,
+      "step": 21466
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1811529403065166e-06,
+      "loss": 0.18,
+      "step": 21467
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.180407907018621e-06,
+      "loss": 0.1771,
+      "step": 21468
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1796629854260097e-06,
+      "loss": 0.1682,
+      "step": 21469
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.178918175539323e-06,
+      "loss": 0.1738,
+      "step": 21470
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1781734773691988e-06,
+      "loss": 0.1873,
+      "step": 21471
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.177428890926273e-06,
+      "loss": 0.1709,
+      "step": 21472
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1766844162211866e-06,
+      "loss": 0.1821,
+      "step": 21473
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.175940053264568e-06,
+      "loss": 0.1766,
+      "step": 21474
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.175195802067054e-06,
+      "loss": 0.1845,
+      "step": 21475
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1744516626392743e-06,
+      "loss": 0.1648,
+      "step": 21476
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1737076349918552e-06,
+      "loss": 0.1835,
+      "step": 21477
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.172963719135429e-06,
+      "loss": 0.1863,
+      "step": 21478
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1722199150806177e-06,
+      "loss": 0.1688,
+      "step": 21479
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1714762228380517e-06,
+      "loss": 0.1626,
+      "step": 21480
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.170732642418348e-06,
+      "loss": 0.169,
+      "step": 21481
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.169989173832129e-06,
+      "loss": 0.1649,
+      "step": 21482
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.16924581709002e-06,
+      "loss": 0.1835,
+      "step": 21483
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1685025722026286e-06,
+      "loss": 0.1791,
+      "step": 21484
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.16775943918058e-06,
+      "loss": 0.1739,
+      "step": 21485
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.167016418034484e-06,
+      "loss": 0.1722,
+      "step": 21486
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1662735087749597e-06,
+      "loss": 0.1721,
+      "step": 21487
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1655307114126156e-06,
+      "loss": 0.1753,
+      "step": 21488
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1647880259580587e-06,
+      "loss": 0.1556,
+      "step": 21489
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1640454524219034e-06,
+      "loss": 0.1718,
+      "step": 21490
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1633029908147506e-06,
+      "loss": 0.1697,
+      "step": 21491
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1625606411472123e-06,
+      "loss": 0.1915,
+      "step": 21492
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1618184034298884e-06,
+      "loss": 0.1877,
+      "step": 21493
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1610762776733795e-06,
+      "loss": 0.1894,
+      "step": 21494
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1603342638882896e-06,
+      "loss": 0.2034,
+      "step": 21495
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1595923620852175e-06,
+      "loss": 0.1772,
+      "step": 21496
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.158850572274759e-06,
+      "loss": 0.1828,
+      "step": 21497
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.158108894467509e-06,
+      "loss": 0.1645,
+      "step": 21498
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1573673286740603e-06,
+      "loss": 0.1653,
+      "step": 21499
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1566258749050105e-06,
+      "loss": 0.1748,
+      "step": 21500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1558845331709453e-06,
+      "loss": 0.1717,
+      "step": 21501
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1551433034824598e-06,
+      "loss": 0.1933,
+      "step": 21502
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1544021858501384e-06,
+      "loss": 0.1668,
+      "step": 21503
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1536611802845654e-06,
+      "loss": 0.1831,
+      "step": 21504
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1529202867963294e-06,
+      "loss": 0.1667,
+      "step": 21505
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1521795053960082e-06,
+      "loss": 0.1703,
+      "step": 21506
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1514388360941898e-06,
+      "loss": 0.1642,
+      "step": 21507
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1506982789014496e-06,
+      "loss": 0.1971,
+      "step": 21508
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1499578338283677e-06,
+      "loss": 0.1785,
+      "step": 21509
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149217500885519e-06,
+      "loss": 0.1703,
+      "step": 21510
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.148477280083475e-06,
+      "loss": 0.1865,
+      "step": 21511
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.147737171432818e-06,
+      "loss": 0.175,
+      "step": 21512
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1469971749441098e-06,
+      "loss": 0.2005,
+      "step": 21513
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1462572906279288e-06,
+      "loss": 0.1799,
+      "step": 21514
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1455175184948407e-06,
+      "loss": 0.1818,
+      "step": 21515
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.144777858555409e-06,
+      "loss": 0.1721,
+      "step": 21516
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1440383108202044e-06,
+      "loss": 0.1866,
+      "step": 21517
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.143298875299785e-06,
+      "loss": 0.19,
+      "step": 21518
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1425595520047195e-06,
+      "loss": 0.1872,
+      "step": 21519
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.141820340945565e-06,
+      "loss": 0.1539,
+      "step": 21520
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1410812421328775e-06,
+      "loss": 0.1797,
+      "step": 21521
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1403422555772226e-06,
+      "loss": 0.1757,
+      "step": 21522
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.139603381289145e-06,
+      "loss": 0.1855,
+      "step": 21523
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.138864619279207e-06,
+      "loss": 0.1702,
+      "step": 21524
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.138125969557957e-06,
+      "loss": 0.1764,
+      "step": 21525
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137387432135949e-06,
+      "loss": 0.1746,
+      "step": 21526
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1366490070237313e-06,
+      "loss": 0.1825,
+      "step": 21527
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1359106942318475e-06,
+      "loss": 0.1792,
+      "step": 21528
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1351724937708495e-06,
+      "loss": 0.1815,
+      "step": 21529
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.13443440565128e-06,
+      "loss": 0.1688,
+      "step": 21530
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1336964298836795e-06,
+      "loss": 0.1794,
+      "step": 21531
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1329585664785922e-06,
+      "loss": 0.1698,
+      "step": 21532
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1322208154465552e-06,
+      "loss": 0.1843,
+      "step": 21533
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1314831767981102e-06,
+      "loss": 0.1696,
+      "step": 21534
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1307456505437918e-06,
+      "loss": 0.1703,
+      "step": 21535
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.130008236694133e-06,
+      "loss": 0.1767,
+      "step": 21536
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.129270935259671e-06,
+      "loss": 0.1769,
+      "step": 21537
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1285337462509304e-06,
+      "loss": 0.1605,
+      "step": 21538
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.127796669678449e-06,
+      "loss": 0.1668,
+      "step": 21539
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1270597055527498e-06,
+      "loss": 0.1691,
+      "step": 21540
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.126322853884364e-06,
+      "loss": 0.1755,
+      "step": 21541
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.125586114683814e-06,
+      "loss": 0.1831,
+      "step": 21542
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.124849487961622e-06,
+      "loss": 0.1668,
+      "step": 21543
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.124112973728315e-06,
+      "loss": 0.1878,
+      "step": 21544
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.123376571994408e-06,
+      "loss": 0.1642,
+      "step": 21545
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.122640282770424e-06,
+      "loss": 0.1713,
+      "step": 21546
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.121904106066879e-06,
+      "loss": 0.1828,
+      "step": 21547
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.121168041894286e-06,
+      "loss": 0.1973,
+      "step": 21548
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.120432090263165e-06,
+      "loss": 0.1991,
+      "step": 21549
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.119696251184019e-06,
+      "loss": 0.1574,
+      "step": 21550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1189605246673674e-06,
+      "loss": 0.1861,
+      "step": 21551
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.118224910723713e-06,
+      "loss": 0.1792,
+      "step": 21552
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.117489409363569e-06,
+      "loss": 0.1943,
+      "step": 21553
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116754020597439e-06,
+      "loss": 0.1716,
+      "step": 21554
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116018744435824e-06,
+      "loss": 0.1746,
+      "step": 21555
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1152835808892326e-06,
+      "loss": 0.1845,
+      "step": 21556
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.114548529968159e-06,
+      "loss": 0.174,
+      "step": 21557
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1138135916831103e-06,
+      "loss": 0.1746,
+      "step": 21558
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1130787660445806e-06,
+      "loss": 0.188,
+      "step": 21559
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1123440530630634e-06,
+      "loss": 0.1746,
+      "step": 21560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1116094527490594e-06,
+      "loss": 0.1775,
+      "step": 21561
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1108749651130587e-06,
+      "loss": 0.169,
+      "step": 21562
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1101405901655526e-06,
+      "loss": 0.1765,
+      "step": 21563
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1094063279170274e-06,
+      "loss": 0.179,
+      "step": 21564
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1086721783779773e-06,
+      "loss": 0.1791,
+      "step": 21565
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1079381415588875e-06,
+      "loss": 0.1727,
+      "step": 21566
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.107204217470238e-06,
+      "loss": 0.1748,
+      "step": 21567
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1064704061225196e-06,
+      "loss": 0.1792,
+      "step": 21568
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10573670752621e-06,
+      "loss": 0.182,
+      "step": 21569
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1050031216917887e-06,
+      "loss": 0.2019,
+      "step": 21570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1042696486297376e-06,
+      "loss": 0.1767,
+      "step": 21571
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1035362883505283e-06,
+      "loss": 0.1785,
+      "step": 21572
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1028030408646415e-06,
+      "loss": 0.1701,
+      "step": 21573
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10206990618255e-06,
+      "loss": 0.1944,
+      "step": 21574
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1013368843147253e-06,
+      "loss": 0.1846,
+      "step": 21575
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100603975271636e-06,
+      "loss": 0.1715,
+      "step": 21576
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0998711790637504e-06,
+      "loss": 0.1861,
+      "step": 21577
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0991384957015406e-06,
+      "loss": 0.1598,
+      "step": 21578
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.098405925195466e-06,
+      "loss": 0.1815,
+      "step": 21579
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0976734675559973e-06,
+      "loss": 0.183,
+      "step": 21580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0969411227935933e-06,
+      "loss": 0.163,
+      "step": 21581
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.096208890918713e-06,
+      "loss": 0.192,
+      "step": 21582
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.09547677194182e-06,
+      "loss": 0.1781,
+      "step": 21583
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0947447658733677e-06,
+      "loss": 0.1925,
+      "step": 21584
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0940128727238175e-06,
+      "loss": 0.1903,
+      "step": 21585
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0932810925036195e-06,
+      "loss": 0.1704,
+      "step": 21586
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.092549425223225e-06,
+      "loss": 0.1645,
+      "step": 21587
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0918178708930923e-06,
+      "loss": 0.197,
+      "step": 21588
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.091086429523662e-06,
+      "loss": 0.168,
+      "step": 21589
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.090355101125389e-06,
+      "loss": 0.1934,
+      "step": 21590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0896238857087125e-06,
+      "loss": 0.1763,
+      "step": 21591
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0888927832840853e-06,
+      "loss": 0.1639,
+      "step": 21592
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.088161793861948e-06,
+      "loss": 0.1887,
+      "step": 21593
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0874309174527365e-06,
+      "loss": 0.1779,
+      "step": 21594
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0867001540668984e-06,
+      "loss": 0.1786,
+      "step": 21595
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085969503714866e-06,
+      "loss": 0.1755,
+      "step": 21596
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0852389664070803e-06,
+      "loss": 0.1685,
+      "step": 21597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0845085421539757e-06,
+      "loss": 0.1748,
+      "step": 21598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0837782309659817e-06,
+      "loss": 0.1824,
+      "step": 21599
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.083048032853534e-06,
+      "loss": 0.1734,
+      "step": 21600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.082317947827063e-06,
+      "loss": 0.1763,
+      "step": 21601
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.081587975896995e-06,
+      "loss": 0.1826,
+      "step": 21602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0808581170737586e-06,
+      "loss": 0.1791,
+      "step": 21603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0801283713677744e-06,
+      "loss": 0.1794,
+      "step": 21604
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0793987387894743e-06,
+      "loss": 0.1684,
+      "step": 21605
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.078669219349272e-06,
+      "loss": 0.1654,
+      "step": 21606
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0779398130575956e-06,
+      "loss": 0.1684,
+      "step": 21607
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0772105199248595e-06,
+      "loss": 0.1837,
+      "step": 21608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0764813399614803e-06,
+      "loss": 0.1686,
+      "step": 21609
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0757522731778766e-06,
+      "loss": 0.1912,
+      "step": 21610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0750233195844595e-06,
+      "loss": 0.1654,
+      "step": 21611
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.074294479191645e-06,
+      "loss": 0.1488,
+      "step": 21612
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.073565752009842e-06,
+      "loss": 0.178,
+      "step": 21613
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0728371380494592e-06,
+      "loss": 0.1621,
+      "step": 21614
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0721086373209055e-06,
+      "loss": 0.1935,
+      "step": 21615
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0713802498345816e-06,
+      "loss": 0.1757,
+      "step": 21616
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.070651975600899e-06,
+      "loss": 0.1952,
+      "step": 21617
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.069923814630255e-06,
+      "loss": 0.1589,
+      "step": 21618
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.069195766933055e-06,
+      "loss": 0.1724,
+      "step": 21619
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.068467832519697e-06,
+      "loss": 0.1656,
+      "step": 21620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.067740011400575e-06,
+      "loss": 0.1752,
+      "step": 21621
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0670123035860923e-06,
+      "loss": 0.1857,
+      "step": 21622
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0662847090866366e-06,
+      "loss": 0.1738,
+      "step": 21623
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.065557227912607e-06,
+      "loss": 0.1905,
+      "step": 21624
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.064829860074391e-06,
+      "loss": 0.1778,
+      "step": 21625
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0641026055823765e-06,
+      "loss": 0.1671,
+      "step": 21626
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.063375464446957e-06,
+      "loss": 0.169,
+      "step": 21627
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.062648436678517e-06,
+      "loss": 0.1919,
+      "step": 21628
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.06192152228744e-06,
+      "loss": 0.1859,
+      "step": 21629
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0611947212841076e-06,
+      "loss": 0.177,
+      "step": 21630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.060468033678906e-06,
+      "loss": 0.1804,
+      "step": 21631
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.059741459482214e-06,
+      "loss": 0.203,
+      "step": 21632
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0590149987044062e-06,
+      "loss": 0.1625,
+      "step": 21633
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0582886513558642e-06,
+      "loss": 0.172,
+      "step": 21634
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.057562417446961e-06,
+      "loss": 0.1583,
+      "step": 21635
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0568362969880685e-06,
+      "loss": 0.1593,
+      "step": 21636
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0561102899895637e-06,
+      "loss": 0.1992,
+      "step": 21637
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.055384396461809e-06,
+      "loss": 0.1649,
+      "step": 21638
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0546586164151827e-06,
+      "loss": 0.1746,
+      "step": 21639
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.053932949860046e-06,
+      "loss": 0.1775,
+      "step": 21640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0532073968067644e-06,
+      "loss": 0.1791,
+      "step": 21641
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.052481957265704e-06,
+      "loss": 0.1664,
+      "step": 21642
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0517566312472213e-06,
+      "loss": 0.1826,
+      "step": 21643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0510314187616852e-06,
+      "loss": 0.1771,
+      "step": 21644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0503063198194474e-06,
+      "loss": 0.1797,
+      "step": 21645
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0495813344308713e-06,
+      "loss": 0.1686,
+      "step": 21646
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0488564626063102e-06,
+      "loss": 0.1935,
+      "step": 21647
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0481317043561144e-06,
+      "loss": 0.1776,
+      "step": 21648
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0474070596906427e-06,
+      "loss": 0.1776,
+      "step": 21649
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0466825286202407e-06,
+      "loss": 0.1783,
+      "step": 21650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0459581111552617e-06,
+      "loss": 0.1717,
+      "step": 21651
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045233807306052e-06,
+      "loss": 0.1949,
+      "step": 21652
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0445096170829547e-06,
+      "loss": 0.1996,
+      "step": 21653
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0437855404963214e-06,
+      "loss": 0.1779,
+      "step": 21654
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0430615775564844e-06,
+      "loss": 0.1634,
+      "step": 21655
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.042337728273793e-06,
+      "loss": 0.1726,
+      "step": 21656
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.041613992658582e-06,
+      "loss": 0.1765,
+      "step": 21657
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0408903707211937e-06,
+      "loss": 0.1606,
+      "step": 21658
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0401668624719628e-06,
+      "loss": 0.1714,
+      "step": 21659
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039443467921219e-06,
+      "loss": 0.1745,
+      "step": 21660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0387201870793027e-06,
+      "loss": 0.1798,
+      "step": 21661
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.037997019956539e-06,
+      "loss": 0.1771,
+      "step": 21662
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0372739665632635e-06,
+      "loss": 0.1739,
+      "step": 21663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.036551026909802e-06,
+      "loss": 0.1794,
+      "step": 21664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0358282010064766e-06,
+      "loss": 0.174,
+      "step": 21665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.035105488863619e-06,
+      "loss": 0.1879,
+      "step": 21666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0343828904915497e-06,
+      "loss": 0.1818,
+      "step": 21667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0336604059005895e-06,
+      "loss": 0.1748,
+      "step": 21668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.032938035101056e-06,
+      "loss": 0.1702,
+      "step": 21669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0322157781032737e-06,
+      "loss": 0.1825,
+      "step": 21670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0314936349175564e-06,
+      "loss": 0.2012,
+      "step": 21671
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.030771605554216e-06,
+      "loss": 0.1799,
+      "step": 21672
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.030049690023571e-06,
+      "loss": 0.1667,
+      "step": 21673
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0293278883359313e-06,
+      "loss": 0.1838,
+      "step": 21674
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.028606200501605e-06,
+      "loss": 0.1896,
+      "step": 21675
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027884626530905e-06,
+      "loss": 0.1914,
+      "step": 21676
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027163166434133e-06,
+      "loss": 0.1913,
+      "step": 21677
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0264418202215998e-06,
+      "loss": 0.1799,
+      "step": 21678
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025720587903607e-06,
+      "loss": 0.167,
+      "step": 21679
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0249994694904564e-06,
+      "loss": 0.1918,
+      "step": 21680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0242784649924487e-06,
+      "loss": 0.1782,
+      "step": 21681
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0235575744198798e-06,
+      "loss": 0.1657,
+      "step": 21682
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.022836797783052e-06,
+      "loss": 0.172,
+      "step": 21683
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0221161350922557e-06,
+      "loss": 0.1847,
+      "step": 21684
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0213955863577915e-06,
+      "loss": 0.1816,
+      "step": 21685
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.020675151589947e-06,
+      "loss": 0.1796,
+      "step": 21686
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0199548307990112e-06,
+      "loss": 0.1821,
+      "step": 21687
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0192346239952785e-06,
+      "loss": 0.1956,
+      "step": 21688
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0185145311890308e-06,
+      "loss": 0.1861,
+      "step": 21689
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0177945523905594e-06,
+      "loss": 0.1924,
+      "step": 21690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0170746876101465e-06,
+      "loss": 0.1656,
+      "step": 21691
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0163549368580714e-06,
+      "loss": 0.1787,
+      "step": 21692
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.015635300144623e-06,
+      "loss": 0.1653,
+      "step": 21693
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.01491577748007e-06,
+      "loss": 0.1974,
+      "step": 21694
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.014196368874698e-06,
+      "loss": 0.1635,
+      "step": 21695
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0134770743387775e-06,
+      "loss": 0.17,
+      "step": 21696
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0127578938825885e-06,
+      "loss": 0.2011,
+      "step": 21697
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0120388275164016e-06,
+      "loss": 0.1853,
+      "step": 21698
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.011319875250485e-06,
+      "loss": 0.1614,
+      "step": 21699
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0106010370951135e-06,
+      "loss": 0.1697,
+      "step": 21700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0098823130605494e-06,
+      "loss": 0.1717,
+      "step": 21701
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0091637031570644e-06,
+      "loss": 0.1849,
+      "step": 21702
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.00844520739492e-06,
+      "loss": 0.1616,
+      "step": 21703
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0077268257843784e-06,
+      "loss": 0.1755,
+      "step": 21704
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0070085583357036e-06,
+      "loss": 0.1742,
+      "step": 21705
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.006290405059156e-06,
+      "loss": 0.1738,
+      "step": 21706
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0055723659649907e-06,
+      "loss": 0.1562,
+      "step": 21707
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.004854441063465e-06,
+      "loss": 0.2036,
+      "step": 21708
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0041366303648324e-06,
+      "loss": 0.1798,
+      "step": 21709
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.00341893387935e-06,
+      "loss": 0.1733,
+      "step": 21710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0027013516172653e-06,
+      "loss": 0.1594,
+      "step": 21711
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0019838835888316e-06,
+      "loss": 0.1876,
+      "step": 21712
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.001266529804298e-06,
+      "loss": 0.1559,
+      "step": 21713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0005492902739042e-06,
+      "loss": 0.1738,
+      "step": 21714
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9998321650079032e-06,
+      "loss": 0.2014,
+      "step": 21715
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9991151540165334e-06,
+      "loss": 0.1637,
+      "step": 21716
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9983982573100413e-06,
+      "loss": 0.1902,
+      "step": 21717
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.997681474898664e-06,
+      "loss": 0.1711,
+      "step": 21718
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9969648067926383e-06,
+      "loss": 0.1674,
+      "step": 21719
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9962482530022074e-06,
+      "loss": 0.1769,
+      "step": 21720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9955318135375977e-06,
+      "loss": 0.1822,
+      "step": 21721
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994815488409051e-06,
+      "loss": 0.1763,
+      "step": 21722
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994099277626792e-06,
+      "loss": 0.1684,
+      "step": 21723
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9933831812010574e-06,
+      "loss": 0.1823,
+      "step": 21724
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9926671991420742e-06,
+      "loss": 0.176,
+      "step": 21725
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.991951331460066e-06,
+      "loss": 0.1928,
+      "step": 21726
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9912355781652635e-06,
+      "loss": 0.1527,
+      "step": 21727
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9905199392678854e-06,
+      "loss": 0.1816,
+      "step": 21728
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.989804414778159e-06,
+      "loss": 0.1904,
+      "step": 21729
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.989089004706304e-06,
+      "loss": 0.1784,
+      "step": 21730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.988373709062533e-06,
+      "loss": 0.1724,
+      "step": 21731
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9876585278570716e-06,
+      "loss": 0.1737,
+      "step": 21732
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9869434611001324e-06,
+      "loss": 0.1627,
+      "step": 21733
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9862285088019282e-06,
+      "loss": 0.1916,
+      "step": 21734
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985513670972671e-06,
+      "loss": 0.1688,
+      "step": 21735
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9847989476225736e-06,
+      "loss": 0.1902,
+      "step": 21736
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.984084338761846e-06,
+      "loss": 0.1904,
+      "step": 21737
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.98336984440069e-06,
+      "loss": 0.1585,
+      "step": 21738
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982655464549319e-06,
+      "loss": 0.1646,
+      "step": 21739
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9819411992179328e-06,
+      "loss": 0.177,
+      "step": 21740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.981227048416733e-06,
+      "loss": 0.1688,
+      "step": 21741
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9805130121559245e-06,
+      "loss": 0.1908,
+      "step": 21742
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.979799090445702e-06,
+      "loss": 0.1846,
+      "step": 21743
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9790852832962683e-06,
+      "loss": 0.1725,
+      "step": 21744
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.978371590717817e-06,
+      "loss": 0.1995,
+      "step": 21745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9776580127205424e-06,
+      "loss": 0.1977,
+      "step": 21746
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9769445493146365e-06,
+      "loss": 0.1826,
+      "step": 21747
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9762312005102903e-06,
+      "loss": 0.1754,
+      "step": 21748
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.975517966317695e-06,
+      "loss": 0.1871,
+      "step": 21749
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974804846747036e-06,
+      "loss": 0.1706,
+      "step": 21750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974091841808503e-06,
+      "loss": 0.1884,
+      "step": 21751
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.973378951512279e-06,
+      "loss": 0.187,
+      "step": 21752
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9726661758685438e-06,
+      "loss": 0.1876,
+      "step": 21753
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9719535148874848e-06,
+      "loss": 0.1747,
+      "step": 21754
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9712409685792755e-06,
+      "loss": 0.1638,
+      "step": 21755
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9705285369540994e-06,
+      "loss": 0.1649,
+      "step": 21756
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9698162200221306e-06,
+      "loss": 0.1801,
+      "step": 21757
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9691040177935415e-06,
+      "loss": 0.1504,
+      "step": 21758
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9683919302785114e-06,
+      "loss": 0.1568,
+      "step": 21759
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967679957487203e-06,
+      "loss": 0.1742,
+      "step": 21760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9669680994297934e-06,
+      "loss": 0.1756,
+      "step": 21761
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.966256356116446e-06,
+      "loss": 0.1617,
+      "step": 21762
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9655447275573326e-06,
+      "loss": 0.1795,
+      "step": 21763
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9648332137626137e-06,
+      "loss": 0.1807,
+      "step": 21764
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9641218147424524e-06,
+      "loss": 0.204,
+      "step": 21765
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.963410530507015e-06,
+      "loss": 0.1919,
+      "step": 21766
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.962699361066455e-06,
+      "loss": 0.1925,
+      "step": 21767
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9619883064309375e-06,
+      "loss": 0.1589,
+      "step": 21768
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9612773666106155e-06,
+      "loss": 0.1916,
+      "step": 21769
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9605665416156417e-06,
+      "loss": 0.1856,
+      "step": 21770
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9598558314561756e-06,
+      "loss": 0.1636,
+      "step": 21771
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9591452361423623e-06,
+      "loss": 0.1858,
+      "step": 21772
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.958434755684362e-06,
+      "loss": 0.1798,
+      "step": 21773
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9577243900923106e-06,
+      "loss": 0.1659,
+      "step": 21774
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957014139376364e-06,
+      "loss": 0.1881,
+      "step": 21775
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.956304003546664e-06,
+      "loss": 0.1849,
+      "step": 21776
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.955593982613353e-06,
+      "loss": 0.1746,
+      "step": 21777
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9548840765865775e-06,
+      "loss": 0.2017,
+      "step": 21778
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9541742854764735e-06,
+      "loss": 0.1948,
+      "step": 21779
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9534646092931796e-06,
+      "loss": 0.1827,
+      "step": 21780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9527550480468373e-06,
+      "loss": 0.1719,
+      "step": 21781
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9520456017475754e-06,
+      "loss": 0.1838,
+      "step": 21782
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.951336270405535e-06,
+      "loss": 0.18,
+      "step": 21783
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9506270540308435e-06,
+      "loss": 0.1756,
+      "step": 21784
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949917952633631e-06,
+      "loss": 0.1803,
+      "step": 21785
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949208966224032e-06,
+      "loss": 0.1899,
+      "step": 21786
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9485000948121646e-06,
+      "loss": 0.1634,
+      "step": 21787
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9477913384081614e-06,
+      "loss": 0.1912,
+      "step": 21788
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9470826970221425e-06,
+      "loss": 0.1862,
+      "step": 21789
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9463741706642337e-06,
+      "loss": 0.1762,
+      "step": 21790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9456657593445538e-06,
+      "loss": 0.1738,
+      "step": 21791
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9449574630732184e-06,
+      "loss": 0.1905,
+      "step": 21792
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944249281860352e-06,
+      "loss": 0.1924,
+      "step": 21793
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.943541215716063e-06,
+      "loss": 0.1885,
+      "step": 21794
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94283326465047e-06,
+      "loss": 0.1653,
+      "step": 21795
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.942125428673686e-06,
+      "loss": 0.174,
+      "step": 21796
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9414177077958164e-06,
+      "loss": 0.1732,
+      "step": 21797
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.940710102026977e-06,
+      "loss": 0.1912,
+      "step": 21798
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9400026113772718e-06,
+      "loss": 0.1691,
+      "step": 21799
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9392952358568064e-06,
+      "loss": 0.2126,
+      "step": 21800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.938587975475683e-06,
+      "loss": 0.1855,
+      "step": 21801
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93788083024401e-06,
+      "loss": 0.1766,
+      "step": 21802
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9371738001718843e-06,
+      "loss": 0.165,
+      "step": 21803
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9364668852694023e-06,
+      "loss": 0.1615,
+      "step": 21804
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9357600855466684e-06,
+      "loss": 0.1872,
+      "step": 21805
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9350534010137733e-06,
+      "loss": 0.1881,
+      "step": 21806
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9343468316808146e-06,
+      "loss": 0.1866,
+      "step": 21807
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9336403775578837e-06,
+      "loss": 0.1864,
+      "step": 21808
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93293403865507e-06,
+      "loss": 0.1728,
+      "step": 21809
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9322278149824657e-06,
+      "loss": 0.1748,
+      "step": 21810
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9315217065501566e-06,
+      "loss": 0.1571,
+      "step": 21811
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9308157133682306e-06,
+      "loss": 0.2108,
+      "step": 21812
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9301098354467707e-06,
+      "loss": 0.1619,
+      "step": 21813
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.929404072795856e-06,
+      "loss": 0.164,
+      "step": 21814
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9286984254255747e-06,
+      "loss": 0.1694,
+      "step": 21815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.927992893346e-06,
+      "loss": 0.1939,
+      "step": 21816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9272874765672165e-06,
+      "loss": 0.1772,
+      "step": 21817
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9265821750992965e-06,
+      "loss": 0.1834,
+      "step": 21818
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.925876988952311e-06,
+      "loss": 0.1938,
+      "step": 21819
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9251719181363395e-06,
+      "loss": 0.1736,
+      "step": 21820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9244669626614478e-06,
+      "loss": 0.1567,
+      "step": 21821
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.923762122537711e-06,
+      "loss": 0.1753,
+      "step": 21822
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.923057397775194e-06,
+      "loss": 0.1548,
+      "step": 21823
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.922352788383961e-06,
+      "loss": 0.1597,
+      "step": 21824
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9216482943740832e-06,
+      "loss": 0.1963,
+      "step": 21825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9209439157556142e-06,
+      "loss": 0.1827,
+      "step": 21826
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9202396525386237e-06,
+      "loss": 0.1737,
+      "step": 21827
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9195355047331655e-06,
+      "loss": 0.1812,
+      "step": 21828
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918831472349302e-06,
+      "loss": 0.1624,
+      "step": 21829
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918127555397089e-06,
+      "loss": 0.1753,
+      "step": 21830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.917423753886577e-06,
+      "loss": 0.1849,
+      "step": 21831
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9167200678278255e-06,
+      "loss": 0.1676,
+      "step": 21832
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916016497230879e-06,
+      "loss": 0.1618,
+      "step": 21833
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9153130421057955e-06,
+      "loss": 0.1732,
+      "step": 21834
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9146097024626177e-06,
+      "loss": 0.1737,
+      "step": 21835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9139064783113913e-06,
+      "loss": 0.1896,
+      "step": 21836
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9132033696621655e-06,
+      "loss": 0.1572,
+      "step": 21837
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9125003765249804e-06,
+      "loss": 0.1824,
+      "step": 21838
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9117974989098786e-06,
+      "loss": 0.1862,
+      "step": 21839
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9110947368268983e-06,
+      "loss": 0.1754,
+      "step": 21840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9103920902860806e-06,
+      "loss": 0.1827,
+      "step": 21841
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9096895592974606e-06,
+      "loss": 0.1762,
+      "step": 21842
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9089871438710716e-06,
+      "loss": 0.194,
+      "step": 21843
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.908284844016951e-06,
+      "loss": 0.1839,
+      "step": 21844
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9075826597451265e-06,
+      "loss": 0.1679,
+      "step": 21845
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9068805910656308e-06,
+      "loss": 0.1819,
+      "step": 21846
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9061786379884927e-06,
+      "loss": 0.1743,
+      "step": 21847
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9054768005237333e-06,
+      "loss": 0.1751,
+      "step": 21848
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9047750786813857e-06,
+      "loss": 0.1854,
+      "step": 21849
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9040734724714694e-06,
+      "loss": 0.1954,
+      "step": 21850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9033719819040041e-06,
+      "loss": 0.1692,
+      "step": 21851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9026706069890165e-06,
+      "loss": 0.1792,
+      "step": 21852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9019693477365153e-06,
+      "loss": 0.1762,
+      "step": 21853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9012682041565255e-06,
+      "loss": 0.1615,
+      "step": 21854
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9005671762590572e-06,
+      "loss": 0.1901,
+      "step": 21855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8998662640541288e-06,
+      "loss": 0.1962,
+      "step": 21856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.899165467551749e-06,
+      "loss": 0.1999,
+      "step": 21857
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8984647867619255e-06,
+      "loss": 0.1759,
+      "step": 21858
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8977642216946735e-06,
+      "loss": 0.1594,
+      "step": 21859
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8970637723599927e-06,
+      "loss": 0.1569,
+      "step": 21860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8963634387678941e-06,
+      "loss": 0.1806,
+      "step": 21861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8956632209283798e-06,
+      "loss": 0.1947,
+      "step": 21862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8949631188514484e-06,
+      "loss": 0.1725,
+      "step": 21863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.894263132547105e-06,
+      "loss": 0.1678,
+      "step": 21864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8935632620253452e-06,
+      "loss": 0.1633,
+      "step": 21865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8928635072961676e-06,
+      "loss": 0.1809,
+      "step": 21866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8921638683695642e-06,
+      "loss": 0.1742,
+      "step": 21867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8914643452555337e-06,
+      "loss": 0.1662,
+      "step": 21868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8907649379640647e-06,
+      "loss": 0.1803,
+      "step": 21869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.890065646505146e-06,
+      "loss": 0.1777,
+      "step": 21870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8893664708887715e-06,
+      "loss": 0.1991,
+      "step": 21871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8886674111249237e-06,
+      "loss": 0.1819,
+      "step": 21872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.887968467223591e-06,
+      "loss": 0.1661,
+      "step": 21873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8872696391947564e-06,
+      "loss": 0.1995,
+      "step": 21874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.886570927048399e-06,
+      "loss": 0.1907,
+      "step": 21875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8858723307945038e-06,
+      "loss": 0.1719,
+      "step": 21876
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.885173850443045e-06,
+      "loss": 0.1818,
+      "step": 21877
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.884475486004007e-06,
+      "loss": 0.1769,
+      "step": 21878
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8837772374873552e-06,
+      "loss": 0.1719,
+      "step": 21879
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8830791049030716e-06,
+      "loss": 0.1733,
+      "step": 21880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8823810882611248e-06,
+      "loss": 0.1907,
+      "step": 21881
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8816831875714836e-06,
+      "loss": 0.1732,
+      "step": 21882
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8809854028441222e-06,
+      "loss": 0.1701,
+      "step": 21883
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8802877340890035e-06,
+      "loss": 0.1734,
+      "step": 21884
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8795901813160922e-06,
+      "loss": 0.1869,
+      "step": 21885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8788927445353567e-06,
+      "loss": 0.1645,
+      "step": 21886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8781954237567535e-06,
+      "loss": 0.1646,
+      "step": 21887
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8774982189902492e-06,
+      "loss": 0.1818,
+      "step": 21888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8768011302458001e-06,
+      "loss": 0.1736,
+      "step": 21889
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8761041575333604e-06,
+      "loss": 0.165,
+      "step": 21890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8754073008628937e-06,
+      "loss": 0.185,
+      "step": 21891
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8747105602443428e-06,
+      "loss": 0.1796,
+      "step": 21892
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8740139356876686e-06,
+      "loss": 0.1775,
+      "step": 21893
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8733174272028165e-06,
+      "loss": 0.1601,
+      "step": 21894
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.872621034799741e-06,
+      "loss": 0.1704,
+      "step": 21895
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8719247584883848e-06,
+      "loss": 0.1725,
+      "step": 21896
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8712285982786937e-06,
+      "loss": 0.1853,
+      "step": 21897
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8705325541806142e-06,
+      "loss": 0.1665,
+      "step": 21898
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8698366262040858e-06,
+      "loss": 0.1587,
+      "step": 21899
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8691408143590529e-06,
+      "loss": 0.1604,
+      "step": 21900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.868445118655452e-06,
+      "loss": 0.1899,
+      "step": 21901
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8677495391032185e-06,
+      "loss": 0.1481,
+      "step": 21902
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8670540757122923e-06,
+      "loss": 0.1887,
+      "step": 21903
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8663587284926067e-06,
+      "loss": 0.1846,
+      "step": 21904
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.865663497454092e-06,
+      "loss": 0.1875,
+      "step": 21905
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8649683826066768e-06,
+      "loss": 0.1766,
+      "step": 21906
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8642733839602956e-06,
+      "loss": 0.1813,
+      "step": 21907
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8635785015248742e-06,
+      "loss": 0.1899,
+      "step": 21908
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8628837353103336e-06,
+      "loss": 0.1918,
+      "step": 21909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8621890853266045e-06,
+      "loss": 0.1665,
+      "step": 21910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8614945515836035e-06,
+      "loss": 0.1863,
+      "step": 21911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8608001340912573e-06,
+      "loss": 0.1738,
+      "step": 21912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8601058328594824e-06,
+      "loss": 0.1877,
+      "step": 21913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8594116478981928e-06,
+      "loss": 0.1924,
+      "step": 21914
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8587175792173084e-06,
+      "loss": 0.1746,
+      "step": 21915
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8580236268267437e-06,
+      "loss": 0.1767,
+      "step": 21916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8573297907364097e-06,
+      "loss": 0.1731,
+      "step": 21917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8566360709562158e-06,
+      "loss": 0.1728,
+      "step": 21918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8559424674960702e-06,
+      "loss": 0.1938,
+      "step": 21919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8552489803658846e-06,
+      "loss": 0.1623,
+      "step": 21920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8545556095755612e-06,
+      "loss": 0.1977,
+      "step": 21921
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8538623551350076e-06,
+      "loss": 0.1963,
+      "step": 21922
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8531692170541239e-06,
+      "loss": 0.1826,
+      "step": 21923
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8524761953428084e-06,
+      "loss": 0.1801,
+      "step": 21924
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8517832900109667e-06,
+      "loss": 0.1929,
+      "step": 21925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8510905010684887e-06,
+      "loss": 0.173,
+      "step": 21926
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8503978285252778e-06,
+      "loss": 0.18,
+      "step": 21927
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8497052723912245e-06,
+      "loss": 0.1871,
+      "step": 21928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8490128326762192e-06,
+      "loss": 0.1823,
+      "step": 21929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8483205093901568e-06,
+      "loss": 0.1788,
+      "step": 21930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8476283025429243e-06,
+      "loss": 0.1756,
+      "step": 21931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8469362121444101e-06,
+      "loss": 0.1815,
+      "step": 21932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8462442382044966e-06,
+      "loss": 0.1803,
+      "step": 21933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.845552380733072e-06,
+      "loss": 0.169,
+      "step": 21934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8448606397400192e-06,
+      "loss": 0.1692,
+      "step": 21935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8441690152352143e-06,
+      "loss": 0.1731,
+      "step": 21936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8434775072285427e-06,
+      "loss": 0.1796,
+      "step": 21937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8427861157298755e-06,
+      "loss": 0.1696,
+      "step": 21938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8420948407490946e-06,
+      "loss": 0.1741,
+      "step": 21939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8414036822960713e-06,
+      "loss": 0.1879,
+      "step": 21940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8407126403806764e-06,
+      "loss": 0.1826,
+      "step": 21941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8400217150127852e-06,
+      "loss": 0.1837,
+      "step": 21942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8393309062022613e-06,
+      "loss": 0.1756,
+      "step": 21943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.83864021395898e-06,
+      "loss": 0.1595,
+      "step": 21944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8379496382927986e-06,
+      "loss": 0.1768,
+      "step": 21945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8372591792135864e-06,
+      "loss": 0.1854,
+      "step": 21946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8365688367312052e-06,
+      "loss": 0.1851,
+      "step": 21947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8358786108555126e-06,
+      "loss": 0.1764,
+      "step": 21948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8351885015963744e-06,
+      "loss": 0.1789,
+      "step": 21949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8344985089636402e-06,
+      "loss": 0.1928,
+      "step": 21950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8338086329671734e-06,
+      "loss": 0.1715,
+      "step": 21951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8331188736168249e-06,
+      "loss": 0.1784,
+      "step": 21952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8324292309224434e-06,
+      "loss": 0.1918,
+      "step": 21953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8317397048938868e-06,
+      "loss": 0.178,
+      "step": 21954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8310502955410014e-06,
+      "loss": 0.1677,
+      "step": 21955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8303610028736318e-06,
+      "loss": 0.1928,
+      "step": 21956
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8296718269016311e-06,
+      "loss": 0.1849,
+      "step": 21957
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8289827676348337e-06,
+      "loss": 0.1706,
+      "step": 21958
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8282938250830906e-06,
+      "loss": 0.1809,
+      "step": 21959
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.827604999256236e-06,
+      "loss": 0.1787,
+      "step": 21960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8269162901641159e-06,
+      "loss": 0.1654,
+      "step": 21961
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8262276978165639e-06,
+      "loss": 0.1671,
+      "step": 21962
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8255392222234135e-06,
+      "loss": 0.1837,
+      "step": 21963
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.824850863394505e-06,
+      "loss": 0.1924,
+      "step": 21964
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8241626213396645e-06,
+      "loss": 0.1723,
+      "step": 21965
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.823474496068729e-06,
+      "loss": 0.2014,
+      "step": 21966
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8227864875915258e-06,
+      "loss": 0.1834,
+      "step": 21967
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8220985959178773e-06,
+      "loss": 0.1951,
+      "step": 21968
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8214108210576175e-06,
+      "loss": 0.1912,
+      "step": 21969
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8207231630205658e-06,
+      "loss": 0.1933,
+      "step": 21970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.820035621816545e-06,
+      "loss": 0.1692,
+      "step": 21971
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.819348197455375e-06,
+      "loss": 0.1725,
+      "step": 21972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8186608899468784e-06,
+      "loss": 0.1871,
+      "step": 21973
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8179736993008712e-06,
+      "loss": 0.1772,
+      "step": 21974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.817286625527166e-06,
+      "loss": 0.176,
+      "step": 21975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8165996686355824e-06,
+      "loss": 0.1823,
+      "step": 21976
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8159128286359284e-06,
+      "loss": 0.1791,
+      "step": 21977
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8152261055380183e-06,
+      "loss": 0.1704,
+      "step": 21978
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8145394993516607e-06,
+      "loss": 0.1678,
+      "step": 21979
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8138530100866592e-06,
+      "loss": 0.1886,
+      "step": 21980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8131666377528256e-06,
+      "loss": 0.1898,
+      "step": 21981
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8124803823599579e-06,
+      "loss": 0.1554,
+      "step": 21982
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.811794243917867e-06,
+      "loss": 0.1719,
+      "step": 21983
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.811108222436344e-06,
+      "loss": 0.1726,
+      "step": 21984
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8104223179251944e-06,
+      "loss": 0.1866,
+      "step": 21985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8097365303942138e-06,
+      "loss": 0.1554,
+      "step": 21986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8090508598531952e-06,
+      "loss": 0.195,
+      "step": 21987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8083653063119388e-06,
+      "loss": 0.1829,
+      "step": 21988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8076798697802344e-06,
+      "loss": 0.1615,
+      "step": 21989
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8069945502678688e-06,
+      "loss": 0.1729,
+      "step": 21990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8063093477846372e-06,
+      "loss": 0.1749,
+      "step": 21991
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.805624262340322e-06,
+      "loss": 0.1643,
+      "step": 21992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.804939293944713e-06,
+      "loss": 0.1717,
+      "step": 21993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8042544426075936e-06,
+      "loss": 0.1762,
+      "step": 21994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8035697083387427e-06,
+      "loss": 0.1764,
+      "step": 21995
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8028850911479489e-06,
+      "loss": 0.171,
+      "step": 21996
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8022005910449813e-06,
+      "loss": 0.1876,
+      "step": 21997
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8015162080396242e-06,
+      "loss": 0.1792,
+      "step": 21998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8008319421416487e-06,
+      "loss": 0.1777,
+      "step": 21999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8001477933608335e-06,
+      "loss": 0.1522,
+      "step": 22000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7994637617069498e-06,
+      "loss": 0.1909,
+      "step": 22001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7987798471897645e-06,
+      "loss": 0.1889,
+      "step": 22002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7980960498190514e-06,
+      "loss": 0.1834,
+      "step": 22003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7974123696045742e-06,
+      "loss": 0.1817,
+      "step": 22004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.796728806556104e-06,
+      "loss": 0.176,
+      "step": 22005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7960453606834004e-06,
+      "loss": 0.1724,
+      "step": 22006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7953620319962238e-06,
+      "loss": 0.1898,
+      "step": 22007
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7946788205043408e-06,
+      "loss": 0.1708,
+      "step": 22008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7939957262175046e-06,
+      "loss": 0.1704,
+      "step": 22009
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7933127491454815e-06,
+      "loss": 0.2014,
+      "step": 22010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7926298892980142e-06,
+      "loss": 0.1838,
+      "step": 22011
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.791947146684867e-06,
+      "loss": 0.1773,
+      "step": 22012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7912645213157885e-06,
+      "loss": 0.1674,
+      "step": 22013
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.790582013200528e-06,
+      "loss": 0.1763,
+      "step": 22014
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7898996223488375e-06,
+      "loss": 0.1746,
+      "step": 22015
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.78921734877046e-06,
+      "loss": 0.1769,
+      "step": 22016
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.788535192475147e-06,
+      "loss": 0.1985,
+      "step": 22017
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7878531534726396e-06,
+      "loss": 0.1693,
+      "step": 22018
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7871712317726763e-06,
+      "loss": 0.1864,
+      "step": 22019
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7864894273850043e-06,
+      "loss": 0.1763,
+      "step": 22020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7858077403193596e-06,
+      "loss": 0.1798,
+      "step": 22021
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7851261705854771e-06,
+      "loss": 0.1566,
+      "step": 22022
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.784444718193099e-06,
+      "loss": 0.1835,
+      "step": 22023
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7837633831519497e-06,
+      "loss": 0.1811,
+      "step": 22024
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7830821654717689e-06,
+      "loss": 0.178,
+      "step": 22025
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7824010651622824e-06,
+      "loss": 0.1768,
+      "step": 22026
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7817200822332238e-06,
+      "loss": 0.1777,
+      "step": 22027
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.781039216694318e-06,
+      "loss": 0.1777,
+      "step": 22028
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7803584685552877e-06,
+      "loss": 0.1735,
+      "step": 22029
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7796778378258606e-06,
+      "loss": 0.1749,
+      "step": 22030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7789973245157565e-06,
+      "loss": 0.1809,
+      "step": 22031
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7783169286346991e-06,
+      "loss": 0.1698,
+      "step": 22032
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7776366501924047e-06,
+      "loss": 0.1608,
+      "step": 22033
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7769564891985891e-06,
+      "loss": 0.1816,
+      "step": 22034
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.776276445662971e-06,
+      "loss": 0.1657,
+      "step": 22035
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.775596519595263e-06,
+      "loss": 0.1833,
+      "step": 22036
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7749167110051767e-06,
+      "loss": 0.1736,
+      "step": 22037
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7742370199024207e-06,
+      "loss": 0.1751,
+      "step": 22038
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7735574462967064e-06,
+      "loss": 0.1927,
+      "step": 22039
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.772877990197741e-06,
+      "loss": 0.1707,
+      "step": 22040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7721986516152267e-06,
+      "loss": 0.1858,
+      "step": 22041
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7715194305588711e-06,
+      "loss": 0.1839,
+      "step": 22042
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7708403270383735e-06,
+      "loss": 0.1876,
+      "step": 22043
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.1701,
+      "step": 22044
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7694824726437576e-06,
+      "loss": 0.163,
+      "step": 22045
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7688037217890308e-06,
+      "loss": 0.1643,
+      "step": 22046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7681250885089573e-06,
+      "loss": 0.187,
+      "step": 22047
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7674465728132261e-06,
+      "loss": 0.1675,
+      "step": 22048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7667681747115339e-06,
+      "loss": 0.1756,
+      "step": 22049
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.766089894213564e-06,
+      "loss": 0.1925,
+      "step": 22050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7654117313290108e-06,
+      "loss": 0.199,
+      "step": 22051
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7647336860675602e-06,
+      "loss": 0.154,
+      "step": 22052
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7640557584388929e-06,
+      "loss": 0.1851,
+      "step": 22053
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.763377948452698e-06,
+      "loss": 0.1748,
+      "step": 22054
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7627002561186534e-06,
+      "loss": 0.1897,
+      "step": 22055
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7620226814464436e-06,
+      "loss": 0.1756,
+      "step": 22056
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.761345224445744e-06,
+      "loss": 0.1606,
+      "step": 22057
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7606678851262304e-06,
+      "loss": 0.1839,
+      "step": 22058
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7599906634975817e-06,
+      "loss": 0.1815,
+      "step": 22059
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.75931355956947e-06,
+      "loss": 0.1883,
+      "step": 22060
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7586365733515631e-06,
+      "loss": 0.1753,
+      "step": 22061
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.757959704853539e-06,
+      "loss": 0.1638,
+      "step": 22062
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7572829540850579e-06,
+      "loss": 0.1654,
+      "step": 22063
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7566063210557905e-06,
+      "loss": 0.1767,
+      "step": 22064
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7559298057754004e-06,
+      "loss": 0.1683,
+      "step": 22065
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7552534082535544e-06,
+      "loss": 0.1703,
+      "step": 22066
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7545771284999114e-06,
+      "loss": 0.1775,
+      "step": 22067
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7539009665241291e-06,
+      "loss": 0.1683,
+      "step": 22068
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.753224922335871e-06,
+      "loss": 0.1846,
+      "step": 22069
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7525489959447883e-06,
+      "loss": 0.1828,
+      "step": 22070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.751873187360542e-06,
+      "loss": 0.1771,
+      "step": 22071
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7511974965927825e-06,
+      "loss": 0.1754,
+      "step": 22072
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7505219236511572e-06,
+      "loss": 0.1766,
+      "step": 22073
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7498464685453242e-06,
+      "loss": 0.1925,
+      "step": 22074
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7491711312849257e-06,
+      "loss": 0.1767,
+      "step": 22075
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7484959118796109e-06,
+      "loss": 0.1752,
+      "step": 22076
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7478208103390215e-06,
+      "loss": 0.165,
+      "step": 22077
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7471458266728048e-06,
+      "loss": 0.1753,
+      "step": 22078
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7464709608906005e-06,
+      "loss": 0.1764,
+      "step": 22079
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7457962130020466e-06,
+      "loss": 0.1639,
+      "step": 22080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7451215830167845e-06,
+      "loss": 0.1695,
+      "step": 22081
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7444470709444473e-06,
+      "loss": 0.1768,
+      "step": 22082
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7437726767946727e-06,
+      "loss": 0.1706,
+      "step": 22083
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7430984005770935e-06,
+      "loss": 0.1689,
+      "step": 22084
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7424242423013371e-06,
+      "loss": 0.1801,
+      "step": 22085
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7417502019770395e-06,
+      "loss": 0.1863,
+      "step": 22086
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7410762796138225e-06,
+      "loss": 0.1654,
+      "step": 22087
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7404024752213177e-06,
+      "loss": 0.1799,
+      "step": 22088
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7397287888091486e-06,
+      "loss": 0.1696,
+      "step": 22089
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7390552203869348e-06,
+      "loss": 0.178,
+      "step": 22090
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7383817699643013e-06,
+      "loss": 0.1951,
+      "step": 22091
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7377084375508635e-06,
+      "loss": 0.175,
+      "step": 22092
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7370352231562437e-06,
+      "loss": 0.1731,
+      "step": 22093
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7363621267900578e-06,
+      "loss": 0.1637,
+      "step": 22094
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7356891484619154e-06,
+      "loss": 0.1787,
+      "step": 22095
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7350162881814348e-06,
+      "loss": 0.1813,
+      "step": 22096
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7343435459582235e-06,
+      "loss": 0.1657,
+      "step": 22097
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7336709218018942e-06,
+      "loss": 0.1613,
+      "step": 22098
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7329984157220537e-06,
+      "loss": 0.1599,
+      "step": 22099
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.732326027728305e-06,
+      "loss": 0.1905,
+      "step": 22100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7316537578302582e-06,
+      "loss": 0.1695,
+      "step": 22101
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7309816060375117e-06,
+      "loss": 0.1845,
+      "step": 22102
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7303095723596686e-06,
+      "loss": 0.1724,
+      "step": 22103
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7296376568063245e-06,
+      "loss": 0.1983,
+      "step": 22104
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7289658593870818e-06,
+      "loss": 0.1883,
+      "step": 22105
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7282941801115349e-06,
+      "loss": 0.184,
+      "step": 22106
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7276226189892763e-06,
+      "loss": 0.1728,
+      "step": 22107
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7269511760299018e-06,
+      "loss": 0.1826,
+      "step": 22108
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7262798512429978e-06,
+      "loss": 0.1882,
+      "step": 22109
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7256086446381593e-06,
+      "loss": 0.1997,
+      "step": 22110
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7249375562249703e-06,
+      "loss": 0.1743,
+      "step": 22111
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7242665860130147e-06,
+      "loss": 0.1919,
+      "step": 22112
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7235957340118814e-06,
+      "loss": 0.1774,
+      "step": 22113
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7229250002311482e-06,
+      "loss": 0.1645,
+      "step": 22114
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7222543846804041e-06,
+      "loss": 0.1772,
+      "step": 22115
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.721583887369217e-06,
+      "loss": 0.204,
+      "step": 22116
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7209135083071716e-06,
+      "loss": 0.188,
+      "step": 22117
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7202432475038422e-06,
+      "loss": 0.1676,
+      "step": 22118
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7195731049688004e-06,
+      "loss": 0.165,
+      "step": 22119
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7189030807116213e-06,
+      "loss": 0.2005,
+      "step": 22120
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7182331747418734e-06,
+      "loss": 0.1886,
+      "step": 22121
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7175633870691299e-06,
+      "loss": 0.1766,
+      "step": 22122
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7168937177029543e-06,
+      "loss": 0.1925,
+      "step": 22123
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7162241666529101e-06,
+      "loss": 0.1797,
+      "step": 22124
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7155547339285672e-06,
+      "loss": 0.1655,
+      "step": 22125
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7148854195394848e-06,
+      "loss": 0.1943,
+      "step": 22126
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.714216223495221e-06,
+      "loss": 0.1717,
+      "step": 22127
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.713547145805341e-06,
+      "loss": 0.1915,
+      "step": 22128
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7128781864793931e-06,
+      "loss": 0.1888,
+      "step": 22129
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7122093455269407e-06,
+      "loss": 0.1839,
+      "step": 22130
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7115406229575316e-06,
+      "loss": 0.1906,
+      "step": 22131
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7108720187807227e-06,
+      "loss": 0.1957,
+      "step": 22132
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7102035330060617e-06,
+      "loss": 0.1766,
+      "step": 22133
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7095351656430959e-06,
+      "loss": 0.1681,
+      "step": 22134
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.708866916701377e-06,
+      "loss": 0.1863,
+      "step": 22135
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7081987861904448e-06,
+      "loss": 0.1757,
+      "step": 22136
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7075307741198477e-06,
+      "loss": 0.1695,
+      "step": 22137
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7068628804991262e-06,
+      "loss": 0.172,
+      "step": 22138
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7061951053378156e-06,
+      "loss": 0.1843,
+      "step": 22139
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.705527448645463e-06,
+      "loss": 0.1651,
+      "step": 22140
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7048599104316e-06,
+      "loss": 0.1797,
+      "step": 22141
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7041924907057627e-06,
+      "loss": 0.16,
+      "step": 22142
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7035251894774819e-06,
+      "loss": 0.2015,
+      "step": 22143
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7028580067562939e-06,
+      "loss": 0.1655,
+      "step": 22144
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.702190942551727e-06,
+      "loss": 0.1734,
+      "step": 22145
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7015239968733066e-06,
+      "loss": 0.1828,
+      "step": 22146
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7008571697305643e-06,
+      "loss": 0.1766,
+      "step": 22147
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.700190461133021e-06,
+      "loss": 0.1813,
+      "step": 22148
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.699523871090204e-06,
+      "loss": 0.1769,
+      "step": 22149
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.698857399611633e-06,
+      "loss": 0.1748,
+      "step": 22150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6981910467068253e-06,
+      "loss": 0.1722,
+      "step": 22151
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6975248123853029e-06,
+      "loss": 0.1842,
+      "step": 22152
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6968586966565792e-06,
+      "loss": 0.1692,
+      "step": 22153
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6961926995301747e-06,
+      "loss": 0.1632,
+      "step": 22154
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.695526821015595e-06,
+      "loss": 0.1842,
+      "step": 22155
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.694861061122356e-06,
+      "loss": 0.1968,
+      "step": 22156
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6941954198599676e-06,
+      "loss": 0.2023,
+      "step": 22157
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6935298972379333e-06,
+      "loss": 0.1802,
+      "step": 22158
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6928644932657657e-06,
+      "loss": 0.1758,
+      "step": 22159
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6921992079529636e-06,
+      "loss": 0.1877,
+      "step": 22160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6915340413090353e-06,
+      "loss": 0.172,
+      "step": 22161
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6908689933434808e-06,
+      "loss": 0.1841,
+      "step": 22162
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6902040640657946e-06,
+      "loss": 0.1704,
+      "step": 22163
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6895392534854815e-06,
+      "loss": 0.1778,
+      "step": 22164
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.688874561612035e-06,
+      "loss": 0.1746,
+      "step": 22165
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6882099884549463e-06,
+      "loss": 0.1939,
+      "step": 22166
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6875455340237145e-06,
+      "loss": 0.1913,
+      "step": 22167
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6868811983278266e-06,
+      "loss": 0.1789,
+      "step": 22168
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6862169813767727e-06,
+      "loss": 0.1718,
+      "step": 22169
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6855528831800382e-06,
+      "loss": 0.1722,
+      "step": 22170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6848889037471138e-06,
+      "loss": 0.1687,
+      "step": 22171
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6842250430874818e-06,
+      "loss": 0.1628,
+      "step": 22172
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.683561301210621e-06,
+      "loss": 0.1681,
+      "step": 22173
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6828976781260187e-06,
+      "loss": 0.1721,
+      "step": 22174
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.682234173843149e-06,
+      "loss": 0.1654,
+      "step": 22175
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6815707883714939e-06,
+      "loss": 0.203,
+      "step": 22176
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6809075217205262e-06,
+      "loss": 0.1992,
+      "step": 22177
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6802443738997188e-06,
+      "loss": 0.1802,
+      "step": 22178
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6795813449185482e-06,
+      "loss": 0.1936,
+      "step": 22179
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6789184347864806e-06,
+      "loss": 0.1643,
+      "step": 22180
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6782556435129926e-06,
+      "loss": 0.1651,
+      "step": 22181
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.677592971107541e-06,
+      "loss": 0.1832,
+      "step": 22182
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6769304175795998e-06,
+      "loss": 0.174,
+      "step": 22183
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6762679829386286e-06,
+      "loss": 0.1692,
+      "step": 22184
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6756056671940902e-06,
+      "loss": 0.2102,
+      "step": 22185
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6749434703554468e-06,
+      "loss": 0.1839,
+      "step": 22186
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6742813924321544e-06,
+      "loss": 0.1781,
+      "step": 22187
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6736194334336741e-06,
+      "loss": 0.1928,
+      "step": 22188
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6729575933694598e-06,
+      "loss": 0.1763,
+      "step": 22189
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6722958722489613e-06,
+      "loss": 0.1767,
+      "step": 22190
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6716342700816368e-06,
+      "loss": 0.1702,
+      "step": 22191
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.670972786876932e-06,
+      "loss": 0.1635,
+      "step": 22192
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6703114226442995e-06,
+      "loss": 0.163,
+      "step": 22193
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.669650177393185e-06,
+      "loss": 0.1766,
+      "step": 22194
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.668989051133032e-06,
+      "loss": 0.1844,
+      "step": 22195
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6683280438732852e-06,
+      "loss": 0.1862,
+      "step": 22196
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6676671556233838e-06,
+      "loss": 0.1865,
+      "step": 22197
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.667006386392772e-06,
+      "loss": 0.1845,
+      "step": 22198
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6663457361908875e-06,
+      "loss": 0.1931,
+      "step": 22199
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6656852050271644e-06,
+      "loss": 0.1675,
+      "step": 22200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6650247929110408e-06,
+      "loss": 0.1887,
+      "step": 22201
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.664364499851946e-06,
+      "loss": 0.1901,
+      "step": 22202
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6637043258593176e-06,
+      "loss": 0.1572,
+      "step": 22203
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.663044270942582e-06,
+      "loss": 0.1755,
+      "step": 22204
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6623843351111657e-06,
+      "loss": 0.1886,
+      "step": 22205
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6617245183745e-06,
+      "loss": 0.1626,
+      "step": 22206
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6610648207420065e-06,
+      "loss": 0.176,
+      "step": 22207
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6604052422231087e-06,
+      "loss": 0.1736,
+      "step": 22208
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6597457828272268e-06,
+      "loss": 0.1787,
+      "step": 22209
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.659086442563783e-06,
+      "loss": 0.1794,
+      "step": 22210
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.658427221442196e-06,
+      "loss": 0.1592,
+      "step": 22211
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6577681194718765e-06,
+      "loss": 0.1611,
+      "step": 22212
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.657109136662246e-06,
+      "loss": 0.2089,
+      "step": 22213
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6564502730227128e-06,
+      "loss": 0.1779,
+      "step": 22214
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6557915285626913e-06,
+      "loss": 0.1697,
+      "step": 22215
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6551329032915907e-06,
+      "loss": 0.1682,
+      "step": 22216
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6544743972188138e-06,
+      "loss": 0.1973,
+      "step": 22217
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6538160103537747e-06,
+      "loss": 0.1807,
+      "step": 22218
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6531577427058697e-06,
+      "loss": 0.1999,
+      "step": 22219
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6524995942845112e-06,
+      "loss": 0.1886,
+      "step": 22220
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6518415650990894e-06,
+      "loss": 0.1616,
+      "step": 22221
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.651183655159012e-06,
+      "loss": 0.1931,
+      "step": 22222
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6505258644736722e-06,
+      "loss": 0.1738,
+      "step": 22223
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6498681930524652e-06,
+      "loss": 0.1858,
+      "step": 22224
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6492106409047892e-06,
+      "loss": 0.1818,
+      "step": 22225
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6485532080400334e-06,
+      "loss": 0.1777,
+      "step": 22226
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6478958944675904e-06,
+      "loss": 0.1793,
+      "step": 22227
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6472387001968493e-06,
+      "loss": 0.1827,
+      "step": 22228
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6465816252371946e-06,
+      "loss": 0.1913,
+      "step": 22229
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.645924669598017e-06,
+      "loss": 0.1799,
+      "step": 22230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6452678332886972e-06,
+      "loss": 0.1735,
+      "step": 22231
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6446111163186152e-06,
+      "loss": 0.1719,
+      "step": 22232
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6439545186971573e-06,
+      "loss": 0.1816,
+      "step": 22233
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6432980404336996e-06,
+      "loss": 0.1713,
+      "step": 22234
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6426416815376178e-06,
+      "loss": 0.1743,
+      "step": 22235
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.641985442018288e-06,
+      "loss": 0.1694,
+      "step": 22236
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6413293218850857e-06,
+      "loss": 0.1672,
+      "step": 22237
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6406733211473813e-06,
+      "loss": 0.1786,
+      "step": 22238
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6400174398145442e-06,
+      "loss": 0.1875,
+      "step": 22239
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6393616778959465e-06,
+      "loss": 0.1726,
+      "step": 22240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6387060354009499e-06,
+      "loss": 0.1734,
+      "step": 22241
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6380505123389257e-06,
+      "loss": 0.1713,
+      "step": 22242
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6373951087192352e-06,
+      "loss": 0.1659,
+      "step": 22243
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6367398245512367e-06,
+      "loss": 0.1817,
+      "step": 22244
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6360846598442947e-06,
+      "loss": 0.1692,
+      "step": 22245
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6354296146077642e-06,
+      "loss": 0.1793,
+      "step": 22246
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6347746888510086e-06,
+      "loss": 0.1833,
+      "step": 22247
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6341198825833726e-06,
+      "loss": 0.1901,
+      "step": 22248
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6334651958142178e-06,
+      "loss": 0.1718,
+      "step": 22249
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6328106285528921e-06,
+      "loss": 0.1917,
+      "step": 22250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6321561808087438e-06,
+      "loss": 0.1682,
+      "step": 22251
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6315018525911264e-06,
+      "loss": 0.167,
+      "step": 22252
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6308476439093801e-06,
+      "loss": 0.176,
+      "step": 22253
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6301935547728553e-06,
+      "loss": 0.1721,
+      "step": 22254
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6295395851908924e-06,
+      "loss": 0.1676,
+      "step": 22255
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6288857351728305e-06,
+      "loss": 0.1882,
+      "step": 22256
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6282320047280132e-06,
+      "loss": 0.1661,
+      "step": 22257
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6275783938657752e-06,
+      "loss": 0.188,
+      "step": 22258
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.626924902595457e-06,
+      "loss": 0.1689,
+      "step": 22259
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6262715309263888e-06,
+      "loss": 0.1768,
+      "step": 22260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6256182788679066e-06,
+      "loss": 0.1965,
+      "step": 22261
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.624965146429338e-06,
+      "loss": 0.1827,
+      "step": 22262
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.624312133620013e-06,
+      "loss": 0.1694,
+      "step": 22263
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6236592404492624e-06,
+      "loss": 0.1828,
+      "step": 22264
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6230064669264079e-06,
+      "loss": 0.1845,
+      "step": 22265
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6223538130607786e-06,
+      "loss": 0.1713,
+      "step": 22266
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6217012788616937e-06,
+      "loss": 0.195,
+      "step": 22267
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6210488643384725e-06,
+      "loss": 0.1929,
+      "step": 22268
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6203965695004386e-06,
+      "loss": 0.1761,
+      "step": 22269
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.619744394356908e-06,
+      "loss": 0.1613,
+      "step": 22270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6190923389171919e-06,
+      "loss": 0.1895,
+      "step": 22271
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.618440403190611e-06,
+      "loss": 0.1605,
+      "step": 22272
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.617788587186474e-06,
+      "loss": 0.1633,
+      "step": 22273
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6171368909140927e-06,
+      "loss": 0.1844,
+      "step": 22274
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6164853143827718e-06,
+      "loss": 0.1811,
+      "step": 22275
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6158338576018251e-06,
+      "loss": 0.1748,
+      "step": 22276
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6151825205805538e-06,
+      "loss": 0.1709,
+      "step": 22277
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6145313033282605e-06,
+      "loss": 0.1719,
+      "step": 22278
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6138802058542514e-06,
+      "loss": 0.1836,
+      "step": 22279
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.613229228167823e-06,
+      "loss": 0.1848,
+      "step": 22280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6125783702782771e-06,
+      "loss": 0.1641,
+      "step": 22281
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6119276321949095e-06,
+      "loss": 0.1681,
+      "step": 22282
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6112770139270128e-06,
+      "loss": 0.1619,
+      "step": 22283
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6106265154838841e-06,
+      "loss": 0.2006,
+      "step": 22284
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6099761368748113e-06,
+      "loss": 0.1778,
+      "step": 22285
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6093258781090915e-06,
+      "loss": 0.1818,
+      "step": 22286
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.608675739196004e-06,
+      "loss": 0.1662,
+      "step": 22287
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6080257201448412e-06,
+      "loss": 0.1792,
+      "step": 22288
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.607375820964887e-06,
+      "loss": 0.1882,
+      "step": 22289
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6067260416654207e-06,
+      "loss": 0.1747,
+      "step": 22290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.60607638225573e-06,
+      "loss": 0.1998,
+      "step": 22291
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6054268427450892e-06,
+      "loss": 0.1926,
+      "step": 22292
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6047774231427816e-06,
+      "loss": 0.1793,
+      "step": 22293
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6041281234580808e-06,
+      "loss": 0.1862,
+      "step": 22294
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6034789437002584e-06,
+      "loss": 0.1872,
+      "step": 22295
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6028298838785928e-06,
+      "loss": 0.1704,
+      "step": 22296
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6021809440023505e-06,
+      "loss": 0.1761,
+      "step": 22297
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6015321240808045e-06,
+      "loss": 0.1682,
+      "step": 22298
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.600883424123223e-06,
+      "loss": 0.1735,
+      "step": 22299
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6002348441388693e-06,
+      "loss": 0.1838,
+      "step": 22300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5995863841370085e-06,
+      "loss": 0.1594,
+      "step": 22301
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.598938044126901e-06,
+      "loss": 0.1778,
+      "step": 22302
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5982898241178125e-06,
+      "loss": 0.1515,
+      "step": 22303
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.597641724119e-06,
+      "loss": 0.1939,
+      "step": 22304
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5969937441397188e-06,
+      "loss": 0.1737,
+      "step": 22305
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5963458841892287e-06,
+      "loss": 0.1867,
+      "step": 22306
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5956981442767793e-06,
+      "loss": 0.1713,
+      "step": 22307
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5950505244116276e-06,
+      "loss": 0.1802,
+      "step": 22308
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.594403024603023e-06,
+      "loss": 0.1734,
+      "step": 22309
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5937556448602097e-06,
+      "loss": 0.1772,
+      "step": 22310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.593108385192441e-06,
+      "loss": 0.1792,
+      "step": 22311
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5924612456089605e-06,
+      "loss": 0.1939,
+      "step": 22312
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5918142261190116e-06,
+      "loss": 0.1898,
+      "step": 22313
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5911673267318328e-06,
+      "loss": 0.1797,
+      "step": 22314
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.590520547456671e-06,
+      "loss": 0.1687,
+      "step": 22315
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5898738883027609e-06,
+      "loss": 0.1735,
+      "step": 22316
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5892273492793387e-06,
+      "loss": 0.1719,
+      "step": 22317
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5885809303956423e-06,
+      "loss": 0.1756,
+      "step": 22318
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5879346316609024e-06,
+      "loss": 0.1585,
+      "step": 22319
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5872884530843534e-06,
+      "loss": 0.1644,
+      "step": 22320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5866423946752253e-06,
+      "loss": 0.175,
+      "step": 22321
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5859964564427422e-06,
+      "loss": 0.1856,
+      "step": 22322
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.585350638396136e-06,
+      "loss": 0.1909,
+      "step": 22323
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.584704940544627e-06,
+      "loss": 0.1776,
+      "step": 22324
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5840593628974432e-06,
+      "loss": 0.1811,
+      "step": 22325
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5834139054638032e-06,
+      "loss": 0.1615,
+      "step": 22326
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.582768568252928e-06,
+      "loss": 0.179,
+      "step": 22327
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5821233512740342e-06,
+      "loss": 0.1848,
+      "step": 22328
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.581478254536336e-06,
+      "loss": 0.1658,
+      "step": 22329
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5808332780490531e-06,
+      "loss": 0.174,
+      "step": 22330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5801884218213936e-06,
+      "loss": 0.1875,
+      "step": 22331
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5795436858625735e-06,
+      "loss": 0.1782,
+      "step": 22332
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5788990701817985e-06,
+      "loss": 0.1757,
+      "step": 22333
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5782545747882761e-06,
+      "loss": 0.1846,
+      "step": 22334
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5776101996912153e-06,
+      "loss": 0.1823,
+      "step": 22335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5769659448998165e-06,
+      "loss": 0.1752,
+      "step": 22336
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.576321810423287e-06,
+      "loss": 0.1766,
+      "step": 22337
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5756777962708247e-06,
+      "loss": 0.1659,
+      "step": 22338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5750339024516292e-06,
+      "loss": 0.1652,
+      "step": 22339
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5743901289748986e-06,
+      "loss": 0.1769,
+      "step": 22340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5737464758498243e-06,
+      "loss": 0.1874,
+      "step": 22341
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.573102943085607e-06,
+      "loss": 0.1561,
+      "step": 22342
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5724595306914348e-06,
+      "loss": 0.1835,
+      "step": 22343
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5718162386764968e-06,
+      "loss": 0.178,
+      "step": 22344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.571173067049987e-06,
+      "loss": 0.1755,
+      "step": 22345
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5705300158210868e-06,
+      "loss": 0.1693,
+      "step": 22346
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5698870849989866e-06,
+      "loss": 0.1743,
+      "step": 22347
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5692442745928683e-06,
+      "loss": 0.1825,
+      "step": 22348
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5686015846119107e-06,
+      "loss": 0.1816,
+      "step": 22349
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.567959015065299e-06,
+      "loss": 0.1796,
+      "step": 22350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5673165659622068e-06,
+      "loss": 0.1813,
+      "step": 22351
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5666742373118182e-06,
+      "loss": 0.1975,
+      "step": 22352
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.566032029123299e-06,
+      "loss": 0.1682,
+      "step": 22353
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5653899414058282e-06,
+      "loss": 0.1746,
+      "step": 22354
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5647479741685767e-06,
+      "loss": 0.1762,
+      "step": 22355
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5641061274207114e-06,
+      "loss": 0.1899,
+      "step": 22356
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.563464401171405e-06,
+      "loss": 0.2038,
+      "step": 22357
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5628227954298204e-06,
+      "loss": 0.1789,
+      "step": 22358
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5621813102051254e-06,
+      "loss": 0.1966,
+      "step": 22359
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5615399455064805e-06,
+      "loss": 0.1873,
+      "step": 22360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5608987013430467e-06,
+      "loss": 0.173,
+      "step": 22361
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5602575777239858e-06,
+      "loss": 0.1631,
+      "step": 22362
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5596165746584523e-06,
+      "loss": 0.1732,
+      "step": 22363
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5589756921556077e-06,
+      "loss": 0.1901,
+      "step": 22364
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5583349302246031e-06,
+      "loss": 0.1953,
+      "step": 22365
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5576942888745905e-06,
+      "loss": 0.173,
+      "step": 22366
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5570537681147214e-06,
+      "loss": 0.159,
+      "step": 22367
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5564133679541427e-06,
+      "loss": 0.1787,
+      "step": 22368
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5557730884020073e-06,
+      "loss": 0.1793,
+      "step": 22369
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5551329294674555e-06,
+      "loss": 0.1709,
+      "step": 22370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5544928911596357e-06,
+      "loss": 0.1957,
+      "step": 22371
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5538529734876894e-06,
+      "loss": 0.1755,
+      "step": 22372
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.553213176460754e-06,
+      "loss": 0.1768,
+      "step": 22373
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5525735000879715e-06,
+      "loss": 0.1764,
+      "step": 22374
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5519339443784787e-06,
+      "loss": 0.1785,
+      "step": 22375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5512945093414089e-06,
+      "loss": 0.1715,
+      "step": 22376
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5506551949858984e-06,
+      "loss": 0.1743,
+      "step": 22377
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5500160013210796e-06,
+      "loss": 0.1885,
+      "step": 22378
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.54937692835608e-06,
+      "loss": 0.1752,
+      "step": 22379
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5487379761000276e-06,
+      "loss": 0.1625,
+      "step": 22380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5480991445620541e-06,
+      "loss": 0.1721,
+      "step": 22381
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5474604337512811e-06,
+      "loss": 0.1585,
+      "step": 22382
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546821843676829e-06,
+      "loss": 0.1833,
+      "step": 22383
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546183374347826e-06,
+      "loss": 0.1823,
+      "step": 22384
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5455450257733873e-06,
+      "loss": 0.1633,
+      "step": 22385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5449067979626341e-06,
+      "loss": 0.1585,
+      "step": 22386
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5442686909246818e-06,
+      "loss": 0.2008,
+      "step": 22387
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5436307046686426e-06,
+      "loss": 0.1791,
+      "step": 22388
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5429928392036331e-06,
+      "loss": 0.17,
+      "step": 22389
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5423550945387623e-06,
+      "loss": 0.1855,
+      "step": 22390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.541717470683144e-06,
+      "loss": 0.1623,
+      "step": 22391
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5410799676458787e-06,
+      "loss": 0.1874,
+      "step": 22392
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5404425854360783e-06,
+      "loss": 0.1912,
+      "step": 22393
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5398053240628464e-06,
+      "loss": 0.1821,
+      "step": 22394
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.539168183535281e-06,
+      "loss": 0.1754,
+      "step": 22395
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.53853116386249e-06,
+      "loss": 0.1745,
+      "step": 22396
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5378942650535666e-06,
+      "loss": 0.1699,
+      "step": 22397
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5372574871176128e-06,
+      "loss": 0.2044,
+      "step": 22398
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5366208300637231e-06,
+      "loss": 0.1774,
+      "step": 22399
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5359842939009884e-06,
+      "loss": 0.1748,
+      "step": 22400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5353478786385057e-06,
+      "loss": 0.1725,
+      "step": 22401
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5347115842853611e-06,
+      "loss": 0.1872,
+      "step": 22402
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5340754108506483e-06,
+      "loss": 0.1725,
+      "step": 22403
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5334393583434526e-06,
+      "loss": 0.1677,
+      "step": 22404
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5328034267728575e-06,
+      "loss": 0.1909,
+      "step": 22405
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5321676161479482e-06,
+      "loss": 0.186,
+      "step": 22406
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.531531926477804e-06,
+      "loss": 0.1766,
+      "step": 22407
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5308963577715096e-06,
+      "loss": 0.1701,
+      "step": 22408
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5302609100381415e-06,
+      "loss": 0.1964,
+      "step": 22409
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5296255832867735e-06,
+      "loss": 0.1669,
+      "step": 22410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.528990377526487e-06,
+      "loss": 0.1821,
+      "step": 22411
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.528355292766348e-06,
+      "loss": 0.1769,
+      "step": 22412
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5277203290154342e-06,
+      "loss": 0.1651,
+      "step": 22413
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5270854862828133e-06,
+      "loss": 0.1907,
+      "step": 22414
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5264507645775505e-06,
+      "loss": 0.1851,
+      "step": 22415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5258161639087166e-06,
+      "loss": 0.1606,
+      "step": 22416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5251816842853729e-06,
+      "loss": 0.1806,
+      "step": 22417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5245473257165878e-06,
+      "loss": 0.1582,
+      "step": 22418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.523913088211415e-06,
+      "loss": 0.1605,
+      "step": 22419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5232789717789187e-06,
+      "loss": 0.1869,
+      "step": 22420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5226449764281558e-06,
+      "loss": 0.1886,
+      "step": 22421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5220111021681793e-06,
+      "loss": 0.1758,
+      "step": 22422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5213773490080496e-06,
+      "loss": 0.1732,
+      "step": 22423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5207437169568117e-06,
+      "loss": 0.186,
+      "step": 22424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.520110206023524e-06,
+      "loss": 0.175,
+      "step": 22425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5194768162172324e-06,
+      "loss": 0.1709,
+      "step": 22426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.518843547546981e-06,
+      "loss": 0.1962,
+      "step": 22427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5182104000218202e-06,
+      "loss": 0.17,
+      "step": 22428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5175773736507903e-06,
+      "loss": 0.1806,
+      "step": 22429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5169444684429379e-06,
+      "loss": 0.1712,
+      "step": 22430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5163116844073e-06,
+      "loss": 0.1778,
+      "step": 22431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.515679021552917e-06,
+      "loss": 0.1521,
+      "step": 22432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5150464798888243e-06,
+      "loss": 0.1683,
+      "step": 22433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5144140594240553e-06,
+      "loss": 0.2033,
+      "step": 22434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5137817601676485e-06,
+      "loss": 0.1872,
+      "step": 22435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5131495821286313e-06,
+      "loss": 0.1831,
+      "step": 22436
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5125175253160374e-06,
+      "loss": 0.1835,
+      "step": 22437
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511885589738894e-06,
+      "loss": 0.1802,
+      "step": 22438
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511253775406224e-06,
+      "loss": 0.1923,
+      "step": 22439
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5106220823270579e-06,
+      "loss": 0.1607,
+      "step": 22440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.509990510510414e-06,
+      "loss": 0.1536,
+      "step": 22441
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5093590599653185e-06,
+      "loss": 0.1802,
+      "step": 22442
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5087277307007885e-06,
+      "loss": 0.173,
+      "step": 22443
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5080965227258415e-06,
+      "loss": 0.177,
+      "step": 22444
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5074654360494934e-06,
+      "loss": 0.176,
+      "step": 22445
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.506834470680758e-06,
+      "loss": 0.1767,
+      "step": 22446
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5062036266286517e-06,
+      "loss": 0.1619,
+      "step": 22447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5055729039021827e-06,
+      "loss": 0.1716,
+      "step": 22448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5049423025103593e-06,
+      "loss": 0.2023,
+      "step": 22449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5043118224621912e-06,
+      "loss": 0.1661,
+      "step": 22450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5036814637666818e-06,
+      "loss": 0.1922,
+      "step": 22451
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5030512264328401e-06,
+      "loss": 0.1539,
+      "step": 22452
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.502421110469664e-06,
+      "loss": 0.1916,
+      "step": 22453
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5017911158861532e-06,
+      "loss": 0.1677,
+      "step": 22454
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5011612426913114e-06,
+      "loss": 0.1741,
+      "step": 22455
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5005314908941305e-06,
+      "loss": 0.1897,
+      "step": 22456
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4999018605036132e-06,
+      "loss": 0.1766,
+      "step": 22457
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4992723515287423e-06,
+      "loss": 0.153,
+      "step": 22458
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4986429639785194e-06,
+      "loss": 0.175,
+      "step": 22459
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.498013697861931e-06,
+      "loss": 0.1838,
+      "step": 22460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4973845531879615e-06,
+      "loss": 0.1849,
+      "step": 22461
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4967555299656056e-06,
+      "loss": 0.1722,
+      "step": 22462
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4961266282038423e-06,
+      "loss": 0.1677,
+      "step": 22463
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4954978479116578e-06,
+      "loss": 0.1579,
+      "step": 22464
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4948691890980337e-06,
+      "loss": 0.1973,
+      "step": 22465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4942406517719467e-06,
+      "loss": 0.1817,
+      "step": 22466
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4936122359423788e-06,
+      "loss": 0.171,
+      "step": 22467
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4929839416183023e-06,
+      "loss": 0.1862,
+      "step": 22468
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4923557688086964e-06,
+      "loss": 0.1786,
+      "step": 22469
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.491727717522531e-06,
+      "loss": 0.1764,
+      "step": 22470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4910997877687783e-06,
+      "loss": 0.2025,
+      "step": 22471
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.490471979556406e-06,
+      "loss": 0.1676,
+      "step": 22472
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4898442928943813e-06,
+      "loss": 0.1721,
+      "step": 22473
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4892167277916735e-06,
+      "loss": 0.1812,
+      "step": 22474
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4885892842572436e-06,
+      "loss": 0.1699,
+      "step": 22475
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4879619623000562e-06,
+      "loss": 0.1776,
+      "step": 22476
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4873347619290712e-06,
+      "loss": 0.1879,
+      "step": 22477
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4867076831532457e-06,
+      "loss": 0.1685,
+      "step": 22478
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4860807259815413e-06,
+      "loss": 0.1546,
+      "step": 22479
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4854538904229098e-06,
+      "loss": 0.1748,
+      "step": 22480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4848271764863043e-06,
+      "loss": 0.2152,
+      "step": 22481
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4842005841806805e-06,
+      "loss": 0.162,
+      "step": 22482
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4835741135149838e-06,
+      "loss": 0.2111,
+      "step": 22483
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4829477644981704e-06,
+      "loss": 0.1957,
+      "step": 22484
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4823215371391775e-06,
+      "loss": 0.1687,
+      "step": 22485
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4816954314469579e-06,
+      "loss": 0.1774,
+      "step": 22486
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.481069447430451e-06,
+      "loss": 0.178,
+      "step": 22487
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4804435850985966e-06,
+      "loss": 0.1796,
+      "step": 22488
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4798178444603405e-06,
+      "loss": 0.1749,
+      "step": 22489
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.479192225524615e-06,
+      "loss": 0.1783,
+      "step": 22490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4785667283003601e-06,
+      "loss": 0.1756,
+      "step": 22491
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4779413527965103e-06,
+      "loss": 0.1798,
+      "step": 22492
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4773160990219948e-06,
+      "loss": 0.1815,
+      "step": 22493
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4766909669857499e-06,
+      "loss": 0.1816,
+      "step": 22494
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4760659566966995e-06,
+      "loss": 0.1744,
+      "step": 22495
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4754410681637776e-06,
+      "loss": 0.1762,
+      "step": 22496
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.474816301395906e-06,
+      "loss": 0.1834,
+      "step": 22497
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4741916564020097e-06,
+      "loss": 0.1938,
+      "step": 22498
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4735671331910117e-06,
+      "loss": 0.1826,
+      "step": 22499
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4729427317718291e-06,
+      "loss": 0.1889,
+      "step": 22500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4723184521533861e-06,
+      "loss": 0.1817,
+      "step": 22501
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4716942943445955e-06,
+      "loss": 0.1806,
+      "step": 22502
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.471070258354378e-06,
+      "loss": 0.1688,
+      "step": 22503
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4704463441916439e-06,
+      "loss": 0.1874,
+      "step": 22504
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.469822551865303e-06,
+      "loss": 0.1825,
+      "step": 22505
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4691988813842706e-06,
+      "loss": 0.1777,
+      "step": 22506
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4685753327574515e-06,
+      "loss": 0.1841,
+      "step": 22507
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4679519059937552e-06,
+      "loss": 0.1789,
+      "step": 22508
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4673286011020848e-06,
+      "loss": 0.186,
+      "step": 22509
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4667054180913453e-06,
+      "loss": 0.1779,
+      "step": 22510
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4660823569704374e-06,
+      "loss": 0.1575,
+      "step": 22511
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4654594177482573e-06,
+      "loss": 0.1676,
+      "step": 22512
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.464836600433709e-06,
+      "loss": 0.1621,
+      "step": 22513
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4642139050356864e-06,
+      "loss": 0.1771,
+      "step": 22514
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4635913315630824e-06,
+      "loss": 0.1735,
+      "step": 22515
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4629688800247933e-06,
+      "loss": 0.1664,
+      "step": 22516
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4623465504297052e-06,
+      "loss": 0.1636,
+      "step": 22517
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4617243427867133e-06,
+      "loss": 0.1807,
+      "step": 22518
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.461102257104703e-06,
+      "loss": 0.1783,
+      "step": 22519
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4604802933925576e-06,
+      "loss": 0.1868,
+      "step": 22520
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4598584516591652e-06,
+      "loss": 0.1655,
+      "step": 22521
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4592367319134048e-06,
+      "loss": 0.1618,
+      "step": 22522
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.458615134164163e-06,
+      "loss": 0.1874,
+      "step": 22523
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4579936584203091e-06,
+      "loss": 0.1699,
+      "step": 22524
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4573723046907284e-06,
+      "loss": 0.1775,
+      "step": 22525
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4567510729842937e-06,
+      "loss": 0.1834,
+      "step": 22526
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4561299633098759e-06,
+      "loss": 0.172,
+      "step": 22527
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.455508975676352e-06,
+      "loss": 0.198,
+      "step": 22528
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4548881100925872e-06,
+      "loss": 0.1625,
+      "step": 22529
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4542673665674555e-06,
+      "loss": 0.1705,
+      "step": 22530
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4536467451098201e-06,
+      "loss": 0.1576,
+      "step": 22531
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4530262457285448e-06,
+      "loss": 0.1964,
+      "step": 22532
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.452405868432496e-06,
+      "loss": 0.1729,
+      "step": 22533
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.451785613230532e-06,
+      "loss": 0.1708,
+      "step": 22534
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4511654801315168e-06,
+      "loss": 0.1589,
+      "step": 22535
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4505454691443043e-06,
+      "loss": 0.1814,
+      "step": 22536
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4499255802777545e-06,
+      "loss": 0.1725,
+      "step": 22537
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.449305813540718e-06,
+      "loss": 0.1547,
+      "step": 22538
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4486861689420485e-06,
+      "loss": 0.1821,
+      "step": 22539
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.448066646490599e-06,
+      "loss": 0.1961,
+      "step": 22540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.447447246195216e-06,
+      "loss": 0.1753,
+      "step": 22541
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.44682796806475e-06,
+      "loss": 0.1666,
+      "step": 22542
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4462088121080465e-06,
+      "loss": 0.1652,
+      "step": 22543
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4455897783339457e-06,
+      "loss": 0.1877,
+      "step": 22544
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4449708667512953e-06,
+      "loss": 0.1738,
+      "step": 22545
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.44435207736893e-06,
+      "loss": 0.1674,
+      "step": 22546
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4437334101956946e-06,
+      "loss": 0.1604,
+      "step": 22547
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4431148652404236e-06,
+      "loss": 0.1781,
+      "step": 22548
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4424964425119513e-06,
+      "loss": 0.1769,
+      "step": 22549
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4418781420191131e-06,
+      "loss": 0.1856,
+      "step": 22550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4412599637707369e-06,
+      "loss": 0.1872,
+      "step": 22551
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4406419077756585e-06,
+      "loss": 0.174,
+      "step": 22552
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4400239740427035e-06,
+      "loss": 0.1869,
+      "step": 22553
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4394061625806953e-06,
+      "loss": 0.1497,
+      "step": 22554
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4387884733984647e-06,
+      "loss": 0.1714,
+      "step": 22555
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4381709065048289e-06,
+      "loss": 0.1811,
+      "step": 22556
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.437553461908615e-06,
+      "loss": 0.1816,
+      "step": 22557
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.43693613961864e-06,
+      "loss": 0.1785,
+      "step": 22558
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4363189396437193e-06,
+      "loss": 0.1903,
+      "step": 22559
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4357018619926732e-06,
+      "loss": 0.1788,
+      "step": 22560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4350849066743123e-06,
+      "loss": 0.1517,
+      "step": 22561
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4344680736974536e-06,
+      "loss": 0.1642,
+      "step": 22562
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.433851363070905e-06,
+      "loss": 0.1787,
+      "step": 22563
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4332347748034758e-06,
+      "loss": 0.1974,
+      "step": 22564
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4326183089039747e-06,
+      "loss": 0.1694,
+      "step": 22565
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4320019653812024e-06,
+      "loss": 0.1886,
+      "step": 22566
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4313857442439693e-06,
+      "loss": 0.1846,
+      "step": 22567
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4307696455010734e-06,
+      "loss": 0.1666,
+      "step": 22568
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4301536691613183e-06,
+      "loss": 0.1632,
+      "step": 22569
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4295378152335004e-06,
+      "loss": 0.1571,
+      "step": 22570
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.428922083726415e-06,
+      "loss": 0.1813,
+      "step": 22571
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4283064746488606e-06,
+      "loss": 0.1704,
+      "step": 22572
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4276909880096278e-06,
+      "loss": 0.1875,
+      "step": 22573
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.427075623817512e-06,
+      "loss": 0.2003,
+      "step": 22574
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4264603820813006e-06,
+      "loss": 0.1767,
+      "step": 22575
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.425845262809782e-06,
+      "loss": 0.1851,
+      "step": 22576
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4252302660117422e-06,
+      "loss": 0.173,
+      "step": 22577
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4246153916959637e-06,
+      "loss": 0.1668,
+      "step": 22578
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4240006398712337e-06,
+      "loss": 0.1769,
+      "step": 22579
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4233860105463293e-06,
+      "loss": 0.1772,
+      "step": 22580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4227715037300349e-06,
+      "loss": 0.2064,
+      "step": 22581
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4221571194311246e-06,
+      "loss": 0.1854,
+      "step": 22582
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4215428576583722e-06,
+      "loss": 0.1737,
+      "step": 22583
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4209287184205567e-06,
+      "loss": 0.1889,
+      "step": 22584
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4203147017264496e-06,
+      "loss": 0.1931,
+      "step": 22585
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4197008075848161e-06,
+      "loss": 0.1615,
+      "step": 22586
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4190870360044328e-06,
+      "loss": 0.167,
+      "step": 22587
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4184733869940604e-06,
+      "loss": 0.1933,
+      "step": 22588
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4178598605624717e-06,
+      "loss": 0.1838,
+      "step": 22589
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.417246456718422e-06,
+      "loss": 0.1735,
+      "step": 22590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4166331754706774e-06,
+      "loss": 0.164,
+      "step": 22591
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4160200168279991e-06,
+      "loss": 0.1862,
+      "step": 22592
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.415406980799141e-06,
+      "loss": 0.1608,
+      "step": 22593
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.414794067392865e-06,
+      "loss": 0.1814,
+      "step": 22594
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.414181276617922e-06,
+      "loss": 0.1921,
+      "step": 22595
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.413568608483068e-06,
+      "loss": 0.1679,
+      "step": 22596
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.412956062997053e-06,
+      "loss": 0.1724,
+      "step": 22597
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4123436401686253e-06,
+      "loss": 0.1647,
+      "step": 22598
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4117313400065359e-06,
+      "loss": 0.1751,
+      "step": 22599
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4111191625195274e-06,
+      "loss": 0.1742,
+      "step": 22600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.410507107716348e-06,
+      "loss": 0.1639,
+      "step": 22601
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4098951756057378e-06,
+      "loss": 0.1663,
+      "step": 22602
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4092833661964388e-06,
+      "loss": 0.1904,
+      "step": 22603
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4086716794971889e-06,
+      "loss": 0.17,
+      "step": 22604
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.408060115516724e-06,
+      "loss": 0.1684,
+      "step": 22605
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4074486742637839e-06,
+      "loss": 0.1776,
+      "step": 22606
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4068373557470971e-06,
+      "loss": 0.2005,
+      "step": 22607
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4062261599754024e-06,
+      "loss": 0.157,
+      "step": 22608
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.405615086957426e-06,
+      "loss": 0.1654,
+      "step": 22609
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4050041367018952e-06,
+      "loss": 0.177,
+      "step": 22610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.40439330921754e-06,
+      "loss": 0.1633,
+      "step": 22611
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4037826045130832e-06,
+      "loss": 0.1782,
+      "step": 22612
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.40317202259725e-06,
+      "loss": 0.1764,
+      "step": 22613
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4025615634787616e-06,
+      "loss": 0.1734,
+      "step": 22614
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4019512271663372e-06,
+      "loss": 0.1847,
+      "step": 22615
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4013410136686955e-06,
+      "loss": 0.1689,
+      "step": 22616
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4007309229945487e-06,
+      "loss": 0.176,
+      "step": 22617
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4001209551526184e-06,
+      "loss": 0.1764,
+      "step": 22618
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3995111101516134e-06,
+      "loss": 0.1844,
+      "step": 22619
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3989013880002423e-06,
+      "loss": 0.1819,
+      "step": 22620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.39829178870722e-06,
+      "loss": 0.1745,
+      "step": 22621
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3976823122812499e-06,
+      "loss": 0.1864,
+      "step": 22622
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3970729587310406e-06,
+      "loss": 0.1858,
+      "step": 22623
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3964637280652948e-06,
+      "loss": 0.1735,
+      "step": 22624
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3958546202927125e-06,
+      "loss": 0.1633,
+      "step": 22625
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.395245635421999e-06,
+      "loss": 0.1775,
+      "step": 22626
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3946367734618481e-06,
+      "loss": 0.1867,
+      "step": 22627
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3940280344209611e-06,
+      "loss": 0.167,
+      "step": 22628
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3934194183080318e-06,
+      "loss": 0.1698,
+      "step": 22629
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3928109251317534e-06,
+      "loss": 0.1587,
+      "step": 22630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3922025549008167e-06,
+      "loss": 0.1831,
+      "step": 22631
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3915943076239113e-06,
+      "loss": 0.1819,
+      "step": 22632
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3909861833097283e-06,
+      "loss": 0.1564,
+      "step": 22633
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3903781819669493e-06,
+      "loss": 0.1783,
+      "step": 22634
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3897703036042654e-06,
+      "loss": 0.173,
+      "step": 22635
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3891625482303562e-06,
+      "loss": 0.1676,
+      "step": 22636
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3885549158539002e-06,
+      "loss": 0.1884,
+      "step": 22637
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3879474064835818e-06,
+      "loss": 0.1916,
+      "step": 22638
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3873400201280751e-06,
+      "loss": 0.1774,
+      "step": 22639
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3867327567960587e-06,
+      "loss": 0.1606,
+      "step": 22640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3861256164962066e-06,
+      "loss": 0.1653,
+      "step": 22641
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3855185992371888e-06,
+      "loss": 0.1829,
+      "step": 22642
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3849117050276784e-06,
+      "loss": 0.1691,
+      "step": 22643
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3843049338763403e-06,
+      "loss": 0.1939,
+      "step": 22644
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3836982857918469e-06,
+      "loss": 0.1664,
+      "step": 22645
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3830917607828587e-06,
+      "loss": 0.1788,
+      "step": 22646
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3824853588580445e-06,
+      "loss": 0.1805,
+      "step": 22647
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3818790800260628e-06,
+      "loss": 0.1866,
+      "step": 22648
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3812729242955724e-06,
+      "loss": 0.1871,
+      "step": 22649
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3806668916752363e-06,
+      "loss": 0.1973,
+      "step": 22650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3800609821737066e-06,
+      "loss": 0.1899,
+      "step": 22651
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3794551957996405e-06,
+      "loss": 0.1692,
+      "step": 22652
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3788495325616912e-06,
+      "loss": 0.1809,
+      "step": 22653
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3782439924685076e-06,
+      "loss": 0.1621,
+      "step": 22654
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3776385755287457e-06,
+      "loss": 0.1802,
+      "step": 22655
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3770332817510434e-06,
+      "loss": 0.1892,
+      "step": 22656
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3764281111440537e-06,
+      "loss": 0.1632,
+      "step": 22657
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3758230637164193e-06,
+      "loss": 0.1523,
+      "step": 22658
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3752181394767805e-06,
+      "loss": 0.1812,
+      "step": 22659
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3746133384337823e-06,
+      "loss": 0.1722,
+      "step": 22660
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.374008660596059e-06,
+      "loss": 0.1797,
+      "step": 22661
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3734041059722524e-06,
+      "loss": 0.1843,
+      "step": 22662
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.372799674570997e-06,
+      "loss": 0.1638,
+      "step": 22663
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3721953664009213e-06,
+      "loss": 0.1849,
+      "step": 22664
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.371591181470665e-06,
+      "loss": 0.1693,
+      "step": 22665
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3709871197888514e-06,
+      "loss": 0.1648,
+      "step": 22666
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3703831813641154e-06,
+      "loss": 0.1763,
+      "step": 22667
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3697793662050795e-06,
+      "loss": 0.1846,
+      "step": 22668
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.36917567432037e-06,
+      "loss": 0.1829,
+      "step": 22669
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3685721057186086e-06,
+      "loss": 0.1764,
+      "step": 22670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3679686604084163e-06,
+      "loss": 0.1866,
+      "step": 22671
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3673653383984164e-06,
+      "loss": 0.1908,
+      "step": 22672
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.366762139697222e-06,
+      "loss": 0.1785,
+      "step": 22673
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3661590643134548e-06,
+      "loss": 0.1771,
+      "step": 22674
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.365556112255725e-06,
+      "loss": 0.1904,
+      "step": 22675
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.364953283532644e-06,
+      "loss": 0.1636,
+      "step": 22676
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3643505781528277e-06,
+      "loss": 0.1796,
+      "step": 22677
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.363747996124879e-06,
+      "loss": 0.1858,
+      "step": 22678
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.363145537457411e-06,
+      "loss": 0.1804,
+      "step": 22679
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.362543202159028e-06,
+      "loss": 0.1637,
+      "step": 22680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3619409902383308e-06,
+      "loss": 0.1978,
+      "step": 22681
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3613389017039246e-06,
+      "loss": 0.1775,
+      "step": 22682
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3607369365644053e-06,
+      "loss": 0.1642,
+      "step": 22683
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3601350948283766e-06,
+      "loss": 0.1659,
+      "step": 22684
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3595333765044293e-06,
+      "loss": 0.1517,
+      "step": 22685
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.358931781601166e-06,
+      "loss": 0.1797,
+      "step": 22686
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.358330310127175e-06,
+      "loss": 0.1698,
+      "step": 22687
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3577289620910471e-06,
+      "loss": 0.1732,
+      "step": 22688
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3571277375013747e-06,
+      "loss": 0.177,
+      "step": 22689
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.356526636366745e-06,
+      "loss": 0.1908,
+      "step": 22690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3559256586957413e-06,
+      "loss": 0.1639,
+      "step": 22691
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3553248044969525e-06,
+      "loss": 0.1617,
+      "step": 22692
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3547240737789568e-06,
+      "loss": 0.1696,
+      "step": 22693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3541234665503422e-06,
+      "loss": 0.1748,
+      "step": 22694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3535229828196771e-06,
+      "loss": 0.1803,
+      "step": 22695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3529226225955483e-06,
+      "loss": 0.1681,
+      "step": 22696
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3523223858865264e-06,
+      "loss": 0.1932,
+      "step": 22697
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3517222727011847e-06,
+      "loss": 0.1752,
+      "step": 22698
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3511222830480997e-06,
+      "loss": 0.1895,
+      "step": 22699
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3505224169358365e-06,
+      "loss": 0.1764,
+      "step": 22700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3499226743729677e-06,
+      "loss": 0.1936,
+      "step": 22701
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3493230553680591e-06,
+      "loss": 0.1638,
+      "step": 22702
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3487235599296721e-06,
+      "loss": 0.1873,
+      "step": 22703
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3481241880663753e-06,
+      "loss": 0.1724,
+      "step": 22704
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3475249397867263e-06,
+      "loss": 0.186,
+      "step": 22705
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.346925815099287e-06,
+      "loss": 0.1671,
+      "step": 22706
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.346326814012615e-06,
+      "loss": 0.1949,
+      "step": 22707
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.345727936535266e-06,
+      "loss": 0.1807,
+      "step": 22708
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.345129182675794e-06,
+      "loss": 0.173,
+      "step": 22709
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.34453055244275e-06,
+      "loss": 0.2041,
+      "step": 22710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.343932045844688e-06,
+      "loss": 0.1586,
+      "step": 22711
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.343333662890155e-06,
+      "loss": 0.1909,
+      "step": 22712
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3427354035877005e-06,
+      "loss": 0.1752,
+      "step": 22713
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3421372679458688e-06,
+      "loss": 0.1864,
+      "step": 22714
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.341539255973201e-06,
+      "loss": 0.1843,
+      "step": 22715
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3409413676782447e-06,
+      "loss": 0.175,
+      "step": 22716
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3403436030695338e-06,
+      "loss": 0.1695,
+      "step": 22717
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.1742,
+      "step": 22718
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3391484449450164e-06,
+      "loss": 0.1686,
+      "step": 22719
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3385510514462762e-06,
+      "loss": 0.1811,
+      "step": 22720
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3379537816679332e-06,
+      "loss": 0.1771,
+      "step": 22721
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3373566356185086e-06,
+      "loss": 0.1602,
+      "step": 22722
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3367596133065397e-06,
+      "loss": 0.1787,
+      "step": 22723
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3361627147405487e-06,
+      "loss": 0.1516,
+      "step": 22724
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3355659399290676e-06,
+      "loss": 0.1706,
+      "step": 22725
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3349692888806188e-06,
+      "loss": 0.185,
+      "step": 22726
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3343727616037217e-06,
+      "loss": 0.1644,
+      "step": 22727
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3337763581069007e-06,
+      "loss": 0.1752,
+      "step": 22728
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3331800783986748e-06,
+      "loss": 0.1797,
+      "step": 22729
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.332583922487557e-06,
+      "loss": 0.1798,
+      "step": 22730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3319878903820682e-06,
+      "loss": 0.1752,
+      "step": 22731
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3313919820907185e-06,
+      "loss": 0.1821,
+      "step": 22732
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.330796197622023e-06,
+      "loss": 0.1812,
+      "step": 22733
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3302005369844905e-06,
+      "loss": 0.1768,
+      "step": 22734
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.329605000186628e-06,
+      "loss": 0.1787,
+      "step": 22735
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3290095872369446e-06,
+      "loss": 0.167,
+      "step": 22736
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3284142981439408e-06,
+      "loss": 0.1998,
+      "step": 22737
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327819132916125e-06,
+      "loss": 0.1909,
+      "step": 22738
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327224091561995e-06,
+      "loss": 0.1646,
+      "step": 22739
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3266291740900528e-06,
+      "loss": 0.1774,
+      "step": 22740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.326034380508796e-06,
+      "loss": 0.1719,
+      "step": 22741
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.325439710826717e-06,
+      "loss": 0.1798,
+      "step": 22742
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3248451650523153e-06,
+      "loss": 0.1897,
+      "step": 22743
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3242507431940787e-06,
+      "loss": 0.1639,
+      "step": 22744
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3236564452605027e-06,
+      "loss": 0.169,
+      "step": 22745
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3230622712600738e-06,
+      "loss": 0.1956,
+      "step": 22746
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3224682212012784e-06,
+      "loss": 0.1619,
+      "step": 22747
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3218742950926034e-06,
+      "loss": 0.1619,
+      "step": 22748
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3212804929425293e-06,
+      "loss": 0.1767,
+      "step": 22749
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.320686814759543e-06,
+      "loss": 0.1601,
+      "step": 22750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.32009326055212e-06,
+      "loss": 0.1788,
+      "step": 22751
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.319499830328742e-06,
+      "loss": 0.1714,
+      "step": 22752
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.318906524097885e-06,
+      "loss": 0.1846,
+      "step": 22753
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3183133418680205e-06,
+      "loss": 0.1732,
+      "step": 22754
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3177202836476267e-06,
+      "loss": 0.1768,
+      "step": 22755
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.317127349445171e-06,
+      "loss": 0.1696,
+      "step": 22756
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3165345392691253e-06,
+      "loss": 0.1637,
+      "step": 22757
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3159418531279556e-06,
+      "loss": 0.1754,
+      "step": 22758
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.315349291030127e-06,
+      "loss": 0.1716,
+      "step": 22759
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3147568529841105e-06,
+      "loss": 0.1825,
+      "step": 22760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.314164538998357e-06,
+      "loss": 0.1775,
+      "step": 22761
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3135723490813367e-06,
+      "loss": 0.157,
+      "step": 22762
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.312980283241504e-06,
+      "loss": 0.1689,
+      "step": 22763
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.312388341487315e-06,
+      "loss": 0.1956,
+      "step": 22764
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.31179652382723e-06,
+      "loss": 0.1777,
+      "step": 22765
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3112048302696968e-06,
+      "loss": 0.1487,
+      "step": 22766
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3106132608231725e-06,
+      "loss": 0.1809,
+      "step": 22767
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3100218154961042e-06,
+      "loss": 0.1633,
+      "step": 22768
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3094304942969382e-06,
+      "loss": 0.1517,
+      "step": 22769
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3088392972341256e-06,
+      "loss": 0.1768,
+      "step": 22770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3082482243161066e-06,
+      "loss": 0.1931,
+      "step": 22771
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3076572755513273e-06,
+      "loss": 0.1892,
+      "step": 22772
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3070664509482288e-06,
+      "loss": 0.1726,
+      "step": 22773
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3064757505152493e-06,
+      "loss": 0.1738,
+      "step": 22774
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3058851742608259e-06,
+      "loss": 0.1796,
+      "step": 22775
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3052947221933932e-06,
+      "loss": 0.1591,
+      "step": 22776
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.304704394321389e-06,
+      "loss": 0.1725,
+      "step": 22777
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3041141906532417e-06,
+      "loss": 0.1626,
+      "step": 22778
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3035241111973851e-06,
+      "loss": 0.1658,
+      "step": 22779
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3029341559622465e-06,
+      "loss": 0.1737,
+      "step": 22780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3023443249562518e-06,
+      "loss": 0.1746,
+      "step": 22781
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3017546181878282e-06,
+      "loss": 0.1848,
+      "step": 22782
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3011650356653959e-06,
+      "loss": 0.1735,
+      "step": 22783
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3005755773973805e-06,
+      "loss": 0.1628,
+      "step": 22784
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2999862433922006e-06,
+      "loss": 0.179,
+      "step": 22785
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2993970336582718e-06,
+      "loss": 0.1862,
+      "step": 22786
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2988079482040161e-06,
+      "loss": 0.169,
+      "step": 22787
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2982189870378404e-06,
+      "loss": 0.1655,
+      "step": 22788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.297630150168162e-06,
+      "loss": 0.1752,
+      "step": 22789
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2970414376033902e-06,
+      "loss": 0.2035,
+      "step": 22790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2964528493519368e-06,
+      "loss": 0.1767,
+      "step": 22791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2958643854222074e-06,
+      "loss": 0.1689,
+      "step": 22792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2952760458226055e-06,
+      "loss": 0.1829,
+      "step": 22793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2946878305615396e-06,
+      "loss": 0.1817,
+      "step": 22794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2940997396474087e-06,
+      "loss": 0.194,
+      "step": 22795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2935117730886126e-06,
+      "loss": 0.1783,
+      "step": 22796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2929239308935526e-06,
+      "loss": 0.1867,
+      "step": 22797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2923362130706208e-06,
+      "loss": 0.1685,
+      "step": 22798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.291748619628218e-06,
+      "loss": 0.169,
+      "step": 22799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2911611505747357e-06,
+      "loss": 0.1904,
+      "step": 22800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2905738059185625e-06,
+      "loss": 0.1587,
+      "step": 22801
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2899865856680903e-06,
+      "loss": 0.1823,
+      "step": 22802
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2893994898317053e-06,
+      "loss": 0.1767,
+      "step": 22803
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2888125184177958e-06,
+      "loss": 0.1872,
+      "step": 22804
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.288225671434743e-06,
+      "loss": 0.16,
+      "step": 22805
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2876389488909347e-06,
+      "loss": 0.1777,
+      "step": 22806
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2870523507947474e-06,
+      "loss": 0.1817,
+      "step": 22807
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.286465877154559e-06,
+      "loss": 0.1847,
+      "step": 22808
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2858795279787517e-06,
+      "loss": 0.1912,
+      "step": 22809
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2852933032756954e-06,
+      "loss": 0.1906,
+      "step": 22810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2847072030537687e-06,
+      "loss": 0.1775,
+      "step": 22811
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.284121227321341e-06,
+      "loss": 0.1744,
+      "step": 22812
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.283535376086783e-06,
+      "loss": 0.1494,
+      "step": 22813
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2829496493584625e-06,
+      "loss": 0.1667,
+      "step": 22814
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.282364047144744e-06,
+      "loss": 0.1784,
+      "step": 22815
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2817785694539974e-06,
+      "loss": 0.1902,
+      "step": 22816
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2811932162945795e-06,
+      "loss": 0.1735,
+      "step": 22817
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2806079876748567e-06,
+      "loss": 0.1863,
+      "step": 22818
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.280022883603187e-06,
+      "loss": 0.1972,
+      "step": 22819
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2794379040879257e-06,
+      "loss": 0.176,
+      "step": 22820
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2788530491374317e-06,
+      "loss": 0.1733,
+      "step": 22821
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2782683187600553e-06,
+      "loss": 0.1625,
+      "step": 22822
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.277683712964155e-06,
+      "loss": 0.149,
+      "step": 22823
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2770992317580766e-06,
+      "loss": 0.1646,
+      "step": 22824
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2765148751501678e-06,
+      "loss": 0.1633,
+      "step": 22825
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2759306431487818e-06,
+      "loss": 0.1784,
+      "step": 22826
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2753465357622564e-06,
+      "loss": 0.1835,
+      "step": 22827
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2747625529989394e-06,
+      "loss": 0.1848,
+      "step": 22828
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2741786948671697e-06,
+      "loss": 0.1698,
+      "step": 22829
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2735949613752907e-06,
+      "loss": 0.1711,
+      "step": 22830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.273011352531638e-06,
+      "loss": 0.1852,
+      "step": 22831
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2724278683445467e-06,
+      "loss": 0.1525,
+      "step": 22832
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2718445088223552e-06,
+      "loss": 0.1533,
+      "step": 22833
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2712612739733942e-06,
+      "loss": 0.1891,
+      "step": 22834
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2706781638059917e-06,
+      "loss": 0.1807,
+      "step": 22835
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2700951783284821e-06,
+      "loss": 0.1886,
+      "step": 22836
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2695123175491886e-06,
+      "loss": 0.1753,
+      "step": 22837
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2689295814764402e-06,
+      "loss": 0.1941,
+      "step": 22838
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.268346970118558e-06,
+      "loss": 0.1902,
+      "step": 22839
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2677644834838665e-06,
+      "loss": 0.1668,
+      "step": 22840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2671821215806845e-06,
+      "loss": 0.1664,
+      "step": 22841
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2665998844173277e-06,
+      "loss": 0.1729,
+      "step": 22842
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2660177720021183e-06,
+      "loss": 0.1667,
+      "step": 22843
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2654357843433662e-06,
+      "loss": 0.1654,
+      "step": 22844
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2648539214493882e-06,
+      "loss": 0.1835,
+      "step": 22845
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2642721833284954e-06,
+      "loss": 0.1768,
+      "step": 22846
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2636905699889934e-06,
+      "loss": 0.1761,
+      "step": 22847
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2631090814391945e-06,
+      "loss": 0.1639,
+      "step": 22848
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2625277176874018e-06,
+      "loss": 0.16,
+      "step": 22849
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2619464787419223e-06,
+      "loss": 0.1926,
+      "step": 22850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2613653646110579e-06,
+      "loss": 0.1736,
+      "step": 22851
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2607843753031067e-06,
+      "loss": 0.1747,
+      "step": 22852
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2602035108263698e-06,
+      "loss": 0.188,
+      "step": 22853
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2596227711891417e-06,
+      "loss": 0.1641,
+      "step": 22854
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2590421563997224e-06,
+      "loss": 0.1793,
+      "step": 22855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2584616664663996e-06,
+      "loss": 0.1715,
+      "step": 22856
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.25788130139747e-06,
+      "loss": 0.1833,
+      "step": 22857
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2573010612012215e-06,
+      "loss": 0.1675,
+      "step": 22858
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2567209458859397e-06,
+      "loss": 0.1922,
+      "step": 22859
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2561409554599168e-06,
+      "loss": 0.1708,
+      "step": 22860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2555610899314318e-06,
+      "loss": 0.1636,
+      "step": 22861
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2549813493087726e-06,
+      "loss": 0.1789,
+      "step": 22862
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2544017336002167e-06,
+      "loss": 0.1963,
+      "step": 22863
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2538222428140423e-06,
+      "loss": 0.1807,
+      "step": 22864
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2532428769585315e-06,
+      "loss": 0.1723,
+      "step": 22865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2526636360419564e-06,
+      "loss": 0.1714,
+      "step": 22866
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2520845200725929e-06,
+      "loss": 0.1679,
+      "step": 22867
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2515055290587109e-06,
+      "loss": 0.1881,
+      "step": 22868
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2509266630085802e-06,
+      "loss": 0.1798,
+      "step": 22869
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2503479219304727e-06,
+      "loss": 0.1659,
+      "step": 22870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249769305832652e-06,
+      "loss": 0.1896,
+      "step": 22871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2491908147233866e-06,
+      "loss": 0.1659,
+      "step": 22872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2486124486109363e-06,
+      "loss": 0.1803,
+      "step": 22873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2480342075035622e-06,
+      "loss": 0.1633,
+      "step": 22874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.247456091409528e-06,
+      "loss": 0.1626,
+      "step": 22875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2468781003370868e-06,
+      "loss": 0.1901,
+      "step": 22876
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2463002342944986e-06,
+      "loss": 0.1657,
+      "step": 22877
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2457224932900168e-06,
+      "loss": 0.1699,
+      "step": 22878
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2451448773318919e-06,
+      "loss": 0.1915,
+      "step": 22879
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2445673864283759e-06,
+      "loss": 0.1676,
+      "step": 22880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2439900205877154e-06,
+      "loss": 0.1908,
+      "step": 22881
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2434127798181617e-06,
+      "loss": 0.1915,
+      "step": 22882
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.242835664127956e-06,
+      "loss": 0.1764,
+      "step": 22883
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2422586735253462e-06,
+      "loss": 0.1754,
+      "step": 22884
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2416818080185712e-06,
+      "loss": 0.1749,
+      "step": 22885
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2411050676158687e-06,
+      "loss": 0.1703,
+      "step": 22886
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2405284523254823e-06,
+      "loss": 0.1893,
+      "step": 22887
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2399519621556444e-06,
+      "loss": 0.2036,
+      "step": 22888
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2393755971145915e-06,
+      "loss": 0.183,
+      "step": 22889
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2387993572105571e-06,
+      "loss": 0.1634,
+      "step": 22890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.238223242451768e-06,
+      "loss": 0.1673,
+      "step": 22891
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.237647252846461e-06,
+      "loss": 0.1776,
+      "step": 22892
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2370713884028541e-06,
+      "loss": 0.1717,
+      "step": 22893
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2364956491291803e-06,
+      "loss": 0.1783,
+      "step": 22894
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2359200350336599e-06,
+      "loss": 0.1859,
+      "step": 22895
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2353445461245162e-06,
+      "loss": 0.1772,
+      "step": 22896
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2347691824099705e-06,
+      "loss": 0.1793,
+      "step": 22897
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2341939438982388e-06,
+      "loss": 0.166,
+      "step": 22898
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2336188305975394e-06,
+      "loss": 0.1874,
+      "step": 22899
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2330438425160884e-06,
+      "loss": 0.1796,
+      "step": 22900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2324689796620948e-06,
+      "loss": 0.1778,
+      "step": 22901
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2318942420437752e-06,
+      "loss": 0.1747,
+      "step": 22902
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2313196296693353e-06,
+      "loss": 0.1832,
+      "step": 22903
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2307451425469853e-06,
+      "loss": 0.1788,
+      "step": 22904
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2301707806849305e-06,
+      "loss": 0.1761,
+      "step": 22905
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.229596544091375e-06,
+      "loss": 0.1814,
+      "step": 22906
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2290224327745204e-06,
+      "loss": 0.1815,
+      "step": 22907
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2284484467425662e-06,
+      "loss": 0.1568,
+      "step": 22908
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2278745860037156e-06,
+      "loss": 0.1847,
+      "step": 22909
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2273008505661598e-06,
+      "loss": 0.1844,
+      "step": 22910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2267272404381002e-06,
+      "loss": 0.17,
+      "step": 22911
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.226153755627727e-06,
+      "loss": 0.1818,
+      "step": 22912
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2255803961432299e-06,
+      "loss": 0.1858,
+      "step": 22913
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2250071619928039e-06,
+      "loss": 0.1764,
+      "step": 22914
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.224434053184631e-06,
+      "loss": 0.1971,
+      "step": 22915
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2238610697269037e-06,
+      "loss": 0.171,
+      "step": 22916
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2232882116278034e-06,
+      "loss": 0.188,
+      "step": 22917
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.222715478895513e-06,
+      "loss": 0.1835,
+      "step": 22918
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2221428715382133e-06,
+      "loss": 0.1649,
+      "step": 22919
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2215703895640817e-06,
+      "loss": 0.1936,
+      "step": 22920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2209980329812988e-06,
+      "loss": 0.1947,
+      "step": 22921
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2204258017980375e-06,
+      "loss": 0.1595,
+      "step": 22922
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2198536960224749e-06,
+      "loss": 0.1772,
+      "step": 22923
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2192817156627802e-06,
+      "loss": 0.1767,
+      "step": 22924
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.218709860727123e-06,
+      "loss": 0.1623,
+      "step": 22925
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2181381312236751e-06,
+      "loss": 0.1706,
+      "step": 22926
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2175665271605985e-06,
+      "loss": 0.186,
+      "step": 22927
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2169950485460624e-06,
+      "loss": 0.1673,
+      "step": 22928
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.216423695388228e-06,
+      "loss": 0.1706,
+      "step": 22929
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2158524676952543e-06,
+      "loss": 0.1813,
+      "step": 22930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2152813654753082e-06,
+      "loss": 0.165,
+      "step": 22931
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2147103887365364e-06,
+      "loss": 0.1695,
+      "step": 22932
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2141395374871033e-06,
+      "loss": 0.1674,
+      "step": 22933
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2135688117351574e-06,
+      "loss": 0.1729,
+      "step": 22934
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.212998211488856e-06,
+      "loss": 0.1555,
+      "step": 22935
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2124277367563475e-06,
+      "loss": 0.1687,
+      "step": 22936
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2118573875457773e-06,
+      "loss": 0.1562,
+      "step": 22937
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2112871638652978e-06,
+      "loss": 0.1906,
+      "step": 22938
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2107170657230515e-06,
+      "loss": 0.1621,
+      "step": 22939
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2101470931271798e-06,
+      "loss": 0.157,
+      "step": 22940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.209577246085828e-06,
+      "loss": 0.1674,
+      "step": 22941
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.209007524607132e-06,
+      "loss": 0.1718,
+      "step": 22942
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2084379286992332e-06,
+      "loss": 0.1619,
+      "step": 22943
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2078684583702672e-06,
+      "loss": 0.1784,
+      "step": 22944
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2072991136283662e-06,
+      "loss": 0.1518,
+      "step": 22945
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2067298944816651e-06,
+      "loss": 0.1824,
+      "step": 22946
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2061608009382896e-06,
+      "loss": 0.1824,
+      "step": 22947
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2055918330063765e-06,
+      "loss": 0.1624,
+      "step": 22948
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2050229906940447e-06,
+      "loss": 0.183,
+      "step": 22949
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.204454274009428e-06,
+      "loss": 0.1924,
+      "step": 22950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2038856829606438e-06,
+      "loss": 0.1837,
+      "step": 22951
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.203317217555815e-06,
+      "loss": 0.185,
+      "step": 22952
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2027488778030638e-06,
+      "loss": 0.1779,
+      "step": 22953
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.202180663710505e-06,
+      "loss": 0.1863,
+      "step": 22954
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2016125752862595e-06,
+      "loss": 0.196,
+      "step": 22955
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.201044612538439e-06,
+      "loss": 0.1826,
+      "step": 22956
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2004767754751534e-06,
+      "loss": 0.1643,
+      "step": 22957
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.199909064104523e-06,
+      "loss": 0.1597,
+      "step": 22958
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1993414784346459e-06,
+      "loss": 0.1888,
+      "step": 22959
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1987740184736363e-06,
+      "loss": 0.186,
+      "step": 22960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1982066842295958e-06,
+      "loss": 0.1874,
+      "step": 22961
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1976394757106335e-06,
+      "loss": 0.1746,
+      "step": 22962
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1970723929248473e-06,
+      "loss": 0.1719,
+      "step": 22963
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1965054358803375e-06,
+      "loss": 0.1571,
+      "step": 22964
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.195938604585205e-06,
+      "loss": 0.1607,
+      "step": 22965
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1953718990475438e-06,
+      "loss": 0.173,
+      "step": 22966
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1948053192754506e-06,
+      "loss": 0.1855,
+      "step": 22967
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1942388652770187e-06,
+      "loss": 0.1701,
+      "step": 22968
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1936725370603364e-06,
+      "loss": 0.1707,
+      "step": 22969
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.193106334633497e-06,
+      "loss": 0.1731,
+      "step": 22970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1925402580045864e-06,
+      "loss": 0.1909,
+      "step": 22971
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1919743071816903e-06,
+      "loss": 0.1573,
+      "step": 22972
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1914084821728921e-06,
+      "loss": 0.1808,
+      "step": 22973
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1908427829862734e-06,
+      "loss": 0.1782,
+      "step": 22974
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1902772096299186e-06,
+      "loss": 0.1776,
+      "step": 22975
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1897117621119014e-06,
+      "loss": 0.1912,
+      "step": 22976
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.189146440440303e-06,
+      "loss": 0.196,
+      "step": 22977
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1885812446231959e-06,
+      "loss": 0.1771,
+      "step": 22978
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1880161746686536e-06,
+      "loss": 0.1576,
+      "step": 22979
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1874512305847485e-06,
+      "loss": 0.1771,
+      "step": 22980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1868864123795488e-06,
+      "loss": 0.186,
+      "step": 22981
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1863217200611244e-06,
+      "loss": 0.1804,
+      "step": 22982
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1857571536375412e-06,
+      "loss": 0.1743,
+      "step": 22983
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1851927131168628e-06,
+      "loss": 0.1804,
+      "step": 22984
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1846283985071505e-06,
+      "loss": 0.1959,
+      "step": 22985
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1840642098164645e-06,
+      "loss": 0.1705,
+      "step": 22986
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1835001470528673e-06,
+      "loss": 0.1781,
+      "step": 22987
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1829362102244125e-06,
+      "loss": 0.1726,
+      "step": 22988
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1823723993391577e-06,
+      "loss": 0.169,
+      "step": 22989
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1818087144051559e-06,
+      "loss": 0.1739,
+      "step": 22990
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.181245155430457e-06,
+      "loss": 0.192,
+      "step": 22991
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1806817224231137e-06,
+      "loss": 0.1944,
+      "step": 22992
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1801184153911705e-06,
+      "loss": 0.1824,
+      "step": 22993
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1795552343426775e-06,
+      "loss": 0.176,
+      "step": 22994
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1789921792856783e-06,
+      "loss": 0.1953,
+      "step": 22995
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1784292502282125e-06,
+      "loss": 0.1875,
+      "step": 22996
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1778664471783275e-06,
+      "loss": 0.1947,
+      "step": 22997
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1773037701440537e-06,
+      "loss": 0.1803,
+      "step": 22998
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1767412191334348e-06,
+      "loss": 0.1867,
+      "step": 22999
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1761787941545033e-06,
+      "loss": 0.1602,
+      "step": 23000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.175616495215296e-06,
+      "loss": 0.1866,
+      "step": 23001
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.175054322323842e-06,
+      "loss": 0.1678,
+      "step": 23002
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1744922754881705e-06,
+      "loss": 0.1987,
+      "step": 23003
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1739303547163138e-06,
+      "loss": 0.1827,
+      "step": 23004
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1733685600162958e-06,
+      "loss": 0.1693,
+      "step": 23005
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1728068913961388e-06,
+      "loss": 0.1644,
+      "step": 23006
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.172245348863872e-06,
+      "loss": 0.1903,
+      "step": 23007
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1716839324275086e-06,
+      "loss": 0.1774,
+      "step": 23008
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1711226420950761e-06,
+      "loss": 0.1684,
+      "step": 23009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1705614778745878e-06,
+      "loss": 0.2018,
+      "step": 23010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1700004397740583e-06,
+      "loss": 0.163,
+      "step": 23011
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1694395278015048e-06,
+      "loss": 0.1886,
+      "step": 23012
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1688787419649338e-06,
+      "loss": 0.1743,
+      "step": 23013
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1683180822723616e-06,
+      "loss": 0.1812,
+      "step": 23014
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1677575487317916e-06,
+      "loss": 0.1816,
+      "step": 23015
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1671971413512361e-06,
+      "loss": 0.1847,
+      "step": 23016
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1666368601386968e-06,
+      "loss": 0.1626,
+      "step": 23017
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1660767051021749e-06,
+      "loss": 0.1803,
+      "step": 23018
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1655166762496761e-06,
+      "loss": 0.1637,
+      "step": 23019
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1649567735891943e-06,
+      "loss": 0.1952,
+      "step": 23020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1643969971287338e-06,
+      "loss": 0.1742,
+      "step": 23021
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1638373468762864e-06,
+      "loss": 0.1859,
+      "step": 23022
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1632778228398446e-06,
+      "loss": 0.1676,
+      "step": 23023
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1627184250274072e-06,
+      "loss": 0.1894,
+      "step": 23024
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1621591534469567e-06,
+      "loss": 0.1813,
+      "step": 23025
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.161600008106486e-06,
+      "loss": 0.1702,
+      "step": 23026
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1610409890139806e-06,
+      "loss": 0.1709,
+      "step": 23027
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1604820961774278e-06,
+      "loss": 0.1687,
+      "step": 23028
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1599233296048096e-06,
+      "loss": 0.1802,
+      "step": 23029
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1593646893041043e-06,
+      "loss": 0.1874,
+      "step": 23030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1588061752832979e-06,
+      "loss": 0.1685,
+      "step": 23031
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1582477875503618e-06,
+      "loss": 0.1789,
+      "step": 23032
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.157689526113276e-06,
+      "loss": 0.1662,
+      "step": 23033
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1571313909800153e-06,
+      "loss": 0.1797,
+      "step": 23034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.156573382158548e-06,
+      "loss": 0.1781,
+      "step": 23035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1560154996568485e-06,
+      "loss": 0.1709,
+      "step": 23036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1554577434828851e-06,
+      "loss": 0.202,
+      "step": 23037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1549001136446237e-06,
+      "loss": 0.1851,
+      "step": 23038
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1543426101500266e-06,
+      "loss": 0.1691,
+      "step": 23039
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1537852330070642e-06,
+      "loss": 0.1701,
+      "step": 23040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1532279822236937e-06,
+      "loss": 0.181,
+      "step": 23041
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1526708578078728e-06,
+      "loss": 0.1787,
+      "step": 23042
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.152113859767565e-06,
+      "loss": 0.1982,
+      "step": 23043
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1515569881107237e-06,
+      "loss": 0.1603,
+      "step": 23044
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1510002428453004e-06,
+      "loss": 0.1931,
+      "step": 23045
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1504436239792537e-06,
+      "loss": 0.1786,
+      "step": 23046
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1498871315205284e-06,
+      "loss": 0.1644,
+      "step": 23047
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149330765477078e-06,
+      "loss": 0.1926,
+      "step": 23048
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1487745258568483e-06,
+      "loss": 0.166,
+      "step": 23049
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1482184126677832e-06,
+      "loss": 0.1702,
+      "step": 23050
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1476624259178282e-06,
+      "loss": 0.1813,
+      "step": 23051
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1471065656149216e-06,
+      "loss": 0.1875,
+      "step": 23052
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1465508317670082e-06,
+      "loss": 0.1664,
+      "step": 23053
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1459952243820216e-06,
+      "loss": 0.2016,
+      "step": 23054
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 0.1738,
+      "step": 23055
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1448843890325822e-06,
+      "loss": 0.1792,
+      "step": 23056
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1443291610839924e-06,
+      "loss": 0.1791,
+      "step": 23057
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1437740596300684e-06,
+      "loss": 0.1569,
+      "step": 23058
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.143219084678735e-06,
+      "loss": 0.1643,
+      "step": 23059
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1426642362379226e-06,
+      "loss": 0.1699,
+      "step": 23060
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1421095143155558e-06,
+      "loss": 0.1658,
+      "step": 23061
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1415549189195552e-06,
+      "loss": 0.1756,
+      "step": 23062
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1410004500578497e-06,
+      "loss": 0.1954,
+      "step": 23063
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.140446107738351e-06,
+      "loss": 0.1755,
+      "step": 23064
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1398918919689839e-06,
+      "loss": 0.1862,
+      "step": 23065
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1393378027576584e-06,
+      "loss": 0.1859,
+      "step": 23066
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1387838401122963e-06,
+      "loss": 0.1792,
+      "step": 23067
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1382300040408068e-06,
+      "loss": 0.1692,
+      "step": 23068
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1376762945510988e-06,
+      "loss": 0.1758,
+      "step": 23069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1371227116510863e-06,
+      "loss": 0.1787,
+      "step": 23070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1365692553486718e-06,
+      "loss": 0.1971,
+      "step": 23071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1360159256517655e-06,
+      "loss": 0.1574,
+      "step": 23072
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1354627225682702e-06,
+      "loss": 0.1755,
+      "step": 23073
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1349096461060826e-06,
+      "loss": 0.1886,
+      "step": 23074
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.13435669627311e-06,
+      "loss": 0.1703,
+      "step": 23075
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1338038730772484e-06,
+      "loss": 0.1673,
+      "step": 23076
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.133251176526392e-06,
+      "loss": 0.1817,
+      "step": 23077
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1326986066284385e-06,
+      "loss": 0.2063,
+      "step": 23078
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1321461633912767e-06,
+      "loss": 0.1878,
+      "step": 23079
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1315938468228027e-06,
+      "loss": 0.1551,
+      "step": 23080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1310416569309012e-06,
+      "loss": 0.1641,
+      "step": 23081
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.130489593723465e-06,
+      "loss": 0.1597,
+      "step": 23082
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1299376572083766e-06,
+      "loss": 0.1749,
+      "step": 23083
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.129385847393517e-06,
+      "loss": 0.1784,
+      "step": 23084
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.128834164286775e-06,
+      "loss": 0.1645,
+      "step": 23085
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.128282607896025e-06,
+      "loss": 0.1798,
+      "step": 23086
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.127731178229149e-06,
+      "loss": 0.1667,
+      "step": 23087
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1271798752940233e-06,
+      "loss": 0.18,
+      "step": 23088
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.126628699098521e-06,
+      "loss": 0.1758,
+      "step": 23089
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1260776496505167e-06,
+      "loss": 0.1649,
+      "step": 23090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1255267269578785e-06,
+      "loss": 0.1635,
+      "step": 23091
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1249759310284792e-06,
+      "loss": 0.1804,
+      "step": 23092
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1244252618701845e-06,
+      "loss": 0.1896,
+      "step": 23093
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1238747194908628e-06,
+      "loss": 0.1717,
+      "step": 23094
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1233243038983755e-06,
+      "loss": 0.1616,
+      "step": 23095
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1227740151005838e-06,
+      "loss": 0.1747,
+      "step": 23096
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1222238531053532e-06,
+      "loss": 0.1706,
+      "step": 23097
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1216738179205344e-06,
+      "loss": 0.1799,
+      "step": 23098
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1211239095539928e-06,
+      "loss": 0.175,
+      "step": 23099
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1205741280135774e-06,
+      "loss": 0.1721,
+      "step": 23100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1200244733071407e-06,
+      "loss": 0.171,
+      "step": 23101
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.119474945442539e-06,
+      "loss": 0.1853,
+      "step": 23102
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1189255444276192e-06,
+      "loss": 0.1681,
+      "step": 23103
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1183762702702283e-06,
+      "loss": 0.1733,
+      "step": 23104
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.11782712297821e-06,
+      "loss": 0.1984,
+      "step": 23105
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1172781025594137e-06,
+      "loss": 0.1818,
+      "step": 23106
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1167292090216785e-06,
+      "loss": 0.1846,
+      "step": 23107
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.116180442372844e-06,
+      "loss": 0.1786,
+      "step": 23108
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1156318026207514e-06,
+      "loss": 0.1845,
+      "step": 23109
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1150832897732333e-06,
+      "loss": 0.191,
+      "step": 23110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1145349038381293e-06,
+      "loss": 0.1693,
+      "step": 23111
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1139866448232716e-06,
+      "loss": 0.1787,
+      "step": 23112
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1134385127364877e-06,
+      "loss": 0.1798,
+      "step": 23113
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1128905075856133e-06,
+      "loss": 0.2055,
+      "step": 23114
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1123426293784722e-06,
+      "loss": 0.1822,
+      "step": 23115
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1117948781228904e-06,
+      "loss": 0.1657,
+      "step": 23116
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1112472538266928e-06,
+      "loss": 0.1753,
+      "step": 23117
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1106997564976996e-06,
+      "loss": 0.1589,
+      "step": 23118
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1101523861437358e-06,
+      "loss": 0.1605,
+      "step": 23119
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.109605142772614e-06,
+      "loss": 0.1761,
+      "step": 23120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.109058026392158e-06,
+      "loss": 0.1643,
+      "step": 23121
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1085110370101781e-06,
+      "loss": 0.1698,
+      "step": 23122
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.107964174634487e-06,
+      "loss": 0.1843,
+      "step": 23123
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1074174392728999e-06,
+      "loss": 0.1869,
+      "step": 23124
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.106870830933221e-06,
+      "loss": 0.1912,
+      "step": 23125
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1063243496232644e-06,
+      "loss": 0.1799,
+      "step": 23126
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.105777995350833e-06,
+      "loss": 0.1703,
+      "step": 23127
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.105231768123729e-06,
+      "loss": 0.1612,
+      "step": 23128
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1046856679497598e-06,
+      "loss": 0.1863,
+      "step": 23129
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.10413969483672e-06,
+      "loss": 0.1719,
+      "step": 23130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1035938487924114e-06,
+      "loss": 0.1856,
+      "step": 23131
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1030481298246298e-06,
+      "loss": 0.1833,
+      "step": 23132
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1025025379411724e-06,
+      "loss": 0.1828,
+      "step": 23133
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1019570731498308e-06,
+      "loss": 0.1744,
+      "step": 23134
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1014117354583952e-06,
+      "loss": 0.1626,
+      "step": 23135
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1008665248746585e-06,
+      "loss": 0.1879,
+      "step": 23136
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1003214414064045e-06,
+      "loss": 0.1625,
+      "step": 23137
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0997764850614234e-06,
+      "loss": 0.1689,
+      "step": 23138
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0992316558474981e-06,
+      "loss": 0.1801,
+      "step": 23139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0986869537724076e-06,
+      "loss": 0.193,
+      "step": 23140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0981423788439372e-06,
+      "loss": 0.1943,
+      "step": 23141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0975979310698625e-06,
+      "loss": 0.1797,
+      "step": 23142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.097053610457962e-06,
+      "loss": 0.1724,
+      "step": 23143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0965094170160085e-06,
+      "loss": 0.1888,
+      "step": 23144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0959653507517786e-06,
+      "loss": 0.1733,
+      "step": 23145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0954214116730422e-06,
+      "loss": 0.1881,
+      "step": 23146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.094877599787567e-06,
+      "loss": 0.1657,
+      "step": 23147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0943339151031251e-06,
+      "loss": 0.1902,
+      "step": 23148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.093790357627479e-06,
+      "loss": 0.1598,
+      "step": 23149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.093246927368392e-06,
+      "loss": 0.1717,
+      "step": 23150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.092703624333632e-06,
+      "loss": 0.1745,
+      "step": 23151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0921604485309522e-06,
+      "loss": 0.1795,
+      "step": 23152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.091617399968118e-06,
+      "loss": 0.1883,
+      "step": 23153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0910744786528837e-06,
+      "loss": 0.1931,
+      "step": 23154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0905316845930037e-06,
+      "loss": 0.1897,
+      "step": 23155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0899890177962313e-06,
+      "loss": 0.1841,
+      "step": 23156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0894464782703174e-06,
+      "loss": 0.1747,
+      "step": 23157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0889040660230132e-06,
+      "loss": 0.1685,
+      "step": 23158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.088361781062065e-06,
+      "loss": 0.1868,
+      "step": 23159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.087819623395222e-06,
+      "loss": 0.1622,
+      "step": 23160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.087277593030226e-06,
+      "loss": 0.1903,
+      "step": 23161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.086735689974817e-06,
+      "loss": 0.1925,
+      "step": 23162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0861939142367417e-06,
+      "loss": 0.1695,
+      "step": 23163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0856522658237313e-06,
+      "loss": 0.1664,
+      "step": 23164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0851107447435306e-06,
+      "loss": 0.1791,
+      "step": 23165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.084569351003869e-06,
+      "loss": 0.173,
+      "step": 23166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0840280846124806e-06,
+      "loss": 0.171,
+      "step": 23167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0834869455771014e-06,
+      "loss": 0.1874,
+      "step": 23168
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0829459339054538e-06,
+      "loss": 0.1937,
+      "step": 23169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0824050496052717e-06,
+      "loss": 0.1785,
+      "step": 23170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.081864292684276e-06,
+      "loss": 0.1968,
+      "step": 23171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0813236631501966e-06,
+      "loss": 0.1699,
+      "step": 23172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0807831610107521e-06,
+      "loss": 0.1774,
+      "step": 23173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.080242786273662e-06,
+      "loss": 0.1837,
+      "step": 23174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0797025389466486e-06,
+      "loss": 0.1755,
+      "step": 23175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.079162419037426e-06,
+      "loss": 0.1727,
+      "step": 23176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0786224265537115e-06,
+      "loss": 0.1662,
+      "step": 23177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0780825615032175e-06,
+      "loss": 0.1932,
+      "step": 23178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0775428238936524e-06,
+      "loss": 0.1799,
+      "step": 23179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0770032137327323e-06,
+      "loss": 0.1729,
+      "step": 23180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.07646373102816e-06,
+      "loss": 0.1859,
+      "step": 23181
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0759243757876436e-06,
+      "loss": 0.1714,
+      "step": 23182
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.075385148018886e-06,
+      "loss": 0.1716,
+      "step": 23183
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0748460477295875e-06,
+      "loss": 0.1694,
+      "step": 23184
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0743070749274541e-06,
+      "loss": 0.1841,
+      "step": 23185
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.073768229620179e-06,
+      "loss": 0.1742,
+      "step": 23186
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0732295118154633e-06,
+      "loss": 0.1744,
+      "step": 23187
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0726909215210012e-06,
+      "loss": 0.1699,
+      "step": 23188
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.072152458744483e-06,
+      "loss": 0.179,
+      "step": 23189
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.071614123493604e-06,
+      "loss": 0.1797,
+      "step": 23190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0710759157760497e-06,
+      "loss": 0.1901,
+      "step": 23191
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0705378355995133e-06,
+      "loss": 0.1754,
+      "step": 23192
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0699998829716763e-06,
+      "loss": 0.171,
+      "step": 23193
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0694620579002234e-06,
+      "loss": 0.1793,
+      "step": 23194
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0689243603928413e-06,
+      "loss": 0.1725,
+      "step": 23195
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.068386790457202e-06,
+      "loss": 0.1719,
+      "step": 23196
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0678493481009922e-06,
+      "loss": 0.1879,
+      "step": 23197
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0673120333318832e-06,
+      "loss": 0.1778,
+      "step": 23198
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0667748461575544e-06,
+      "loss": 0.1796,
+      "step": 23199
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0662377865856777e-06,
+      "loss": 0.1858,
+      "step": 23200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0657008546239212e-06,
+      "loss": 0.1715,
+      "step": 23201
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0651640502799588e-06,
+      "loss": 0.1732,
+      "step": 23202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0646273735614543e-06,
+      "loss": 0.1919,
+      "step": 23203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0640908244760783e-06,
+      "loss": 0.1709,
+      "step": 23204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0635544030314926e-06,
+      "loss": 0.1757,
+      "step": 23205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0630181092353563e-06,
+      "loss": 0.163,
+      "step": 23206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.062481943095336e-06,
+      "loss": 0.1534,
+      "step": 23207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.061945904619086e-06,
+      "loss": 0.174,
+      "step": 23208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0614099938142653e-06,
+      "loss": 0.1813,
+      "step": 23209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.060874210688525e-06,
+      "loss": 0.1719,
+      "step": 23210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0603385552495226e-06,
+      "loss": 0.1843,
+      "step": 23211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0598030275049088e-06,
+      "loss": 0.1745,
+      "step": 23212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0592676274623292e-06,
+      "loss": 0.19,
+      "step": 23213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.058732355129437e-06,
+      "loss": 0.1641,
+      "step": 23214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0581972105138728e-06,
+      "loss": 0.1718,
+      "step": 23215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0576621936232856e-06,
+      "loss": 0.1748,
+      "step": 23216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0571273044653152e-06,
+      "loss": 0.1627,
+      "step": 23217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0565925430475998e-06,
+      "loss": 0.1993,
+      "step": 23218
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0560579093777822e-06,
+      "loss": 0.1864,
+      "step": 23219
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0555234034634964e-06,
+      "loss": 0.1726,
+      "step": 23220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0549890253123785e-06,
+      "loss": 0.1746,
+      "step": 23221
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0544547749320622e-06,
+      "loss": 0.1677,
+      "step": 23222
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0539206523301738e-06,
+      "loss": 0.1759,
+      "step": 23223
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0533866575143493e-06,
+      "loss": 0.1797,
+      "step": 23224
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0528527904922114e-06,
+      "loss": 0.1652,
+      "step": 23225
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0523190512713898e-06,
+      "loss": 0.1828,
+      "step": 23226
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0517854398595072e-06,
+      "loss": 0.1686,
+      "step": 23227
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0512519562641832e-06,
+      "loss": 0.1735,
+      "step": 23228
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0507186004930414e-06,
+      "loss": 0.1801,
+      "step": 23229
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.050185372553697e-06,
+      "loss": 0.1803,
+      "step": 23230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0496522724537705e-06,
+      "loss": 0.1821,
+      "step": 23231
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.049119300200876e-06,
+      "loss": 0.1613,
+      "step": 23232
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0485864558026216e-06,
+      "loss": 0.1806,
+      "step": 23233
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0480537392666269e-06,
+      "loss": 0.1722,
+      "step": 23234
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0475211506004923e-06,
+      "loss": 0.1712,
+      "step": 23235
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.046988689811832e-06,
+      "loss": 0.169,
+      "step": 23236
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0464563569082465e-06,
+      "loss": 0.1819,
+      "step": 23237
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.045924151897344e-06,
+      "loss": 0.1892,
+      "step": 23238
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0453920747867253e-06,
+      "loss": 0.161,
+      "step": 23239
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0448601255839886e-06,
+      "loss": 0.1908,
+      "step": 23240
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0443283042967356e-06,
+      "loss": 0.1854,
+      "step": 23241
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0437966109325592e-06,
+      "loss": 0.1651,
+      "step": 23242
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0432650454990578e-06,
+      "loss": 0.1807,
+      "step": 23243
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042733608003822e-06,
+      "loss": 0.1722,
+      "step": 23244
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0422022984544423e-06,
+      "loss": 0.1811,
+      "step": 23245
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0416711168585102e-06,
+      "loss": 0.151,
+      "step": 23246
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0411400632236123e-06,
+      "loss": 0.1641,
+      "step": 23247
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0406091375573335e-06,
+      "loss": 0.1699,
+      "step": 23248
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0400783398672564e-06,
+      "loss": 0.1669,
+      "step": 23249
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0395476701609664e-06,
+      "loss": 0.1634,
+      "step": 23250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0390171284460416e-06,
+      "loss": 0.1771,
+      "step": 23251
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.038486714730058e-06,
+      "loss": 0.1848,
+      "step": 23252
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0379564290205957e-06,
+      "loss": 0.1861,
+      "step": 23253
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0374262713252292e-06,
+      "loss": 0.1643,
+      "step": 23254
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0368962416515272e-06,
+      "loss": 0.1684,
+      "step": 23255
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0363663400070655e-06,
+      "loss": 0.1693,
+      "step": 23256
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0358365663994097e-06,
+      "loss": 0.196,
+      "step": 23257
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.035306920836131e-06,
+      "loss": 0.1715,
+      "step": 23258
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0347774033247915e-06,
+      "loss": 0.1742,
+      "step": 23259
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0342480138729538e-06,
+      "loss": 0.1899,
+      "step": 23260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0337187524881864e-06,
+      "loss": 0.1589,
+      "step": 23261
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.033189619178041e-06,
+      "loss": 0.1733,
+      "step": 23262
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0326606139500805e-06,
+      "loss": 0.1847,
+      "step": 23263
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0321317368118588e-06,
+      "loss": 0.1672,
+      "step": 23264
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0316029877709333e-06,
+      "loss": 0.1676,
+      "step": 23265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0310743668348555e-06,
+      "loss": 0.1817,
+      "step": 23266
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0305458740111729e-06,
+      "loss": 0.1797,
+      "step": 23267
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0300175093074405e-06,
+      "loss": 0.1735,
+      "step": 23268
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0294892727312e-06,
+      "loss": 0.1737,
+      "step": 23269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.028961164290001e-06,
+      "loss": 0.162,
+      "step": 23270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0284331839913853e-06,
+      "loss": 0.1808,
+      "step": 23271
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0279053318428921e-06,
+      "loss": 0.1782,
+      "step": 23272
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0273776078520659e-06,
+      "loss": 0.1816,
+      "step": 23273
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0268500120264425e-06,
+      "loss": 0.1821,
+      "step": 23274
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0263225443735581e-06,
+      "loss": 0.1723,
+      "step": 23275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0257952049009444e-06,
+      "loss": 0.1707,
+      "step": 23276
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0252679936161392e-06,
+      "loss": 0.1685,
+      "step": 23277
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0247409105266703e-06,
+      "loss": 0.1817,
+      "step": 23278
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0242139556400642e-06,
+      "loss": 0.1695,
+      "step": 23279
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.023687128963854e-06,
+      "loss": 0.1875,
+      "step": 23280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0231604305055586e-06,
+      "loss": 0.1637,
+      "step": 23281
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.022633860272706e-06,
+      "loss": 0.1678,
+      "step": 23282
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0221074182728163e-06,
+      "loss": 0.1712,
+      "step": 23283
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.021581104513406e-06,
+      "loss": 0.188,
+      "step": 23284
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0210549190019991e-06,
+      "loss": 0.1736,
+      "step": 23285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0205288617461085e-06,
+      "loss": 0.1777,
+      "step": 23286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0200029327532467e-06,
+      "loss": 0.1654,
+      "step": 23287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0194771320309294e-06,
+      "loss": 0.1738,
+      "step": 23288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0189514595866623e-06,
+      "loss": 0.1785,
+      "step": 23289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0184259154279608e-06,
+      "loss": 0.1803,
+      "step": 23290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0179004995623242e-06,
+      "loss": 0.1832,
+      "step": 23291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0173752119972658e-06,
+      "loss": 0.1765,
+      "step": 23292
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0168500527402835e-06,
+      "loss": 0.1575,
+      "step": 23293
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0163250217988774e-06,
+      "loss": 0.185,
+      "step": 23294
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0158001191805523e-06,
+      "loss": 0.1717,
+      "step": 23295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.015275344892801e-06,
+      "loss": 0.1698,
+      "step": 23296
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0147506989431232e-06,
+      "loss": 0.1824,
+      "step": 23297
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.014226181339011e-06,
+      "loss": 0.1698,
+      "step": 23298
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0137017920879556e-06,
+      "loss": 0.1798,
+      "step": 23299
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0131775311974523e-06,
+      "loss": 0.1738,
+      "step": 23300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0126533986749821e-06,
+      "loss": 0.1662,
+      "step": 23301
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0121293945280375e-06,
+      "loss": 0.1747,
+      "step": 23302
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0116055187641005e-06,
+      "loss": 0.1582,
+      "step": 23303
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0110817713906562e-06,
+      "loss": 0.176,
+      "step": 23304
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0105581524151852e-06,
+      "loss": 0.1711,
+      "step": 23305
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0100346618451652e-06,
+      "loss": 0.1652,
+      "step": 23306
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0095112996880762e-06,
+      "loss": 0.1798,
+      "step": 23307
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0089880659513906e-06,
+      "loss": 0.1569,
+      "step": 23308
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0084649606425866e-06,
+      "loss": 0.1527,
+      "step": 23309
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0079419837691352e-06,
+      "loss": 0.1761,
+      "step": 23310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.007419135338502e-06,
+      "loss": 0.1832,
+      "step": 23311
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0068964153581618e-06,
+      "loss": 0.1753,
+      "step": 23312
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0063738238355792e-06,
+      "loss": 0.1848,
+      "step": 23313
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0058513607782161e-06,
+      "loss": 0.1831,
+      "step": 23314
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0053290261935366e-06,
+      "loss": 0.1574,
+      "step": 23315
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0048068200890037e-06,
+      "loss": 0.1735,
+      "step": 23316
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0042847424720758e-06,
+      "loss": 0.188,
+      "step": 23317
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0037627933502082e-06,
+      "loss": 0.1779,
+      "step": 23318
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.003240972730859e-06,
+      "loss": 0.1675,
+      "step": 23319
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0027192806214804e-06,
+      "loss": 0.1911,
+      "step": 23320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.002197717029526e-06,
+      "loss": 0.1858,
+      "step": 23321
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0016762819624436e-06,
+      "loss": 0.1893,
+      "step": 23322
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0011549754276816e-06,
+      "loss": 0.1778,
+      "step": 23323
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0006337974326897e-06,
+      "loss": 0.1759,
+      "step": 23324
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0001127479849082e-06,
+      "loss": 0.1788,
+      "step": 23325
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.995918270917826e-07,
+      "loss": 0.1642,
+      "step": 23326
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990710347607523e-07,
+      "loss": 0.1707,
+      "step": 23327
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98550370999255e-07,
+      "loss": 0.1812,
+      "step": 23328
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.980298358147323e-07,
+      "loss": 0.1954,
+      "step": 23329
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.975094292146138e-07,
+      "loss": 0.1688,
+      "step": 23330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96989151206338e-07,
+      "loss": 0.1618,
+      "step": 23331
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.964690017973344e-07,
+      "loss": 0.1628,
+      "step": 23332
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.959489809950317e-07,
+      "loss": 0.1875,
+      "step": 23333
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.954290888068608e-07,
+      "loss": 0.1823,
+      "step": 23334
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.949093252402432e-07,
+      "loss": 0.1818,
+      "step": 23335
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943896903026095e-07,
+      "loss": 0.1773,
+      "step": 23336
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.938701840013787e-07,
+      "loss": 0.1662,
+      "step": 23337
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93350806343969e-07,
+      "loss": 0.1796,
+      "step": 23338
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.928315573378044e-07,
+      "loss": 0.1941,
+      "step": 23339
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.923124369902993e-07,
+      "loss": 0.2021,
+      "step": 23340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.917934453088695e-07,
+      "loss": 0.1683,
+      "step": 23341
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.91274582300925e-07,
+      "loss": 0.165,
+      "step": 23342
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.907558479738822e-07,
+      "loss": 0.1769,
+      "step": 23343
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.902372423351481e-07,
+      "loss": 0.1718,
+      "step": 23344
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.897187653921293e-07,
+      "loss": 0.199,
+      "step": 23345
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.89200417152234e-07,
+      "loss": 0.1659,
+      "step": 23346
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.886821976228645e-07,
+      "loss": 0.1797,
+      "step": 23347
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.881641068114268e-07,
+      "loss": 0.1749,
+      "step": 23348
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.876461447253172e-07,
+      "loss": 0.165,
+      "step": 23349
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.871283113719342e-07,
+      "loss": 0.1699,
+      "step": 23350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.866106067586789e-07,
+      "loss": 0.1696,
+      "step": 23351
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.860930308929405e-07,
+      "loss": 0.1794,
+      "step": 23352
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.85575583782119e-07,
+      "loss": 0.1837,
+      "step": 23353
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.850582654335983e-07,
+      "loss": 0.1569,
+      "step": 23354
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.845410758547724e-07,
+      "loss": 0.1954,
+      "step": 23355
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.840240150530279e-07,
+      "loss": 0.1786,
+      "step": 23356
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.835070830357473e-07,
+      "loss": 0.1758,
+      "step": 23357
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.829902798103208e-07,
+      "loss": 0.1814,
+      "step": 23358
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.824736053841254e-07,
+      "loss": 0.1948,
+      "step": 23359
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.819570597645411e-07,
+      "loss": 0.1654,
+      "step": 23360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.814406429589506e-07,
+      "loss": 0.1852,
+      "step": 23361
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.80924354974726e-07,
+      "loss": 0.1783,
+      "step": 23362
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.804081958192458e-07,
+      "loss": 0.1837,
+      "step": 23363
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.798921654998794e-07,
+      "loss": 0.1726,
+      "step": 23364
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.79376264023999e-07,
+      "loss": 0.1652,
+      "step": 23365
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.788604913989762e-07,
+      "loss": 0.1877,
+      "step": 23366
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.783448476321733e-07,
+      "loss": 0.179,
+      "step": 23367
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.778293327309607e-07,
+      "loss": 0.1965,
+      "step": 23368
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.773139467026972e-07,
+      "loss": 0.1912,
+      "step": 23369
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.767986895547487e-07,
+      "loss": 0.1817,
+      "step": 23370
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.762835612944742e-07,
+      "loss": 0.1735,
+      "step": 23371
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.757685619292301e-07,
+      "loss": 0.1613,
+      "step": 23372
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.752536914663758e-07,
+      "loss": 0.1879,
+      "step": 23373
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.747389499132609e-07,
+      "loss": 0.1622,
+      "step": 23374
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.74224337277243e-07,
+      "loss": 0.2255,
+      "step": 23375
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.737098535656708e-07,
+      "loss": 0.1772,
+      "step": 23376
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.731954987858916e-07,
+      "loss": 0.1589,
+      "step": 23377
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.72681272945255e-07,
+      "loss": 0.182,
+      "step": 23378
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.721671760511065e-07,
+      "loss": 0.1888,
+      "step": 23379
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.716532081107877e-07,
+      "loss": 0.1821,
+      "step": 23380
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.711393691316383e-07,
+      "loss": 0.1657,
+      "step": 23381
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.706256591210017e-07,
+      "loss": 0.1895,
+      "step": 23382
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.70112078086215e-07,
+      "loss": 0.1855,
+      "step": 23383
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.695986260346124e-07,
+      "loss": 0.1829,
+      "step": 23384
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.690853029735303e-07,
+      "loss": 0.1923,
+      "step": 23385
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.685721089102984e-07,
+      "loss": 0.1766,
+      "step": 23386
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.680590438522497e-07,
+      "loss": 0.1841,
+      "step": 23387
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.675461078067128e-07,
+      "loss": 0.1813,
+      "step": 23388
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.670333007810117e-07,
+      "loss": 0.1626,
+      "step": 23389
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.665206227824742e-07,
+      "loss": 0.1903,
+      "step": 23390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66008073818423e-07,
+      "loss": 0.1746,
+      "step": 23391
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.65495653896179e-07,
+      "loss": 0.1716,
+      "step": 23392
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.649833630230609e-07,
+      "loss": 0.1581,
+      "step": 23393
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.64471201206385e-07,
+      "loss": 0.1634,
+      "step": 23394
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.63959168453471e-07,
+      "loss": 0.1772,
+      "step": 23395
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.634472647716287e-07,
+      "loss": 0.1717,
+      "step": 23396
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.629354901681732e-07,
+      "loss": 0.1783,
+      "step": 23397
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.624238446504141e-07,
+      "loss": 0.1783,
+      "step": 23398
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.61912328225657e-07,
+      "loss": 0.1925,
+      "step": 23399
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.614009409012126e-07,
+      "loss": 0.1773,
+      "step": 23400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.608896826843828e-07,
+      "loss": 0.1808,
+      "step": 23401
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.603785535824717e-07,
+      "loss": 0.1765,
+      "step": 23402
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.598675536027802e-07,
+      "loss": 0.1807,
+      "step": 23403
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.593566827526058e-07,
+      "loss": 0.1764,
+      "step": 23404
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.58845941039248e-07,
+      "loss": 0.1786,
+      "step": 23405
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.583353284700015e-07,
+      "loss": 0.172,
+      "step": 23406
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.578248450521598e-07,
+      "loss": 0.1904,
+      "step": 23407
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.573144907930132e-07,
+      "loss": 0.1735,
+      "step": 23408
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.568042656998534e-07,
+      "loss": 0.1744,
+      "step": 23409
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.5629416977997e-07,
+      "loss": 0.1963,
+      "step": 23410
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.557842030406438e-07,
+      "loss": 0.1861,
+      "step": 23411
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.55274365489165e-07,
+      "loss": 0.1789,
+      "step": 23412
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.547646571328118e-07,
+      "loss": 0.1793,
+      "step": 23413
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.542550779788695e-07,
+      "loss": 0.1705,
+      "step": 23414
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.537456280346136e-07,
+      "loss": 0.2013,
+      "step": 23415
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.532363073073203e-07,
+      "loss": 0.1834,
+      "step": 23416
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.527271158042672e-07,
+      "loss": 0.1789,
+      "step": 23417
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.522180535327252e-07,
+      "loss": 0.1768,
+      "step": 23418
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.517091204999718e-07,
+      "loss": 0.1769,
+      "step": 23419
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.512003167132678e-07,
+      "loss": 0.1779,
+      "step": 23420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.506916421798873e-07,
+      "loss": 0.2001,
+      "step": 23421
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.501830969070935e-07,
+      "loss": 0.1559,
+      "step": 23422
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.496746809021495e-07,
+      "loss": 0.1594,
+      "step": 23423
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.491663941723217e-07,
+      "loss": 0.1546,
+      "step": 23424
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.486582367248653e-07,
+      "loss": 0.1667,
+      "step": 23425
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481502085670435e-07,
+      "loss": 0.1916,
+      "step": 23426
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.476423097061105e-07,
+      "loss": 0.167,
+      "step": 23427
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.471345401493204e-07,
+      "loss": 0.1926,
+      "step": 23428
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.466268999039285e-07,
+      "loss": 0.1604,
+      "step": 23429
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.461193889771858e-07,
+      "loss": 0.1797,
+      "step": 23430
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.456120073763375e-07,
+      "loss": 0.18,
+      "step": 23431
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.45104755108639e-07,
+      "loss": 0.1501,
+      "step": 23432
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.445976321813277e-07,
+      "loss": 0.1943,
+      "step": 23433
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.440906386016513e-07,
+      "loss": 0.1833,
+      "step": 23434
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.435837743768505e-07,
+      "loss": 0.1605,
+      "step": 23435
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.430770395141664e-07,
+      "loss": 0.1771,
+      "step": 23436
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.425704340208386e-07,
+      "loss": 0.18,
+      "step": 23437
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.420639579040991e-07,
+      "loss": 0.1699,
+      "step": 23438
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.415576111711866e-07,
+      "loss": 0.1767,
+      "step": 23439
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.410513938293309e-07,
+      "loss": 0.1812,
+      "step": 23440
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.40545305885765e-07,
+      "loss": 0.1856,
+      "step": 23441
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.400393473477176e-07,
+      "loss": 0.1815,
+      "step": 23442
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.39533518222413e-07,
+      "loss": 0.1973,
+      "step": 23443
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.39027818517082e-07,
+      "loss": 0.1652,
+      "step": 23444
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.385222482389434e-07,
+      "loss": 0.1657,
+      "step": 23445
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.3801680739522e-07,
+      "loss": 0.1686,
+      "step": 23446
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.375114959931309e-07,
+      "loss": 0.1623,
+      "step": 23447
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.370063140398955e-07,
+      "loss": 0.1653,
+      "step": 23448
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.365012615427293e-07,
+      "loss": 0.1674,
+      "step": 23449
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.359963385088444e-07,
+      "loss": 0.1809,
+      "step": 23450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.354915449454571e-07,
+      "loss": 0.1738,
+      "step": 23451
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.34986880859774e-07,
+      "loss": 0.1651,
+      "step": 23452
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.344823462590069e-07,
+      "loss": 0.1748,
+      "step": 23453
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.339779411503614e-07,
+      "loss": 0.1575,
+      "step": 23454
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.334736655410404e-07,
+      "loss": 0.1768,
+      "step": 23455
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.329695194382505e-07,
+      "loss": 0.1878,
+      "step": 23456
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.324655028491891e-07,
+      "loss": 0.1691,
+      "step": 23457
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.319616157810618e-07,
+      "loss": 0.1806,
+      "step": 23458
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.314578582410593e-07,
+      "loss": 0.1719,
+      "step": 23459
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.309542302363806e-07,
+      "loss": 0.1742,
+      "step": 23460
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.304507317742195e-07,
+      "loss": 0.1714,
+      "step": 23461
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.299473628617661e-07,
+      "loss": 0.1789,
+      "step": 23462
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.294441235062135e-07,
+      "loss": 0.1886,
+      "step": 23463
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.289410137147481e-07,
+      "loss": 0.164,
+      "step": 23464
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.284380334945541e-07,
+      "loss": 0.1677,
+      "step": 23465
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.279351828528215e-07,
+      "loss": 0.1561,
+      "step": 23466
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.27432461796729e-07,
+      "loss": 0.1896,
+      "step": 23467
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.269298703334595e-07,
+      "loss": 0.1854,
+      "step": 23468
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.264274084701908e-07,
+      "loss": 0.169,
+      "step": 23469
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.259250762140992e-07,
+      "loss": 0.1694,
+      "step": 23470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254228735723648e-07,
+      "loss": 0.1625,
+      "step": 23471
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.249208005521538e-07,
+      "loss": 0.1691,
+      "step": 23472
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.244188571606438e-07,
+      "loss": 0.1848,
+      "step": 23473
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.239170434050004e-07,
+      "loss": 0.1568,
+      "step": 23474
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.234153592923956e-07,
+      "loss": 0.1757,
+      "step": 23475
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.229138048299924e-07,
+      "loss": 0.1726,
+      "step": 23476
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.224123800249551e-07,
+      "loss": 0.1785,
+      "step": 23477
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.219110848844482e-07,
+      "loss": 0.1647,
+      "step": 23478
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.21409919415629e-07,
+      "loss": 0.18,
+      "step": 23479
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.209088836256608e-07,
+      "loss": 0.1688,
+      "step": 23480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.204079775216967e-07,
+      "loss": 0.1859,
+      "step": 23481
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.199072011108922e-07,
+      "loss": 0.1947,
+      "step": 23482
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.194065544004016e-07,
+      "loss": 0.1753,
+      "step": 23483
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.189060373973735e-07,
+      "loss": 0.1675,
+      "step": 23484
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.184056501089633e-07,
+      "loss": 0.1801,
+      "step": 23485
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.179053925423109e-07,
+      "loss": 0.1713,
+      "step": 23486
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.174052647045684e-07,
+      "loss": 0.1449,
+      "step": 23487
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.169052666028755e-07,
+      "loss": 0.1831,
+      "step": 23488
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.164053982443743e-07,
+      "loss": 0.1615,
+      "step": 23489
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.159056596362071e-07,
+      "loss": 0.1747,
+      "step": 23490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.154060507855101e-07,
+      "loss": 0.1896,
+      "step": 23491
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.149065716994221e-07,
+      "loss": 0.1686,
+      "step": 23492
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.144072223850775e-07,
+      "loss": 0.1574,
+      "step": 23493
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.13908002849605e-07,
+      "loss": 0.164,
+      "step": 23494
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.134089131001412e-07,
+      "loss": 0.1814,
+      "step": 23495
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.129099531438124e-07,
+      "loss": 0.1748,
+      "step": 23496
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.124111229877431e-07,
+      "loss": 0.1807,
+      "step": 23497
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.11912422639063e-07,
+      "loss": 0.1711,
+      "step": 23498
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.114138521048943e-07,
+      "loss": 0.1639,
+      "step": 23499
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.10915411392358e-07,
+      "loss": 0.17,
+      "step": 23500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.104171005085716e-07,
+      "loss": 0.1822,
+      "step": 23501
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.099189194606573e-07,
+      "loss": 0.1776,
+      "step": 23502
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.094208682557293e-07,
+      "loss": 0.182,
+      "step": 23503
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.089229469008998e-07,
+      "loss": 0.1661,
+      "step": 23504
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.084251554032842e-07,
+      "loss": 0.2008,
+      "step": 23505
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.0792749376999e-07,
+      "loss": 0.1883,
+      "step": 23506
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.074299620081295e-07,
+      "loss": 0.1623,
+      "step": 23507
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.069325601248069e-07,
+      "loss": 0.1757,
+      "step": 23508
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.064352881271265e-07,
+      "loss": 0.1889,
+      "step": 23509
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.05938146022195e-07,
+      "loss": 0.1902,
+      "step": 23510
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.054411338171099e-07,
+      "loss": 0.1646,
+      "step": 23511
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.049442515189721e-07,
+      "loss": 0.1737,
+      "step": 23512
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.044474991348761e-07,
+      "loss": 0.1803,
+      "step": 23513
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.039508766719229e-07,
+      "loss": 0.1672,
+      "step": 23514
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.034543841372035e-07,
+      "loss": 0.1687,
+      "step": 23515
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.029580215378075e-07,
+      "loss": 0.1685,
+      "step": 23516
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.024617888808296e-07,
+      "loss": 0.1846,
+      "step": 23517
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.019656861733539e-07,
+      "loss": 0.1868,
+      "step": 23518
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.014697134224704e-07,
+      "loss": 0.1849,
+      "step": 23519
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.009738706352611e-07,
+      "loss": 0.172,
+      "step": 23520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.004781578188082e-07,
+      "loss": 0.1623,
+      "step": 23521
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.99982574980195e-07,
+      "loss": 0.179,
+      "step": 23522
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.994871221264966e-07,
+      "loss": 0.1611,
+      "step": 23523
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.989917992647968e-07,
+      "loss": 0.162,
+      "step": 23524
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.984966064021616e-07,
+      "loss": 0.1854,
+      "step": 23525
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.980015435456724e-07,
+      "loss": 0.1794,
+      "step": 23526
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.975066107023966e-07,
+      "loss": 0.1769,
+      "step": 23527
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.970118078794033e-07,
+      "loss": 0.1618,
+      "step": 23528
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.965171350837631e-07,
+      "loss": 0.176,
+      "step": 23529
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.960225923225385e-07,
+      "loss": 0.1619,
+      "step": 23530
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.955281796027982e-07,
+      "loss": 0.1889,
+      "step": 23531
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.950338969316009e-07,
+      "loss": 0.1718,
+      "step": 23532
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.945397443160064e-07,
+      "loss": 0.165,
+      "step": 23533
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.940457217630771e-07,
+      "loss": 0.1809,
+      "step": 23534
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.935518292798662e-07,
+      "loss": 0.1801,
+      "step": 23535
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.930580668734279e-07,
+      "loss": 0.1895,
+      "step": 23536
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.9256443455082e-07,
+      "loss": 0.1616,
+      "step": 23537
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.920709323190869e-07,
+      "loss": 0.1734,
+      "step": 23538
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.915775601852827e-07,
+      "loss": 0.173,
+      "step": 23539
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.91084318156451e-07,
+      "loss": 0.1779,
+      "step": 23540
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.905912062396416e-07,
+      "loss": 0.1766,
+      "step": 23541
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.900982244418954e-07,
+      "loss": 0.1766,
+      "step": 23542
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.896053727702536e-07,
+      "loss": 0.1646,
+      "step": 23543
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.891126512317594e-07,
+      "loss": 0.1657,
+      "step": 23544
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.88620059833446e-07,
+      "loss": 0.1719,
+      "step": 23545
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.881275985823534e-07,
+      "loss": 0.1777,
+      "step": 23546
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.87635267485516e-07,
+      "loss": 0.1887,
+      "step": 23547
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.871430665499625e-07,
+      "loss": 0.177,
+      "step": 23548
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.866509957827286e-07,
+      "loss": 0.1983,
+      "step": 23549
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.861590551908405e-07,
+      "loss": 0.1806,
+      "step": 23550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.856672447813242e-07,
+      "loss": 0.1689,
+      "step": 23551
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.851755645612047e-07,
+      "loss": 0.1799,
+      "step": 23552
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.84684014537508e-07,
+      "loss": 0.1629,
+      "step": 23553
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.841925947172525e-07,
+      "loss": 0.1644,
+      "step": 23554
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.837013051074572e-07,
+      "loss": 0.1691,
+      "step": 23555
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.832101457151432e-07,
+      "loss": 0.1929,
+      "step": 23556
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.827191165473225e-07,
+      "loss": 0.1845,
+      "step": 23557
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.822282176110131e-07,
+      "loss": 0.1715,
+      "step": 23558
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.817374489132236e-07,
+      "loss": 0.1644,
+      "step": 23559
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.812468104609639e-07,
+      "loss": 0.1606,
+      "step": 23560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.807563022612453e-07,
+      "loss": 0.1667,
+      "step": 23561
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.802659243210709e-07,
+      "loss": 0.1893,
+      "step": 23562
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.797756766474474e-07,
+      "loss": 0.1672,
+      "step": 23563
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.792855592473781e-07,
+      "loss": 0.1621,
+      "step": 23564
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.787955721278629e-07,
+      "loss": 0.1659,
+      "step": 23565
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.783057152959007e-07,
+      "loss": 0.1682,
+      "step": 23566
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.778159887584859e-07,
+      "loss": 0.1648,
+      "step": 23567
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.773263925226183e-07,
+      "loss": 0.169,
+      "step": 23568
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.768369265952903e-07,
+      "loss": 0.185,
+      "step": 23569
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.763475909834896e-07,
+      "loss": 0.1696,
+      "step": 23570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.758583856942116e-07,
+      "loss": 0.1781,
+      "step": 23571
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.753693107344385e-07,
+      "loss": 0.1684,
+      "step": 23572
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.748803661111627e-07,
+      "loss": 0.1825,
+      "step": 23573
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.743915518313628e-07,
+      "loss": 0.188,
+      "step": 23574
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.739028679020223e-07,
+      "loss": 0.166,
+      "step": 23575
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.734143143301243e-07,
+      "loss": 0.1629,
+      "step": 23576
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.729258911226446e-07,
+      "loss": 0.1779,
+      "step": 23577
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.724375982865607e-07,
+      "loss": 0.1758,
+      "step": 23578
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.71949435828845e-07,
+      "loss": 0.1779,
+      "step": 23579
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.714614037564751e-07,
+      "loss": 0.1786,
+      "step": 23580
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.709735020764187e-07,
+      "loss": 0.18,
+      "step": 23581
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.704857307956449e-07,
+      "loss": 0.1843,
+      "step": 23582
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.699980899211235e-07,
+      "loss": 0.1658,
+      "step": 23583
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.695105794598169e-07,
+      "loss": 0.1729,
+      "step": 23584
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.690231994186915e-07,
+      "loss": 0.1715,
+      "step": 23585
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.685359498047086e-07,
+      "loss": 0.1849,
+      "step": 23586
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.680488306248258e-07,
+      "loss": 0.189,
+      "step": 23587
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.675618418860043e-07,
+      "loss": 0.1772,
+      "step": 23588
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.670749835951964e-07,
+      "loss": 0.1622,
+      "step": 23589
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.665882557593619e-07,
+      "loss": 0.1834,
+      "step": 23590
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.661016583854476e-07,
+      "loss": 0.1822,
+      "step": 23591
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.656151914804078e-07,
+      "loss": 0.1726,
+      "step": 23592
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.651288550511894e-07,
+      "loss": 0.1967,
+      "step": 23593
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.646426491047377e-07,
+      "loss": 0.1746,
+      "step": 23594
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.641565736480017e-07,
+      "loss": 0.1696,
+      "step": 23595
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.636706286879215e-07,
+      "loss": 0.1803,
+      "step": 23596
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.631848142314403e-07,
+      "loss": 0.1836,
+      "step": 23597
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.62699130285497e-07,
+      "loss": 0.1714,
+      "step": 23598
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.622135768570261e-07,
+      "loss": 0.1814,
+      "step": 23599
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.617281539529687e-07,
+      "loss": 0.1605,
+      "step": 23600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.612428615802537e-07,
+      "loss": 0.1661,
+      "step": 23601
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.607576997458167e-07,
+      "loss": 0.1722,
+      "step": 23602
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.602726684565854e-07,
+      "loss": 0.1727,
+      "step": 23603
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.597877677194899e-07,
+      "loss": 0.1929,
+      "step": 23604
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.593029975414546e-07,
+      "loss": 0.1762,
+      "step": 23605
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.588183579294029e-07,
+      "loss": 0.1796,
+      "step": 23606
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.583338488902615e-07,
+      "loss": 0.1738,
+      "step": 23607
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.578494704309481e-07,
+      "loss": 0.1769,
+      "step": 23608
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.573652225583806e-07,
+      "loss": 0.1794,
+      "step": 23609
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.568811052794801e-07,
+      "loss": 0.1695,
+      "step": 23610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.563971186011577e-07,
+      "loss": 0.1904,
+      "step": 23611
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.559132625303302e-07,
+      "loss": 0.18,
+      "step": 23612
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.554295370739074e-07,
+      "loss": 0.1987,
+      "step": 23613
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.549459422387962e-07,
+      "loss": 0.1924,
+      "step": 23614
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.544624780319089e-07,
+      "loss": 0.1723,
+      "step": 23615
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.539791444601486e-07,
+      "loss": 0.1786,
+      "step": 23616
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.5349594153042e-07,
+      "loss": 0.1797,
+      "step": 23617
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.530128692496232e-07,
+      "loss": 0.1568,
+      "step": 23618
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.525299276246624e-07,
+      "loss": 0.1636,
+      "step": 23619
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.520471166624333e-07,
+      "loss": 0.1757,
+      "step": 23620
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.515644363698316e-07,
+      "loss": 0.1597,
+      "step": 23621
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.510818867537552e-07,
+      "loss": 0.1653,
+      "step": 23622
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.505994678210916e-07,
+      "loss": 0.1844,
+      "step": 23623
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.501171795787378e-07,
+      "loss": 0.1612,
+      "step": 23624
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.496350220335792e-07,
+      "loss": 0.1599,
+      "step": 23625
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.491529951925027e-07,
+      "loss": 0.1646,
+      "step": 23626
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.486710990623958e-07,
+      "loss": 0.1882,
+      "step": 23627
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.48189333650139e-07,
+      "loss": 0.1792,
+      "step": 23628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.477076989626198e-07,
+      "loss": 0.1562,
+      "step": 23629
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.472261950067095e-07,
+      "loss": 0.174,
+      "step": 23630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.467448217892915e-07,
+      "loss": 0.1784,
+      "step": 23631
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.462635793172403e-07,
+      "loss": 0.1824,
+      "step": 23632
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.45782467597428e-07,
+      "loss": 0.1801,
+      "step": 23633
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.453014866367315e-07,
+      "loss": 0.1547,
+      "step": 23634
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.448206364420164e-07,
+      "loss": 0.1791,
+      "step": 23635
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.443399170201538e-07,
+      "loss": 0.1682,
+      "step": 23636
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.438593283780105e-07,
+      "loss": 0.1811,
+      "step": 23637
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.433788705224488e-07,
+      "loss": 0.1668,
+      "step": 23638
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.428985434603354e-07,
+      "loss": 0.1869,
+      "step": 23639
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.424183471985281e-07,
+      "loss": 0.171,
+      "step": 23640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.419382817438859e-07,
+      "loss": 0.18,
+      "step": 23641
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.414583471032689e-07,
+      "loss": 0.1694,
+      "step": 23642
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.409785432835315e-07,
+      "loss": 0.1715,
+      "step": 23643
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.404988702915263e-07,
+      "loss": 0.1549,
+      "step": 23644
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.400193281341029e-07,
+      "loss": 0.1902,
+      "step": 23645
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.395399168181151e-07,
+      "loss": 0.1733,
+      "step": 23646
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.390606363504095e-07,
+      "loss": 0.1806,
+      "step": 23647
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.385814867378306e-07,
+      "loss": 0.1517,
+      "step": 23648
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.38102467987224e-07,
+      "loss": 0.1528,
+      "step": 23649
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.376235801054311e-07,
+      "loss": 0.1811,
+      "step": 23650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.371448230992951e-07,
+      "loss": 0.1853,
+      "step": 23651
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.366661969756518e-07,
+      "loss": 0.1947,
+      "step": 23652
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.361877017413377e-07,
+      "loss": 0.1679,
+      "step": 23653
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.357093374031899e-07,
+      "loss": 0.1575,
+      "step": 23654
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.352311039680383e-07,
+      "loss": 0.1741,
+      "step": 23655
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.347530014427197e-07,
+      "loss": 0.1675,
+      "step": 23656
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.342750298340552e-07,
+      "loss": 0.1967,
+      "step": 23657
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.337971891488784e-07,
+      "loss": 0.1673,
+      "step": 23658
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.333194793940125e-07,
+      "loss": 0.1911,
+      "step": 23659
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.3284190057628e-07,
+      "loss": 0.1741,
+      "step": 23660
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.323644527025054e-07,
+      "loss": 0.1878,
+      "step": 23661
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.318871357795055e-07,
+      "loss": 0.1859,
+      "step": 23662
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.314099498141026e-07,
+      "loss": 0.1753,
+      "step": 23663
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.30932894813109e-07,
+      "loss": 0.1814,
+      "step": 23664
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.304559707833393e-07,
+      "loss": 0.1995,
+      "step": 23665
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.299791777316079e-07,
+      "loss": 0.2035,
+      "step": 23666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.29502515664723e-07,
+      "loss": 0.1701,
+      "step": 23667
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.290259845894965e-07,
+      "loss": 0.1751,
+      "step": 23668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.285495845127323e-07,
+      "loss": 0.1857,
+      "step": 23669
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.280733154412369e-07,
+      "loss": 0.1722,
+      "step": 23670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.275971773818125e-07,
+      "loss": 0.1875,
+      "step": 23671
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.271211703412585e-07,
+      "loss": 0.1861,
+      "step": 23672
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.26645294326377e-07,
+      "loss": 0.17,
+      "step": 23673
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.261695493439659e-07,
+      "loss": 0.1793,
+      "step": 23674
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.256939354008164e-07,
+      "loss": 0.1594,
+      "step": 23675
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.252184525037277e-07,
+      "loss": 0.1799,
+      "step": 23676
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.247431006594874e-07,
+      "loss": 0.1879,
+      "step": 23677
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.242678798748894e-07,
+      "loss": 0.1512,
+      "step": 23678
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.237927901567178e-07,
+      "loss": 0.1818,
+      "step": 23679
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.233178315117596e-07,
+      "loss": 0.1928,
+      "step": 23680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.228430039468028e-07,
+      "loss": 0.1757,
+      "step": 23681
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.223683074686251e-07,
+      "loss": 0.193,
+      "step": 23682
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.218937420840101e-07,
+      "loss": 0.1756,
+      "step": 23683
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.214193077997335e-07,
+      "loss": 0.1593,
+      "step": 23684
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.209450046225765e-07,
+      "loss": 0.1684,
+      "step": 23685
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.204708325593114e-07,
+      "loss": 0.1814,
+      "step": 23686
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.199967916167084e-07,
+      "loss": 0.1971,
+      "step": 23687
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.195228818015444e-07,
+      "loss": 0.1691,
+      "step": 23688
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.190491031205849e-07,
+      "loss": 0.1586,
+      "step": 23689
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.185754555806002e-07,
+      "loss": 0.1871,
+      "step": 23690
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.181019391883538e-07,
+      "loss": 0.1901,
+      "step": 23691
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.17628553950609e-07,
+      "loss": 0.1773,
+      "step": 23692
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.171552998741295e-07,
+      "loss": 0.1776,
+      "step": 23693
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.16682176965673e-07,
+      "loss": 0.166,
+      "step": 23694
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.16209185232003e-07,
+      "loss": 0.1818,
+      "step": 23695
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.157363246798677e-07,
+      "loss": 0.156,
+      "step": 23696
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.152635953160271e-07,
+      "loss": 0.1697,
+      "step": 23697
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.147909971472323e-07,
+      "loss": 0.1788,
+      "step": 23698
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.143185301802326e-07,
+      "loss": 0.1777,
+      "step": 23699
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.13846194421778e-07,
+      "loss": 0.1595,
+      "step": 23700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.133739898786141e-07,
+      "loss": 0.1657,
+      "step": 23701
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.129019165574892e-07,
+      "loss": 0.1826,
+      "step": 23702
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.124299744651432e-07,
+      "loss": 0.1627,
+      "step": 23703
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.119581636083173e-07,
+      "loss": 0.1761,
+      "step": 23704
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.114864839937531e-07,
+      "loss": 0.1778,
+      "step": 23705
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.110149356281848e-07,
+      "loss": 0.1854,
+      "step": 23706
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.10543518518353e-07,
+      "loss": 0.1809,
+      "step": 23707
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.100722326709876e-07,
+      "loss": 0.171,
+      "step": 23708
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.09601078092821e-07,
+      "loss": 0.1761,
+      "step": 23709
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.091300547905834e-07,
+      "loss": 0.1808,
+      "step": 23710
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.086591627710027e-07,
+      "loss": 0.1733,
+      "step": 23711
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.081884020408059e-07,
+      "loss": 0.1707,
+      "step": 23712
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.077177726067175e-07,
+      "loss": 0.1904,
+      "step": 23713
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.072472744754567e-07,
+      "loss": 0.1688,
+      "step": 23714
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.067769076537491e-07,
+      "loss": 0.1856,
+      "step": 23715
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.063066721483104e-07,
+      "loss": 0.1558,
+      "step": 23716
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.058365679658586e-07,
+      "loss": 0.1745,
+      "step": 23717
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.053665951131084e-07,
+      "loss": 0.1834,
+      "step": 23718
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.048967535967722e-07,
+      "loss": 0.1882,
+      "step": 23719
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.044270434235624e-07,
+      "loss": 0.1844,
+      "step": 23720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.039574646001868e-07,
+      "loss": 0.1857,
+      "step": 23721
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.03488017133357e-07,
+      "loss": 0.198,
+      "step": 23722
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.030187010297719e-07,
+      "loss": 0.1837,
+      "step": 23723
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.025495162961405e-07,
+      "loss": 0.1816,
+      "step": 23724
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.02080462939162e-07,
+      "loss": 0.1728,
+      "step": 23725
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.016115409655367e-07,
+      "loss": 0.1826,
+      "step": 23726
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.011427503819646e-07,
+      "loss": 0.1974,
+      "step": 23727
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.006740911951394e-07,
+      "loss": 0.176,
+      "step": 23728
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.002055634117578e-07,
+      "loss": 0.1803,
+      "step": 23729
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.997371670385123e-07,
+      "loss": 0.166,
+      "step": 23730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.992689020820887e-07,
+      "loss": 0.1772,
+      "step": 23731
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.988007685491827e-07,
+      "loss": 0.1662,
+      "step": 23732
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.983327664464746e-07,
+      "loss": 0.1765,
+      "step": 23733
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.978648957806556e-07,
+      "loss": 0.1672,
+      "step": 23734
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.973971565584049e-07,
+      "loss": 0.1917,
+      "step": 23735
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.969295487864048e-07,
+      "loss": 0.2096,
+      "step": 23736
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.964620724713335e-07,
+      "loss": 0.1807,
+      "step": 23737
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.959947276198688e-07,
+      "loss": 0.1811,
+      "step": 23738
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.955275142386875e-07,
+      "loss": 0.19,
+      "step": 23739
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.950604323344613e-07,
+      "loss": 0.1919,
+      "step": 23740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.945934819138645e-07,
+      "loss": 0.182,
+      "step": 23741
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.941266629835664e-07,
+      "loss": 0.1664,
+      "step": 23742
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.936599755502316e-07,
+      "loss": 0.1801,
+      "step": 23743
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.931934196205315e-07,
+      "loss": 0.1768,
+      "step": 23744
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.927269952011285e-07,
+      "loss": 0.1628,
+      "step": 23745
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.922607022986817e-07,
+      "loss": 0.2,
+      "step": 23746
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.917945409198569e-07,
+      "loss": 0.1734,
+      "step": 23747
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.913285110713098e-07,
+      "loss": 0.1621,
+      "step": 23748
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.908626127596986e-07,
+      "loss": 0.1889,
+      "step": 23749
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.903968459916744e-07,
+      "loss": 0.1606,
+      "step": 23750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.899312107738954e-07,
+      "loss": 0.2075,
+      "step": 23751
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.894657071130107e-07,
+      "loss": 0.1585,
+      "step": 23752
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.890003350156672e-07,
+      "loss": 0.1798,
+      "step": 23753
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.885350944885162e-07,
+      "loss": 0.1796,
+      "step": 23754
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.880699855382002e-07,
+      "loss": 0.1804,
+      "step": 23755
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.876050081713659e-07,
+      "loss": 0.1695,
+      "step": 23756
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.871401623946528e-07,
+      "loss": 0.1832,
+      "step": 23757
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.866754482146987e-07,
+      "loss": 0.196,
+      "step": 23758
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.862108656381473e-07,
+      "loss": 0.1853,
+      "step": 23759
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.857464146716287e-07,
+      "loss": 0.1763,
+      "step": 23760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.852820953217832e-07,
+      "loss": 0.1665,
+      "step": 23761
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.848179075952367e-07,
+      "loss": 0.1933,
+      "step": 23762
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.843538514986248e-07,
+      "loss": 0.195,
+      "step": 23763
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.838899270385747e-07,
+      "loss": 0.1541,
+      "step": 23764
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.834261342217097e-07,
+      "loss": 0.1708,
+      "step": 23765
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.829624730546603e-07,
+      "loss": 0.1612,
+      "step": 23766
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.824989435440444e-07,
+      "loss": 0.1948,
+      "step": 23767
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.820355456964868e-07,
+      "loss": 0.1693,
+      "step": 23768
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.815722795186053e-07,
+      "loss": 0.1584,
+      "step": 23769
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.811091450170149e-07,
+      "loss": 0.1826,
+      "step": 23770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.806461421983358e-07,
+      "loss": 0.1825,
+      "step": 23771
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.80183271069177e-07,
+      "loss": 0.1665,
+      "step": 23772
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.797205316361534e-07,
+      "loss": 0.177,
+      "step": 23773
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.792579239058739e-07,
+      "loss": 0.1857,
+      "step": 23774
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.787954478849458e-07,
+      "loss": 0.1604,
+      "step": 23775
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.783331035799757e-07,
+      "loss": 0.1836,
+      "step": 23776
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.778708909975652e-07,
+      "loss": 0.1658,
+      "step": 23777
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.774088101443211e-07,
+      "loss": 0.1736,
+      "step": 23778
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.769468610268416e-07,
+      "loss": 0.1683,
+      "step": 23779
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.764850436517223e-07,
+      "loss": 0.1824,
+      "step": 23780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.760233580255649e-07,
+      "loss": 0.2015,
+      "step": 23781
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.755618041549596e-07,
+      "loss": 0.1811,
+      "step": 23782
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.751003820465031e-07,
+      "loss": 0.163,
+      "step": 23783
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.746390917067847e-07,
+      "loss": 0.1559,
+      "step": 23784
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.741779331423916e-07,
+      "loss": 0.1641,
+      "step": 23785
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.737169063599148e-07,
+      "loss": 0.1767,
+      "step": 23786
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.732560113659382e-07,
+      "loss": 0.1801,
+      "step": 23787
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.727952481670442e-07,
+      "loss": 0.1616,
+      "step": 23788
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.723346167698132e-07,
+      "loss": 0.1803,
+      "step": 23789
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.718741171808286e-07,
+      "loss": 0.1747,
+      "step": 23790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.714137494066653e-07,
+      "loss": 0.1574,
+      "step": 23791
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.709535134538993e-07,
+      "loss": 0.168,
+      "step": 23792
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.704934093291061e-07,
+      "loss": 0.1921,
+      "step": 23793
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.700334370388563e-07,
+      "loss": 0.1764,
+      "step": 23794
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.695735965897222e-07,
+      "loss": 0.1514,
+      "step": 23795
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.691138879882709e-07,
+      "loss": 0.1652,
+      "step": 23796
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.686543112410672e-07,
+      "loss": 0.1891,
+      "step": 23797
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.68194866354679e-07,
+      "loss": 0.176,
+      "step": 23798
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.677355533356656e-07,
+      "loss": 0.1827,
+      "step": 23799
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.672763721905918e-07,
+      "loss": 0.1641,
+      "step": 23800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.668173229260145e-07,
+      "loss": 0.178,
+      "step": 23801
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.663584055484908e-07,
+      "loss": 0.169,
+      "step": 23802
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.658996200645752e-07,
+      "loss": 0.1774,
+      "step": 23803
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.654409664808204e-07,
+      "loss": 0.193,
+      "step": 23804
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.649824448037813e-07,
+      "loss": 0.1686,
+      "step": 23805
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.645240550400024e-07,
+      "loss": 0.1789,
+      "step": 23806
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.640657971960364e-07,
+      "loss": 0.1705,
+      "step": 23807
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.636076712784268e-07,
+      "loss": 0.1632,
+      "step": 23808
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.631496772937163e-07,
+      "loss": 0.1637,
+      "step": 23809
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.626918152484486e-07,
+      "loss": 0.1881,
+      "step": 23810
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.622340851491628e-07,
+      "loss": 0.1643,
+      "step": 23811
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.617764870023981e-07,
+      "loss": 0.1824,
+      "step": 23812
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.613190208146914e-07,
+      "loss": 0.1785,
+      "step": 23813
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.608616865925755e-07,
+      "loss": 0.1638,
+      "step": 23814
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.60404484342584e-07,
+      "loss": 0.1772,
+      "step": 23815
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.599474140712449e-07,
+      "loss": 0.1731,
+      "step": 23816
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.594904757850918e-07,
+      "loss": 0.1701,
+      "step": 23817
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.590336694906486e-07,
+      "loss": 0.168,
+      "step": 23818
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.585769951944399e-07,
+      "loss": 0.1681,
+      "step": 23819
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.581204529029907e-07,
+      "loss": 0.1591,
+      "step": 23820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.576640426228199e-07,
+      "loss": 0.1747,
+      "step": 23821
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.572077643604502e-07,
+      "loss": 0.1879,
+      "step": 23822
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.567516181223966e-07,
+      "loss": 0.1925,
+      "step": 23823
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.562956039151747e-07,
+      "loss": 0.1581,
+      "step": 23824
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.558397217452995e-07,
+      "loss": 0.1702,
+      "step": 23825
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.553839716192812e-07,
+      "loss": 0.1844,
+      "step": 23826
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.549283535436346e-07,
+      "loss": 0.1764,
+      "step": 23827
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.544728675248592e-07,
+      "loss": 0.1715,
+      "step": 23828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.540175135694683e-07,
+      "loss": 0.1953,
+      "step": 23829
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.535622916839647e-07,
+      "loss": 0.1845,
+      "step": 23830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.531072018748464e-07,
+      "loss": 0.189,
+      "step": 23831
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.526522441486206e-07,
+      "loss": 0.1937,
+      "step": 23832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.521974185117798e-07,
+      "loss": 0.17,
+      "step": 23833
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.517427249708264e-07,
+      "loss": 0.2026,
+      "step": 23834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.512881635322522e-07,
+      "loss": 0.1709,
+      "step": 23835
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.508337342025485e-07,
+      "loss": 0.1896,
+      "step": 23836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.5037943698821e-07,
+      "loss": 0.1907,
+      "step": 23837
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.499252718957228e-07,
+      "loss": 0.1786,
+      "step": 23838
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.494712389315773e-07,
+      "loss": 0.1866,
+      "step": 23839
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.490173381022581e-07,
+      "loss": 0.1649,
+      "step": 23840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.485635694142468e-07,
+      "loss": 0.1849,
+      "step": 23841
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.481099328740272e-07,
+      "loss": 0.1891,
+      "step": 23842
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.47656428488075e-07,
+      "loss": 0.179,
+      "step": 23843
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.47203056262874e-07,
+      "loss": 0.1699,
+      "step": 23844
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.467498162048947e-07,
+      "loss": 0.1676,
+      "step": 23845
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.462967083206162e-07,
+      "loss": 0.1658,
+      "step": 23846
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.45843732616508e-07,
+      "loss": 0.1686,
+      "step": 23847
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.453908890990391e-07,
+      "loss": 0.1684,
+      "step": 23848
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.449381777746811e-07,
+      "loss": 0.1905,
+      "step": 23849
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.444855986499e-07,
+      "loss": 0.1714,
+      "step": 23850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.440331517311561e-07,
+      "loss": 0.1686,
+      "step": 23851
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.435808370249187e-07,
+      "loss": 0.1506,
+      "step": 23852
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.431286545376437e-07,
+      "loss": 0.169,
+      "step": 23853
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.426766042757938e-07,
+      "loss": 0.1599,
+      "step": 23854
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.422246862458216e-07,
+      "loss": 0.1742,
+      "step": 23855
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.417729004541863e-07,
+      "loss": 0.1818,
+      "step": 23856
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.413212469073395e-07,
+      "loss": 0.1945,
+      "step": 23857
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.408697256117314e-07,
+      "loss": 0.1656,
+      "step": 23858
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.40418336573816e-07,
+      "loss": 0.1906,
+      "step": 23859
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.399670798000347e-07,
+      "loss": 0.1609,
+      "step": 23860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.395159552968389e-07,
+      "loss": 0.1687,
+      "step": 23861
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.390649630706703e-07,
+      "loss": 0.1792,
+      "step": 23862
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.386141031279692e-07,
+      "loss": 0.1837,
+      "step": 23863
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.381633754751793e-07,
+      "loss": 0.1724,
+      "step": 23864
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.377127801187345e-07,
+      "loss": 0.1599,
+      "step": 23865
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.372623170650761e-07,
+      "loss": 0.1712,
+      "step": 23866
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.368119863206336e-07,
+      "loss": 0.177,
+      "step": 23867
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.363617878918427e-07,
+      "loss": 0.1959,
+      "step": 23868
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.35911721785133e-07,
+      "loss": 0.1585,
+      "step": 23869
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.354617880069326e-07,
+      "loss": 0.1826,
+      "step": 23870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.350119865636695e-07,
+      "loss": 0.1775,
+      "step": 23871
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.345623174617678e-07,
+      "loss": 0.1864,
+      "step": 23872
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.341127807076521e-07,
+      "loss": 0.1702,
+      "step": 23873
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.336633763077427e-07,
+      "loss": 0.1776,
+      "step": 23874
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.332141042684571e-07,
+      "loss": 0.1692,
+      "step": 23875
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.327649645962165e-07,
+      "loss": 0.1751,
+      "step": 23876
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.323159572974314e-07,
+      "loss": 0.1582,
+      "step": 23877
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.318670823785212e-07,
+      "loss": 0.1702,
+      "step": 23878
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.31418339845894e-07,
+      "loss": 0.1712,
+      "step": 23879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.309697297059615e-07,
+      "loss": 0.1884,
+      "step": 23880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.305212519651295e-07,
+      "loss": 0.1643,
+      "step": 23881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.30072906629804e-07,
+      "loss": 0.1761,
+      "step": 23882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.296246937063922e-07,
+      "loss": 0.1886,
+      "step": 23883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.291766132012945e-07,
+      "loss": 0.1812,
+      "step": 23884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.287286651209091e-07,
+      "loss": 0.1721,
+      "step": 23885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.282808494716398e-07,
+      "loss": 0.1562,
+      "step": 23886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.27833166259877e-07,
+      "loss": 0.1783,
+      "step": 23887
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.273856154920211e-07,
+      "loss": 0.1684,
+      "step": 23888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.269381971744615e-07,
+      "loss": 0.1875,
+      "step": 23889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.264909113135888e-07,
+      "loss": 0.1833,
+      "step": 23890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.260437579157953e-07,
+      "loss": 0.1652,
+      "step": 23891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.255967369874639e-07,
+      "loss": 0.1653,
+      "step": 23892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.251498485349862e-07,
+      "loss": 0.1826,
+      "step": 23893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.247030925647381e-07,
+      "loss": 0.1786,
+      "step": 23894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.242564690831055e-07,
+      "loss": 0.1857,
+      "step": 23895
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.238099780964669e-07,
+      "loss": 0.184,
+      "step": 23896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.233636196111982e-07,
+      "loss": 0.164,
+      "step": 23897
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.229173936336787e-07,
+      "loss": 0.1823,
+      "step": 23898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.224713001702799e-07,
+      "loss": 0.1692,
+      "step": 23899
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.220253392273758e-07,
+      "loss": 0.1661,
+      "step": 23900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.215795108113343e-07,
+      "loss": 0.1877,
+      "step": 23901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.211338149285241e-07,
+      "loss": 0.1798,
+      "step": 23902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.206882515853131e-07,
+      "loss": 0.1789,
+      "step": 23903
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.202428207880629e-07,
+      "loss": 0.1637,
+      "step": 23904
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.197975225431397e-07,
+      "loss": 0.1834,
+      "step": 23905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.193523568569016e-07,
+      "loss": 0.1611,
+      "step": 23906
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.18907323735708e-07,
+      "loss": 0.1705,
+      "step": 23907
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.184624231859172e-07,
+      "loss": 0.1658,
+      "step": 23908
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.180176552138796e-07,
+      "loss": 0.1957,
+      "step": 23909
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.175730198259533e-07,
+      "loss": 0.159,
+      "step": 23910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.17128517028487e-07,
+      "loss": 0.184,
+      "step": 23911
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.166841468278318e-07,
+      "loss": 0.1912,
+      "step": 23912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.16239909230333e-07,
+      "loss": 0.1579,
+      "step": 23913
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.157958042423362e-07,
+      "loss": 0.1734,
+      "step": 23914
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.153518318701868e-07,
+      "loss": 0.1875,
+      "step": 23915
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.149079921202251e-07,
+      "loss": 0.175,
+      "step": 23916
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.144642849987915e-07,
+      "loss": 0.1656,
+      "step": 23917
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.140207105122243e-07,
+      "loss": 0.161,
+      "step": 23918
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.135772686668596e-07,
+      "loss": 0.1798,
+      "step": 23919
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.1313395946903e-07,
+      "loss": 0.1666,
+      "step": 23920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.126907829250673e-07,
+      "loss": 0.18,
+      "step": 23921
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.122477390413051e-07,
+      "loss": 0.1781,
+      "step": 23922
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.118048278240708e-07,
+      "loss": 0.1694,
+      "step": 23923
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.113620492796868e-07,
+      "loss": 0.1812,
+      "step": 23924
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.109194034144839e-07,
+      "loss": 0.1792,
+      "step": 23925
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.104768902347803e-07,
+      "loss": 0.1657,
+      "step": 23926
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.100345097468997e-07,
+      "loss": 0.1756,
+      "step": 23927
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.095922619571605e-07,
+      "loss": 0.1892,
+      "step": 23928
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.091501468718775e-07,
+      "loss": 0.1788,
+      "step": 23929
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.087081644973692e-07,
+      "loss": 0.1803,
+      "step": 23930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08266314839946e-07,
+      "loss": 0.1908,
+      "step": 23931
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.078245979059228e-07,
+      "loss": 0.1689,
+      "step": 23932
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.073830137016047e-07,
+      "loss": 0.1796,
+      "step": 23933
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.069415622333031e-07,
+      "loss": 0.1826,
+      "step": 23934
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.065002435073221e-07,
+      "loss": 0.182,
+      "step": 23935
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.060590575299642e-07,
+      "loss": 0.1683,
+      "step": 23936
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.056180043075344e-07,
+      "loss": 0.1742,
+      "step": 23937
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.05177083846329e-07,
+      "loss": 0.1658,
+      "step": 23938
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.047362961526494e-07,
+      "loss": 0.1573,
+      "step": 23939
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.042956412327917e-07,
+      "loss": 0.1739,
+      "step": 23940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.038551190930464e-07,
+      "loss": 0.1731,
+      "step": 23941
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.034147297397109e-07,
+      "loss": 0.1822,
+      "step": 23942
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.02974473179071e-07,
+      "loss": 0.1556,
+      "step": 23943
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.025343494174197e-07,
+      "loss": 0.1769,
+      "step": 23944
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.020943584610418e-07,
+      "loss": 0.1834,
+      "step": 23945
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.016545003162223e-07,
+      "loss": 0.1588,
+      "step": 23946
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.012147749892439e-07,
+      "loss": 0.1782,
+      "step": 23947
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.007751824863862e-07,
+      "loss": 0.1549,
+      "step": 23948
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.003357228139318e-07,
+      "loss": 0.1603,
+      "step": 23949
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.998963959781546e-07,
+      "loss": 0.2055,
+      "step": 23950
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.994572019853318e-07,
+      "loss": 0.1859,
+      "step": 23951
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.990181408417373e-07,
+      "loss": 0.1648,
+      "step": 23952
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.985792125536406e-07,
+      "loss": 0.1672,
+      "step": 23953
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.981404171273142e-07,
+      "loss": 0.1702,
+      "step": 23954
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.977017545690234e-07,
+      "loss": 0.175,
+      "step": 23955
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.972632248850342e-07,
+      "loss": 0.1672,
+      "step": 23956
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.968248280816126e-07,
+      "loss": 0.168,
+      "step": 23957
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.963865641650181e-07,
+      "loss": 0.1762,
+      "step": 23958
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.959484331415146e-07,
+      "loss": 0.182,
+      "step": 23959
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.955104350173558e-07,
+      "loss": 0.1916,
+      "step": 23960
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.950725697988003e-07,
+      "loss": 0.1676,
+      "step": 23961
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.946348374921042e-07,
+      "loss": 0.1727,
+      "step": 23962
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.941972381035156e-07,
+      "loss": 0.1989,
+      "step": 23963
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.937597716392896e-07,
+      "loss": 0.1656,
+      "step": 23964
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.933224381056724e-07,
+      "loss": 0.173,
+      "step": 23965
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.928852375089135e-07,
+      "loss": 0.192,
+      "step": 23966
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924481698552554e-07,
+      "loss": 0.1663,
+      "step": 23967
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.920112351509412e-07,
+      "loss": 0.175,
+      "step": 23968
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.915744334022135e-07,
+      "loss": 0.1762,
+      "step": 23969
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.911377646153106e-07,
+      "loss": 0.1819,
+      "step": 23970
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.907012287964709e-07,
+      "loss": 0.1651,
+      "step": 23971
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.902648259519296e-07,
+      "loss": 0.2034,
+      "step": 23972
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.898285560879203e-07,
+      "loss": 0.17,
+      "step": 23973
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.893924192106738e-07,
+      "loss": 0.1954,
+      "step": 23974
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.889564153264183e-07,
+      "loss": 0.1813,
+      "step": 23975
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.885205444413867e-07,
+      "loss": 0.1705,
+      "step": 23976
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.880848065617995e-07,
+      "loss": 0.164,
+      "step": 23977
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.876492016938863e-07,
+      "loss": 0.1679,
+      "step": 23978
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.872137298438653e-07,
+      "loss": 0.1645,
+      "step": 23979
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.867783910179559e-07,
+      "loss": 0.1738,
+      "step": 23980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.863431852223811e-07,
+      "loss": 0.1721,
+      "step": 23981
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.859081124633526e-07,
+      "loss": 0.163,
+      "step": 23982
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.854731727470898e-07,
+      "loss": 0.1688,
+      "step": 23983
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.85038366079801e-07,
+      "loss": 0.1712,
+      "step": 23984
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.846036924677002e-07,
+      "loss": 0.1818,
+      "step": 23985
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.841691519169946e-07,
+      "loss": 0.1727,
+      "step": 23986
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.837347444338893e-07,
+      "loss": 0.1852,
+      "step": 23987
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.833004700245938e-07,
+      "loss": 0.165,
+      "step": 23988
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.828663286953074e-07,
+      "loss": 0.1815,
+      "step": 23989
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.824323204522343e-07,
+      "loss": 0.1708,
+      "step": 23990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.819984453015715e-07,
+      "loss": 0.1612,
+      "step": 23991
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.815647032495176e-07,
+      "loss": 0.1752,
+      "step": 23992
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.811310943022686e-07,
+      "loss": 0.1816,
+      "step": 23993
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.806976184660185e-07,
+      "loss": 0.1629,
+      "step": 23994
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.802642757469568e-07,
+      "loss": 0.1835,
+      "step": 23995
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.798310661512763e-07,
+      "loss": 0.1787,
+      "step": 23996
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.793979896851621e-07,
+      "loss": 0.1627,
+      "step": 23997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.789650463548048e-07,
+      "loss": 0.1673,
+      "step": 23998
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.785322361663827e-07,
+      "loss": 0.1605,
+      "step": 23999
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.780995591260819e-07,
+      "loss": 0.162,
+      "step": 24000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.776670152400821e-07,
+      "loss": 0.1786,
+      "step": 24001
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.772346045145595e-07,
+      "loss": 0.1926,
+      "step": 24002
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.768023269556945e-07,
+      "loss": 0.173,
+      "step": 24003
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.763701825696579e-07,
+      "loss": 0.1925,
+      "step": 24004
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.759381713626267e-07,
+      "loss": 0.1801,
+      "step": 24005
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.755062933407697e-07,
+      "loss": 0.1809,
+      "step": 24006
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.750745485102528e-07,
+      "loss": 0.16,
+      "step": 24007
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.746429368772489e-07,
+      "loss": 0.1692,
+      "step": 24008
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.742114584479176e-07,
+      "loss": 0.1613,
+      "step": 24009
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.737801132284272e-07,
+      "loss": 0.1699,
+      "step": 24010
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.733489012249361e-07,
+      "loss": 0.1626,
+      "step": 24011
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72917822443604e-07,
+      "loss": 0.172,
+      "step": 24012
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72486876890589e-07,
+      "loss": 0.183,
+      "step": 24013
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.720560645720442e-07,
+      "loss": 0.166,
+      "step": 24014
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.716253854941279e-07,
+      "loss": 0.1812,
+      "step": 24015
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.711948396629875e-07,
+      "loss": 0.1772,
+      "step": 24016
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.707644270847768e-07,
+      "loss": 0.1608,
+      "step": 24017
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.703341477656422e-07,
+      "loss": 0.1771,
+      "step": 24018
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.699040017117275e-07,
+      "loss": 0.1703,
+      "step": 24019
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.694739889291813e-07,
+      "loss": 0.1743,
+      "step": 24020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.690441094241418e-07,
+      "loss": 0.1832,
+      "step": 24021
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.68614363202752e-07,
+      "loss": 0.1665,
+      "step": 24022
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.681847502711514e-07,
+      "loss": 0.1851,
+      "step": 24023
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.67755270635474e-07,
+      "loss": 0.1911,
+      "step": 24024
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673259243018549e-07,
+      "loss": 0.1911,
+      "step": 24025
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.668967112764268e-07,
+      "loss": 0.1754,
+      "step": 24026
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.664676315653217e-07,
+      "loss": 0.1614,
+      "step": 24027
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.660386851746692e-07,
+      "loss": 0.1872,
+      "step": 24028
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.65609872110593e-07,
+      "loss": 0.1843,
+      "step": 24029
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.651811923792229e-07,
+      "loss": 0.1771,
+      "step": 24030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.647526459866771e-07,
+      "loss": 0.19,
+      "step": 24031
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.64324232939082e-07,
+      "loss": 0.1619,
+      "step": 24032
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.638959532425548e-07,
+      "loss": 0.1786,
+      "step": 24033
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.634678069032108e-07,
+      "loss": 0.1677,
+      "step": 24034
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.630397939271693e-07,
+      "loss": 0.2002,
+      "step": 24035
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.626119143205423e-07,
+      "loss": 0.1855,
+      "step": 24036
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.621841680894425e-07,
+      "loss": 0.1721,
+      "step": 24037
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.617565552399796e-07,
+      "loss": 0.1614,
+      "step": 24038
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.613290757782609e-07,
+      "loss": 0.1691,
+      "step": 24039
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.609017297103936e-07,
+      "loss": 0.1817,
+      "step": 24040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.604745170424798e-07,
+      "loss": 0.1785,
+      "step": 24041
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.600474377806244e-07,
+      "loss": 0.191,
+      "step": 24042
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.596204919309246e-07,
+      "loss": 0.1699,
+      "step": 24043
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.591936794994835e-07,
+      "loss": 0.1589,
+      "step": 24044
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.587670004923952e-07,
+      "loss": 0.1718,
+      "step": 24045
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.583404549157524e-07,
+      "loss": 0.1843,
+      "step": 24046
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.579140427756514e-07,
+      "loss": 0.1747,
+      "step": 24047
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.574877640781808e-07,
+      "loss": 0.1578,
+      "step": 24048
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.57061618829431e-07,
+      "loss": 0.1656,
+      "step": 24049
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.566356070354896e-07,
+      "loss": 0.1795,
+      "step": 24050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.562097287024393e-07,
+      "loss": 0.1394,
+      "step": 24051
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.557839838363656e-07,
+      "loss": 0.1559,
+      "step": 24052
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.553583724433466e-07,
+      "loss": 0.1751,
+      "step": 24053
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.549328945294664e-07,
+      "loss": 0.1931,
+      "step": 24054
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.54507550100798e-07,
+      "loss": 0.1635,
+      "step": 24055
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.54082339163421e-07,
+      "loss": 0.18,
+      "step": 24056
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.536572617234082e-07,
+      "loss": 0.1875,
+      "step": 24057
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.532323177868282e-07,
+      "loss": 0.1714,
+      "step": 24058
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.52807507359754e-07,
+      "loss": 0.1666,
+      "step": 24059
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.523828304482538e-07,
+      "loss": 0.1729,
+      "step": 24060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.519582870583919e-07,
+      "loss": 0.1786,
+      "step": 24061
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.515338771962343e-07,
+      "loss": 0.1795,
+      "step": 24062
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.511096008678408e-07,
+      "loss": 0.166,
+      "step": 24063
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.506854580792765e-07,
+      "loss": 0.1791,
+      "step": 24064
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.502614488365932e-07,
+      "loss": 0.1634,
+      "step": 24065
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 0.1612,
+      "step": 24066
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.494138310131082e-07,
+      "loss": 0.1688,
+      "step": 24067
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.4899022244441e-07,
+      "loss": 0.1767,
+      "step": 24068
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.485667474458134e-07,
+      "loss": 0.1778,
+      "step": 24069
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.481434060233627e-07,
+      "loss": 0.1602,
+      "step": 24070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.477201981831094e-07,
+      "loss": 0.1675,
+      "step": 24071
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.472971239310966e-07,
+      "loss": 0.1889,
+      "step": 24072
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.468741832733661e-07,
+      "loss": 0.1754,
+      "step": 24073
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.464513762159608e-07,
+      "loss": 0.1721,
+      "step": 24074
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.460287027649192e-07,
+      "loss": 0.1626,
+      "step": 24075
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.456061629262811e-07,
+      "loss": 0.1903,
+      "step": 24076
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.451837567060793e-07,
+      "loss": 0.194,
+      "step": 24077
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.447614841103489e-07,
+      "loss": 0.1708,
+      "step": 24078
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.443393451451219e-07,
+      "loss": 0.1874,
+      "step": 24079
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.439173398164245e-07,
+      "loss": 0.1864,
+      "step": 24080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.434954681302907e-07,
+      "loss": 0.1768,
+      "step": 24081
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.430737300927403e-07,
+      "loss": 0.1867,
+      "step": 24082
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.42652125709804e-07,
+      "loss": 0.1708,
+      "step": 24083
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.422306549874991e-07,
+      "loss": 0.1781,
+      "step": 24084
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.418093179318463e-07,
+      "loss": 0.1861,
+      "step": 24085
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.413881145488665e-07,
+      "loss": 0.1794,
+      "step": 24086
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.409670448445726e-07,
+      "loss": 0.177,
+      "step": 24087
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.405461088249843e-07,
+      "loss": 0.1858,
+      "step": 24088
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.4012530649611e-07,
+      "loss": 0.1663,
+      "step": 24089
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.397046378639627e-07,
+      "loss": 0.1797,
+      "step": 24090
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.392841029345498e-07,
+      "loss": 0.1725,
+      "step": 24091
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.388637017138766e-07,
+      "loss": 0.1719,
+      "step": 24092
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.384434342079526e-07,
+      "loss": 0.1648,
+      "step": 24093
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.380233004227765e-07,
+      "loss": 0.1674,
+      "step": 24094
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.376033003643534e-07,
+      "loss": 0.1893,
+      "step": 24095
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.371834340386807e-07,
+      "loss": 0.1937,
+      "step": 24096
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.367637014517536e-07,
+      "loss": 0.1775,
+      "step": 24097
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.363441026095718e-07,
+      "loss": 0.1819,
+      "step": 24098
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.359246375181271e-07,
+      "loss": 0.1655,
+      "step": 24099
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.355053061834093e-07,
+      "loss": 0.1969,
+      "step": 24100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.350861086114113e-07,
+      "loss": 0.1758,
+      "step": 24101
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.346670448081183e-07,
+      "loss": 0.1869,
+      "step": 24102
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.342481147795187e-07,
+      "loss": 0.1825,
+      "step": 24103
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.338293185315936e-07,
+      "loss": 0.1789,
+      "step": 24104
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.334106560703279e-07,
+      "loss": 0.2017,
+      "step": 24105
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.329921274017003e-07,
+      "loss": 0.1934,
+      "step": 24106
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.325737325316872e-07,
+      "loss": 0.1689,
+      "step": 24107
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.321554714662692e-07,
+      "loss": 0.1789,
+      "step": 24108
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.317373442114172e-07,
+      "loss": 0.1942,
+      "step": 24109
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.313193507731063e-07,
+      "loss": 0.2032,
+      "step": 24110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.309014911573064e-07,
+      "loss": 0.1556,
+      "step": 24111
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.304837653699836e-07,
+      "loss": 0.1628,
+      "step": 24112
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.300661734171087e-07,
+      "loss": 0.1818,
+      "step": 24113
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.296487153046427e-07,
+      "loss": 0.1716,
+      "step": 24114
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.292313910385528e-07,
+      "loss": 0.1645,
+      "step": 24115
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.288142006247977e-07,
+      "loss": 0.1716,
+      "step": 24116
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.283971440693371e-07,
+      "loss": 0.1827,
+      "step": 24117
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.279802213781283e-07,
+      "loss": 0.1764,
+      "step": 24118
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.275634325571234e-07,
+      "loss": 0.1805,
+      "step": 24119
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.271467776122808e-07,
+      "loss": 0.1689,
+      "step": 24120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.267302565495492e-07,
+      "loss": 0.1693,
+      "step": 24121
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.263138693748794e-07,
+      "loss": 0.1857,
+      "step": 24122
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.258976160942199e-07,
+      "loss": 0.1907,
+      "step": 24123
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.254814967135126e-07,
+      "loss": 0.1925,
+      "step": 24124
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.25065511238705e-07,
+      "loss": 0.1852,
+      "step": 24125
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.246496596757379e-07,
+      "loss": 0.1793,
+      "step": 24126
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.242339420305521e-07,
+      "loss": 0.1765,
+      "step": 24127
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.238183583090851e-07,
+      "loss": 0.1855,
+      "step": 24128
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.234029085172711e-07,
+      "loss": 0.1673,
+      "step": 24129
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.229875926610496e-07,
+      "loss": 0.168,
+      "step": 24130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.225724107463471e-07,
+      "loss": 0.1688,
+      "step": 24131
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.221573627790989e-07,
+      "loss": 0.1552,
+      "step": 24132
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.217424487652302e-07,
+      "loss": 0.171,
+      "step": 24133
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.213276687106673e-07,
+      "loss": 0.1721,
+      "step": 24134
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.209130226213378e-07,
+      "loss": 0.1717,
+      "step": 24135
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.204985105031625e-07,
+      "loss": 0.17,
+      "step": 24136
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.200841323620644e-07,
+      "loss": 0.1654,
+      "step": 24137
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.196698882039609e-07,
+      "loss": 0.1764,
+      "step": 24138
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.192557780347675e-07,
+      "loss": 0.1924,
+      "step": 24139
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.188418018604025e-07,
+      "loss": 0.1823,
+      "step": 24140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.18427959686776e-07,
+      "loss": 0.1608,
+      "step": 24141
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.180142515198029e-07,
+      "loss": 0.1708,
+      "step": 24142
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.176006773653909e-07,
+      "loss": 0.1854,
+      "step": 24143
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.171872372294474e-07,
+      "loss": 0.1934,
+      "step": 24144
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.167739311178777e-07,
+      "loss": 0.185,
+      "step": 24145
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.163607590365839e-07,
+      "loss": 0.1747,
+      "step": 24146
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.159477209914722e-07,
+      "loss": 0.1547,
+      "step": 24147
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.155348169884379e-07,
+      "loss": 0.1629,
+      "step": 24148
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.151220470333819e-07,
+      "loss": 0.1618,
+      "step": 24149
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.147094111321994e-07,
+      "loss": 0.1842,
+      "step": 24150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.142969092907836e-07,
+      "loss": 0.1632,
+      "step": 24151
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.138845415150285e-07,
+      "loss": 0.1681,
+      "step": 24152
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.134723078108218e-07,
+      "loss": 0.1536,
+      "step": 24153
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.130602081840542e-07,
+      "loss": 0.1615,
+      "step": 24154
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.126482426406121e-07,
+      "loss": 0.2185,
+      "step": 24155
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.122364111863788e-07,
+      "loss": 0.1824,
+      "step": 24156
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.11824713827237e-07,
+      "loss": 0.1619,
+      "step": 24157
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.114131505690668e-07,
+      "loss": 0.149,
+      "step": 24158
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.110017214177489e-07,
+      "loss": 0.1677,
+      "step": 24159
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.105904263791584e-07,
+      "loss": 0.1662,
+      "step": 24160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.101792654591721e-07,
+      "loss": 0.1495,
+      "step": 24161
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.097682386636616e-07,
+      "loss": 0.159,
+      "step": 24162
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.09357345998498e-07,
+      "loss": 0.1718,
+      "step": 24163
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.08946587469551e-07,
+      "loss": 0.1538,
+      "step": 24164
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.085359630826882e-07,
+      "loss": 0.1838,
+      "step": 24165
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.081254728437736e-07,
+      "loss": 0.1718,
+      "step": 24166
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.077151167586726e-07,
+      "loss": 0.167,
+      "step": 24167
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.073048948332449e-07,
+      "loss": 0.176,
+      "step": 24168
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.068948070733527e-07,
+      "loss": 0.1621,
+      "step": 24169
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.0648485348485e-07,
+      "loss": 0.1898,
+      "step": 24170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.060750340735955e-07,
+      "loss": 0.1884,
+      "step": 24171
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.056653488454423e-07,
+      "loss": 0.1795,
+      "step": 24172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.052557978062412e-07,
+      "loss": 0.1691,
+      "step": 24173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.048463809618444e-07,
+      "loss": 0.1488,
+      "step": 24174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.04437098318097e-07,
+      "loss": 0.1758,
+      "step": 24175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.040279498808487e-07,
+      "loss": 0.1839,
+      "step": 24176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.036189356559419e-07,
+      "loss": 0.1585,
+      "step": 24177
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.032100556492182e-07,
+      "loss": 0.1581,
+      "step": 24178
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.028013098665197e-07,
+      "loss": 0.1756,
+      "step": 24179
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.02392698313683e-07,
+      "loss": 0.1924,
+      "step": 24180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.019842209965487e-07,
+      "loss": 0.1572,
+      "step": 24181
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.015758779209479e-07,
+      "loss": 0.1555,
+      "step": 24182
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.011676690927138e-07,
+      "loss": 0.1881,
+      "step": 24183
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.007595945176792e-07,
+      "loss": 0.1815,
+      "step": 24184
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.003516542016686e-07,
+      "loss": 0.1582,
+      "step": 24185
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.999438481505149e-07,
+      "loss": 0.1934,
+      "step": 24186
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.995361763700381e-07,
+      "loss": 0.164,
+      "step": 24187
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.991286388660644e-07,
+      "loss": 0.1765,
+      "step": 24188
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.987212356444161e-07,
+      "loss": 0.1981,
+      "step": 24189
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.983139667109072e-07,
+      "loss": 0.1792,
+      "step": 24190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.979068320713621e-07,
+      "loss": 0.1698,
+      "step": 24191
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.974998317315905e-07,
+      "loss": 0.1809,
+      "step": 24192
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.9709296569741e-07,
+      "loss": 0.1898,
+      "step": 24193
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.966862339746316e-07,
+      "loss": 0.158,
+      "step": 24194
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.962796365690616e-07,
+      "loss": 0.1852,
+      "step": 24195
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.958731734865142e-07,
+      "loss": 0.1602,
+      "step": 24196
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.954668447327883e-07,
+      "loss": 0.1751,
+      "step": 24197
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.950606503136924e-07,
+      "loss": 0.181,
+      "step": 24198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.946545902350265e-07,
+      "loss": 0.1793,
+      "step": 24199
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.942486645025925e-07,
+      "loss": 0.1816,
+      "step": 24200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.93842873122189e-07,
+      "loss": 0.1815,
+      "step": 24201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.934372160996082e-07,
+      "loss": 0.1823,
+      "step": 24202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.930316934406499e-07,
+      "loss": 0.1904,
+      "step": 24203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.926263051511039e-07,
+      "loss": 0.1813,
+      "step": 24204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.922210512367598e-07,
+      "loss": 0.1593,
+      "step": 24205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.918159317034088e-07,
+      "loss": 0.1652,
+      "step": 24206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.914109465568352e-07,
+      "loss": 0.183,
+      "step": 24207
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.910060958028263e-07,
+      "loss": 0.1788,
+      "step": 24208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.906013794471632e-07,
+      "loss": 0.1707,
+      "step": 24209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.90196797495628e-07,
+      "loss": 0.1798,
+      "step": 24210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.897923499539981e-07,
+      "loss": 0.1856,
+      "step": 24211
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.893880368280514e-07,
+      "loss": 0.1764,
+      "step": 24212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.889838581235641e-07,
+      "loss": 0.164,
+      "step": 24213
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.885798138463073e-07,
+      "loss": 0.1589,
+      "step": 24214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.881759040020563e-07,
+      "loss": 0.1898,
+      "step": 24215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.877721285965776e-07,
+      "loss": 0.1707,
+      "step": 24216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.873684876356367e-07,
+      "loss": 0.1688,
+      "step": 24217
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.869649811250045e-07,
+      "loss": 0.1701,
+      "step": 24218
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.865616090704407e-07,
+      "loss": 0.1845,
+      "step": 24219
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.861583714777086e-07,
+      "loss": 0.1604,
+      "step": 24220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.857552683525691e-07,
+      "loss": 0.1888,
+      "step": 24221
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.853522997007777e-07,
+      "loss": 0.1695,
+      "step": 24222
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.849494655280929e-07,
+      "loss": 0.1772,
+      "step": 24223
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.845467658402648e-07,
+      "loss": 0.166,
+      "step": 24224
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.841442006430498e-07,
+      "loss": 0.1702,
+      "step": 24225
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.837417699421954e-07,
+      "loss": 0.1834,
+      "step": 24226
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.833394737434528e-07,
+      "loss": 0.1759,
+      "step": 24227
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.829373120525672e-07,
+      "loss": 0.1612,
+      "step": 24228
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.825352848752796e-07,
+      "loss": 0.1722,
+      "step": 24229
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.821333922173389e-07,
+      "loss": 0.1692,
+      "step": 24230
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.817316340844792e-07,
+      "loss": 0.183,
+      "step": 24231
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.813300104824448e-07,
+      "loss": 0.1579,
+      "step": 24232
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.809285214169702e-07,
+      "loss": 0.1566,
+      "step": 24233
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.805271668937884e-07,
+      "loss": 0.1481,
+      "step": 24234
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.801259469186382e-07,
+      "loss": 0.1794,
+      "step": 24235
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.797248614972429e-07,
+      "loss": 0.1678,
+      "step": 24236
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.793239106353366e-07,
+      "loss": 0.1866,
+      "step": 24237
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.789230943386459e-07,
+      "loss": 0.1932,
+      "step": 24238
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.78522412612893e-07,
+      "loss": 0.1605,
+      "step": 24239
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.781218654638066e-07,
+      "loss": 0.2009,
+      "step": 24240
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.77721452897102e-07,
+      "loss": 0.176,
+      "step": 24241
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.773211749185049e-07,
+      "loss": 0.1791,
+      "step": 24242
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.769210315337292e-07,
+      "loss": 0.1658,
+      "step": 24243
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.765210227484897e-07,
+      "loss": 0.1828,
+      "step": 24244
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761211485685037e-07,
+      "loss": 0.1785,
+      "step": 24245
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.75721408999479e-07,
+      "loss": 0.1714,
+      "step": 24246
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.753218040471286e-07,
+      "loss": 0.1735,
+      "step": 24247
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.749223337171606e-07,
+      "loss": 0.1956,
+      "step": 24248
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.745229980152788e-07,
+      "loss": 0.1709,
+      "step": 24249
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.74123796947188e-07,
+      "loss": 0.1833,
+      "step": 24250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.7372473051859e-07,
+      "loss": 0.1813,
+      "step": 24251
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.73325798735187e-07,
+      "loss": 0.2011,
+      "step": 24252
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.729270016026755e-07,
+      "loss": 0.173,
+      "step": 24253
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.725283391267533e-07,
+      "loss": 0.1972,
+      "step": 24254
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.721298113131135e-07,
+      "loss": 0.173,
+      "step": 24255
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.717314181674494e-07,
+      "loss": 0.1885,
+      "step": 24256
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.713331596954519e-07,
+      "loss": 0.1524,
+      "step": 24257
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.709350359028087e-07,
+      "loss": 0.1747,
+      "step": 24258
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.705370467952087e-07,
+      "loss": 0.1819,
+      "step": 24259
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.70139192378335e-07,
+      "loss": 0.1673,
+      "step": 24260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.697414726578709e-07,
+      "loss": 0.1808,
+      "step": 24261
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.693438876394997e-07,
+      "loss": 0.1788,
+      "step": 24262
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.689464373288944e-07,
+      "loss": 0.1797,
+      "step": 24263
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.685491217317396e-07,
+      "loss": 0.1676,
+      "step": 24264
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.681519408537039e-07,
+      "loss": 0.1783,
+      "step": 24265
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.677548947004663e-07,
+      "loss": 0.1691,
+      "step": 24266
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.673579832776966e-07,
+      "loss": 0.1679,
+      "step": 24267
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.669612065910601e-07,
+      "loss": 0.1828,
+      "step": 24268
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.665645646462303e-07,
+      "loss": 0.1866,
+      "step": 24269
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.661680574488703e-07,
+      "loss": 0.1947,
+      "step": 24270
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.657716850046435e-07,
+      "loss": 0.1822,
+      "step": 24271
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.653754473192119e-07,
+      "loss": 0.1936,
+      "step": 24272
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.649793443982344e-07,
+      "loss": 0.1786,
+      "step": 24273
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.645833762473719e-07,
+      "loss": 0.1772,
+      "step": 24274
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.641875428722787e-07,
+      "loss": 0.1624,
+      "step": 24275
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.637918442786084e-07,
+      "loss": 0.1706,
+      "step": 24276
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.633962804720139e-07,
+      "loss": 0.2044,
+      "step": 24277
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.630008514581431e-07,
+      "loss": 0.1787,
+      "step": 24278
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.626055572426481e-07,
+      "loss": 0.1748,
+      "step": 24279
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.622103978311722e-07,
+      "loss": 0.2019,
+      "step": 24280
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.618153732293618e-07,
+      "loss": 0.1741,
+      "step": 24281
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.614204834428593e-07,
+      "loss": 0.1841,
+      "step": 24282
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.610257284773035e-07,
+      "loss": 0.1831,
+      "step": 24283
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.606311083383365e-07,
+      "loss": 0.1847,
+      "step": 24284
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.602366230315903e-07,
+      "loss": 0.1619,
+      "step": 24285
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.598422725627051e-07,
+      "loss": 0.1769,
+      "step": 24286
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.594480569373106e-07,
+      "loss": 0.184,
+      "step": 24287
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.590539761610381e-07,
+      "loss": 0.175,
+      "step": 24288
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.586600302395173e-07,
+      "loss": 0.181,
+      "step": 24289
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.582662191783727e-07,
+      "loss": 0.1671,
+      "step": 24290
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.578725429832344e-07,
+      "loss": 0.172,
+      "step": 24291
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.574790016597209e-07,
+      "loss": 0.1815,
+      "step": 24292
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.570855952134568e-07,
+      "loss": 0.184,
+      "step": 24293
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.566923236500609e-07,
+      "loss": 0.1654,
+      "step": 24294
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.562991869751489e-07,
+      "loss": 0.1682,
+      "step": 24295
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.559061851943382e-07,
+      "loss": 0.1756,
+      "step": 24296
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.555133183132411e-07,
+      "loss": 0.1697,
+      "step": 24297
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.55120586337472e-07,
+      "loss": 0.173,
+      "step": 24298
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.547279892726387e-07,
+      "loss": 0.1813,
+      "step": 24299
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.543355271243478e-07,
+      "loss": 0.165,
+      "step": 24300
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.539431998982092e-07,
+      "loss": 0.1556,
+      "step": 24301
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.535510075998219e-07,
+      "loss": 0.173,
+      "step": 24302
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.531589502347912e-07,
+      "loss": 0.1917,
+      "step": 24303
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.527670278087149e-07,
+      "loss": 0.183,
+      "step": 24304
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.523752403271953e-07,
+      "loss": 0.1698,
+      "step": 24305
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.519835877958257e-07,
+      "loss": 0.1753,
+      "step": 24306
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.515920702202005e-07,
+      "loss": 0.2022,
+      "step": 24307
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.512006876059128e-07,
+      "loss": 0.1673,
+      "step": 24308
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.50809439958554e-07,
+      "loss": 0.1824,
+      "step": 24309
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.504183272837094e-07,
+      "loss": 0.1808,
+      "step": 24310
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.500273495869701e-07,
+      "loss": 0.1677,
+      "step": 24311
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.496365068739163e-07,
+      "loss": 0.1719,
+      "step": 24312
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.492457991501354e-07,
+      "loss": 0.1798,
+      "step": 24313
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.488552264212066e-07,
+      "loss": 0.167,
+      "step": 24314
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.484647886927074e-07,
+      "loss": 0.1777,
+      "step": 24315
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.480744859702159e-07,
+      "loss": 0.1666,
+      "step": 24316
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.476843182593062e-07,
+      "loss": 0.1777,
+      "step": 24317
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.472942855655527e-07,
+      "loss": 0.1677,
+      "step": 24318
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.469043878945257e-07,
+      "loss": 0.1791,
+      "step": 24319
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.465146252517973e-07,
+      "loss": 0.1721,
+      "step": 24320
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.461249976429306e-07,
+      "loss": 0.1671,
+      "step": 24321
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.457355050734936e-07,
+      "loss": 0.1854,
+      "step": 24322
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.453461475490496e-07,
+      "loss": 0.183,
+      "step": 24323
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.449569250751585e-07,
+      "loss": 0.1653,
+      "step": 24324
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.445678376573837e-07,
+      "loss": 0.1916,
+      "step": 24325
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.441788853012797e-07,
+      "loss": 0.2019,
+      "step": 24326
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.437900680124031e-07,
+      "loss": 0.1741,
+      "step": 24327
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.434013857963083e-07,
+      "loss": 0.178,
+      "step": 24328
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.430128386585465e-07,
+      "loss": 0.1705,
+      "step": 24329
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.426244266046676e-07,
+      "loss": 0.1647,
+      "step": 24330
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.422361496402195e-07,
+      "loss": 0.1641,
+      "step": 24331
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.418480077707511e-07,
+      "loss": 0.1668,
+      "step": 24332
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.414600010018045e-07,
+      "loss": 0.1744,
+      "step": 24333
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.410721293389198e-07,
+      "loss": 0.1603,
+      "step": 24334
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.406843927876415e-07,
+      "loss": 0.1793,
+      "step": 24335
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.402967913535051e-07,
+      "loss": 0.165,
+      "step": 24336
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.399093250420484e-07,
+      "loss": 0.184,
+      "step": 24337
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.39521993858807e-07,
+      "loss": 0.1581,
+      "step": 24338
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.391347978093098e-07,
+      "loss": 0.175,
+      "step": 24339
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.387477368990923e-07,
+      "loss": 0.1749,
+      "step": 24340
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.383608111336813e-07,
+      "loss": 0.1584,
+      "step": 24341
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.379740205186023e-07,
+      "loss": 0.1804,
+      "step": 24342
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.375873650593822e-07,
+      "loss": 0.1764,
+      "step": 24343
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.372008447615418e-07,
+      "loss": 0.1805,
+      "step": 24344
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.368144596306036e-07,
+      "loss": 0.155,
+      "step": 24345
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.364282096720863e-07,
+      "loss": 0.1875,
+      "step": 24346
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.36042094891509e-07,
+      "loss": 0.1803,
+      "step": 24347
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.356561152943851e-07,
+      "loss": 0.1528,
+      "step": 24348
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.352702708862256e-07,
+      "loss": 0.1888,
+      "step": 24349
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.348845616725473e-07,
+      "loss": 0.1945,
+      "step": 24350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.344989876588557e-07,
+      "loss": 0.2031,
+      "step": 24351
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.3411354885066e-07,
+      "loss": 0.1858,
+      "step": 24352
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.337282452534665e-07,
+      "loss": 0.1853,
+      "step": 24353
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.333430768727765e-07,
+      "loss": 0.1693,
+      "step": 24354
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.329580437140947e-07,
+      "loss": 0.1728,
+      "step": 24355
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.325731457829165e-07,
+      "loss": 0.1755,
+      "step": 24356
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.321883830847441e-07,
+      "loss": 0.1882,
+      "step": 24357
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.31803755625071e-07,
+      "loss": 0.1674,
+      "step": 24358
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.314192634093928e-07,
+      "loss": 0.176,
+      "step": 24359
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.310349064432019e-07,
+      "loss": 0.1784,
+      "step": 24360
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.306506847319859e-07,
+      "loss": 0.1689,
+      "step": 24361
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.302665982812349e-07,
+      "loss": 0.1897,
+      "step": 24362
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.298826470964346e-07,
+      "loss": 0.1843,
+      "step": 24363
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.294988311830717e-07,
+      "loss": 0.1701,
+      "step": 24364
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.291151505466252e-07,
+      "loss": 0.163,
+      "step": 24365
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.287316051925773e-07,
+      "loss": 0.165,
+      "step": 24366
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.28348195126408e-07,
+      "loss": 0.1671,
+      "step": 24367
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.279649203535898e-07,
+      "loss": 0.1921,
+      "step": 24368
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.275817808796013e-07,
+      "loss": 0.1801,
+      "step": 24369
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.271987767099118e-07,
+      "loss": 0.1561,
+      "step": 24370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.268159078499968e-07,
+      "loss": 0.166,
+      "step": 24371
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.264331743053219e-07,
+      "loss": 0.1957,
+      "step": 24372
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.260505760813528e-07,
+      "loss": 0.1776,
+      "step": 24373
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.256681131835595e-07,
+      "loss": 0.1583,
+      "step": 24374
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.252857856173999e-07,
+      "loss": 0.1653,
+      "step": 24375
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.249035933883384e-07,
+      "loss": 0.1912,
+      "step": 24376
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.24521536501833e-07,
+      "loss": 0.1824,
+      "step": 24377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.241396149633415e-07,
+      "loss": 0.176,
+      "step": 24378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.237578287783196e-07,
+      "loss": 0.1838,
+      "step": 24379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.233761779522195e-07,
+      "loss": 0.1784,
+      "step": 24380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.229946624904936e-07,
+      "loss": 0.1661,
+      "step": 24381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.226132823985919e-07,
+      "loss": 0.1906,
+      "step": 24382
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.222320376819601e-07,
+      "loss": 0.1947,
+      "step": 24383
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.21850928346046e-07,
+      "loss": 0.16,
+      "step": 24384
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.21469954396292e-07,
+      "loss": 0.1885,
+      "step": 24385
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.210891158381415e-07,
+      "loss": 0.1723,
+      "step": 24386
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.207084126770334e-07,
+      "loss": 0.1766,
+      "step": 24387
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.203278449184046e-07,
+      "loss": 0.1576,
+      "step": 24388
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.19947412567694e-07,
+      "loss": 0.165,
+      "step": 24389
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.195671156303316e-07,
+      "loss": 0.189,
+      "step": 24390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.191869541117545e-07,
+      "loss": 0.1658,
+      "step": 24391
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.188069280173902e-07,
+      "loss": 0.184,
+      "step": 24392
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.184270373526667e-07,
+      "loss": 0.1723,
+      "step": 24393
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.180472821230109e-07,
+      "loss": 0.1861,
+      "step": 24394
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.176676623338461e-07,
+      "loss": 0.1566,
+      "step": 24395
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.17288177990598e-07,
+      "loss": 0.1708,
+      "step": 24396
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.169088290986823e-07,
+      "loss": 0.1642,
+      "step": 24397
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.165296156635224e-07,
+      "loss": 0.1654,
+      "step": 24398
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.161505376905329e-07,
+      "loss": 0.1595,
+      "step": 24399
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.157715951851283e-07,
+      "loss": 0.2112,
+      "step": 24400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.153927881527221e-07,
+      "loss": 0.1822,
+      "step": 24401
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.150141165987232e-07,
+      "loss": 0.1761,
+      "step": 24402
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.146355805285452e-07,
+      "loss": 0.1657,
+      "step": 24403
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.142571799475904e-07,
+      "loss": 0.1733,
+      "step": 24404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.138789148612656e-07,
+      "loss": 0.1744,
+      "step": 24405
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.135007852749763e-07,
+      "loss": 0.163,
+      "step": 24406
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.131227911941183e-07,
+      "loss": 0.1771,
+      "step": 24407
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.127449326240952e-07,
+      "loss": 0.1831,
+      "step": 24408
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.123672095703025e-07,
+      "loss": 0.1748,
+      "step": 24409
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.119896220381371e-07,
+      "loss": 0.1733,
+      "step": 24410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.116121700329924e-07,
+      "loss": 0.1789,
+      "step": 24411
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.112348535602573e-07,
+      "loss": 0.1503,
+      "step": 24412
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.108576726253244e-07,
+      "loss": 0.1702,
+      "step": 24413
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.104806272335805e-07,
+      "loss": 0.1745,
+      "step": 24414
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.101037173904088e-07,
+      "loss": 0.173,
+      "step": 24415
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.097269431011975e-07,
+      "loss": 0.1635,
+      "step": 24416
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.093503043713255e-07,
+      "loss": 0.1848,
+      "step": 24417
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.08973801206174e-07,
+      "loss": 0.1601,
+      "step": 24418
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.08597433611121e-07,
+      "loss": 0.1769,
+      "step": 24419
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.082212015915411e-07,
+      "loss": 0.1783,
+      "step": 24420
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.078451051528099e-07,
+      "loss": 0.1705,
+      "step": 24421
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.074691443002988e-07,
+      "loss": 0.1867,
+      "step": 24422
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.070933190393778e-07,
+      "loss": 0.1857,
+      "step": 24423
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.06717629375415e-07,
+      "loss": 0.161,
+      "step": 24424
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.063420753137793e-07,
+      "loss": 0.1812,
+      "step": 24425
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.059666568598332e-07,
+      "loss": 0.1632,
+      "step": 24426
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.055913740189366e-07,
+      "loss": 0.1771,
+      "step": 24427
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.052162267964556e-07,
+      "loss": 0.1842,
+      "step": 24428
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.048412151977434e-07,
+      "loss": 0.1811,
+      "step": 24429
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.044663392281612e-07,
+      "loss": 0.1634,
+      "step": 24430
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.040915988930617e-07,
+      "loss": 0.16,
+      "step": 24431
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.037169941977959e-07,
+      "loss": 0.189,
+      "step": 24432
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.033425251477198e-07,
+      "loss": 0.1777,
+      "step": 24433
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.029681917481754e-07,
+      "loss": 0.1808,
+      "step": 24434
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.025939940045155e-07,
+      "loss": 0.1875,
+      "step": 24435
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.02219931922081e-07,
+      "loss": 0.184,
+      "step": 24436
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.018460055062191e-07,
+      "loss": 0.166,
+      "step": 24437
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.014722147622675e-07,
+      "loss": 0.1791,
+      "step": 24438
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.010985596955664e-07,
+      "loss": 0.1695,
+      "step": 24439
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.00725040311455e-07,
+      "loss": 0.1689,
+      "step": 24440
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.003516566152655e-07,
+      "loss": 0.1748,
+      "step": 24441
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.999784086123338e-07,
+      "loss": 0.1556,
+      "step": 24442
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.996052963079911e-07,
+      "loss": 0.1984,
+      "step": 24443
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.992323197075654e-07,
+      "loss": 0.1543,
+      "step": 24444
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.988594788163859e-07,
+      "loss": 0.1521,
+      "step": 24445
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.98486773639778e-07,
+      "loss": 0.1697,
+      "step": 24446
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.981142041830645e-07,
+      "loss": 0.1542,
+      "step": 24447
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.977417704515685e-07,
+      "loss": 0.1697,
+      "step": 24448
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.973694724506073e-07,
+      "loss": 0.1626,
+      "step": 24449
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.96997310185503e-07,
+      "loss": 0.1807,
+      "step": 24450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.96625283661567e-07,
+      "loss": 0.1752,
+      "step": 24451
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.962533928841174e-07,
+      "loss": 0.1833,
+      "step": 24452
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.958816378584641e-07,
+      "loss": 0.1844,
+      "step": 24453
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.955100185899164e-07,
+      "loss": 0.168,
+      "step": 24454
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.951385350837856e-07,
+      "loss": 0.1837,
+      "step": 24455
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.947671873453741e-07,
+      "loss": 0.1733,
+      "step": 24456
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.943959753799899e-07,
+      "loss": 0.1789,
+      "step": 24457
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.940248991929342e-07,
+      "loss": 0.1861,
+      "step": 24458
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.936539587895062e-07,
+      "loss": 0.1762,
+      "step": 24459
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.932831541750061e-07,
+      "loss": 0.1868,
+      "step": 24460
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.929124853547274e-07,
+      "loss": 0.1816,
+      "step": 24461
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.925419523339692e-07,
+      "loss": 0.1698,
+      "step": 24462
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.921715551180195e-07,
+      "loss": 0.182,
+      "step": 24463
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.918012937121741e-07,
+      "loss": 0.1735,
+      "step": 24464
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914311681217188e-07,
+      "loss": 0.1597,
+      "step": 24465
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.910611783519403e-07,
+      "loss": 0.1664,
+      "step": 24466
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.906913244081246e-07,
+      "loss": 0.1801,
+      "step": 24467
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.903216062955529e-07,
+      "loss": 0.1683,
+      "step": 24468
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.899520240195099e-07,
+      "loss": 0.1755,
+      "step": 24469
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.895825775852725e-07,
+      "loss": 0.2003,
+      "step": 24470
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.892132669981165e-07,
+      "loss": 0.1767,
+      "step": 24471
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.888440922633209e-07,
+      "loss": 0.1887,
+      "step": 24472
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.884750533861537e-07,
+      "loss": 0.1744,
+      "step": 24473
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.881061503718909e-07,
+      "loss": 0.1676,
+      "step": 24474
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.877373832257981e-07,
+      "loss": 0.1672,
+      "step": 24475
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.873687519531478e-07,
+      "loss": 0.1841,
+      "step": 24476
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.870002565592013e-07,
+      "loss": 0.1973,
+      "step": 24477
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.866318970492234e-07,
+      "loss": 0.1889,
+      "step": 24478
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.862636734284764e-07,
+      "loss": 0.1643,
+      "step": 24479
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.858955857022185e-07,
+      "loss": 0.1764,
+      "step": 24480
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.855276338757098e-07,
+      "loss": 0.1699,
+      "step": 24481
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.851598179542039e-07,
+      "loss": 0.1662,
+      "step": 24482
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.847921379429555e-07,
+      "loss": 0.204,
+      "step": 24483
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.844245938472181e-07,
+      "loss": 0.1679,
+      "step": 24484
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.840571856722398e-07,
+      "loss": 0.1769,
+      "step": 24485
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.836899134232687e-07,
+      "loss": 0.1796,
+      "step": 24486
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833227771055526e-07,
+      "loss": 0.1716,
+      "step": 24487
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.82955776724332e-07,
+      "loss": 0.1749,
+      "step": 24488
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.825889122848537e-07,
+      "loss": 0.196,
+      "step": 24489
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.822221837923547e-07,
+      "loss": 0.1702,
+      "step": 24490
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.81855591252075e-07,
+      "loss": 0.1823,
+      "step": 24491
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.814891346692507e-07,
+      "loss": 0.1867,
+      "step": 24492
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.811228140491153e-07,
+      "loss": 0.1801,
+      "step": 24493
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.807566293969035e-07,
+      "loss": 0.1804,
+      "step": 24494
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.803905807178433e-07,
+      "loss": 0.1561,
+      "step": 24495
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.80024668017166e-07,
+      "loss": 0.1673,
+      "step": 24496
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796588913000977e-07,
+      "loss": 0.1765,
+      "step": 24497
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.792932505718595e-07,
+      "loss": 0.172,
+      "step": 24498
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.789277458376807e-07,
+      "loss": 0.1899,
+      "step": 24499
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.785623771027759e-07,
+      "loss": 0.1872,
+      "step": 24500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.781971443723688e-07,
+      "loss": 0.1759,
+      "step": 24501
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.778320476516718e-07,
+      "loss": 0.1981,
+      "step": 24502
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.774670869459042e-07,
+      "loss": 0.1813,
+      "step": 24503
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.771022622602783e-07,
+      "loss": 0.1636,
+      "step": 24504
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.767375736000024e-07,
+      "loss": 0.1699,
+      "step": 24505
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.763730209702888e-07,
+      "loss": 0.1854,
+      "step": 24506
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7600860437634235e-07,
+      "loss": 0.1772,
+      "step": 24507
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.75644323823371e-07,
+      "loss": 0.1711,
+      "step": 24508
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.752801793165773e-07,
+      "loss": 0.1797,
+      "step": 24509
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.749161708611616e-07,
+      "loss": 0.1637,
+      "step": 24510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.745522984623241e-07,
+      "loss": 0.1727,
+      "step": 24511
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.741885621252629e-07,
+      "loss": 0.1764,
+      "step": 24512
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7382496185517266e-07,
+      "loss": 0.1675,
+      "step": 24513
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7346149765724715e-07,
+      "loss": 0.1793,
+      "step": 24514
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7309816953667875e-07,
+      "loss": 0.1823,
+      "step": 24515
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.727349774986578e-07,
+      "loss": 0.1612,
+      "step": 24516
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.723719215483691e-07,
+      "loss": 0.1752,
+      "step": 24517
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7200900169100284e-07,
+      "loss": 0.1621,
+      "step": 24518
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.716462179317405e-07,
+      "loss": 0.1678,
+      "step": 24519
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.712835702757623e-07,
+      "loss": 0.1697,
+      "step": 24520
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.709210587282531e-07,
+      "loss": 0.1561,
+      "step": 24521
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.705586832943865e-07,
+      "loss": 0.1739,
+      "step": 24522
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.701964439793416e-07,
+      "loss": 0.179,
+      "step": 24523
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6983434078829215e-07,
+      "loss": 0.1622,
+      "step": 24524
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6947237372640954e-07,
+      "loss": 0.1735,
+      "step": 24525
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6911054279886515e-07,
+      "loss": 0.16,
+      "step": 24526
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6874884801082377e-07,
+      "loss": 0.1628,
+      "step": 24527
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.683872893674579e-07,
+      "loss": 0.1783,
+      "step": 24528
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.680258668739268e-07,
+      "loss": 0.2022,
+      "step": 24529
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6766458053539633e-07,
+      "loss": 0.1908,
+      "step": 24530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6730343035702673e-07,
+      "loss": 0.1603,
+      "step": 24531
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.66942416343974e-07,
+      "loss": 0.1795,
+      "step": 24532
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.665815385013983e-07,
+      "loss": 0.1629,
+      "step": 24533
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.662207968344512e-07,
+      "loss": 0.1797,
+      "step": 24534
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6586019134828965e-07,
+      "loss": 0.1617,
+      "step": 24535
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6549972204806057e-07,
+      "loss": 0.1833,
+      "step": 24536
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.651393889389144e-07,
+      "loss": 0.1725,
+      "step": 24537
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.647791920260003e-07,
+      "loss": 0.1767,
+      "step": 24538
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.644191313144597e-07,
+      "loss": 0.1709,
+      "step": 24539
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.640592068094374e-07,
+      "loss": 0.1748,
+      "step": 24540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.636994185160737e-07,
+      "loss": 0.1811,
+      "step": 24541
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6333976643951006e-07,
+      "loss": 0.1736,
+      "step": 24542
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6298025058488127e-07,
+      "loss": 0.1635,
+      "step": 24543
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6262087095732323e-07,
+      "loss": 0.1877,
+      "step": 24544
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.622616275619718e-07,
+      "loss": 0.1701,
+      "step": 24545
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.61902520403954e-07,
+      "loss": 0.1657,
+      "step": 24546
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6154354948840354e-07,
+      "loss": 0.176,
+      "step": 24547
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6118471482044623e-07,
+      "loss": 0.2024,
+      "step": 24548
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.608260164052059e-07,
+      "loss": 0.1671,
+      "step": 24549
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6046745424780936e-07,
+      "loss": 0.1695,
+      "step": 24550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.601090283533771e-07,
+      "loss": 0.1698,
+      "step": 24551
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5975073872702945e-07,
+      "loss": 0.1589,
+      "step": 24552
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.593925853738823e-07,
+      "loss": 0.1666,
+      "step": 24553
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.590345682990516e-07,
+      "loss": 0.1851,
+      "step": 24554
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5867668750765424e-07,
+      "loss": 0.1801,
+      "step": 24555
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.583189430047985e-07,
+      "loss": 0.1674,
+      "step": 24556
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.57961334795598e-07,
+      "loss": 0.1981,
+      "step": 24557
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.576038628851587e-07,
+      "loss": 0.1713,
+      "step": 24558
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.572465272785853e-07,
+      "loss": 0.1653,
+      "step": 24559
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.568893279809861e-07,
+      "loss": 0.1647,
+      "step": 24560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5653226499745904e-07,
+      "loss": 0.165,
+      "step": 24561
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.56175338333108e-07,
+      "loss": 0.1749,
+      "step": 24562
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.558185479930299e-07,
+      "loss": 0.1843,
+      "step": 24563
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5546189398232075e-07,
+      "loss": 0.1943,
+      "step": 24564
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.551053763060753e-07,
+      "loss": 0.1619,
+      "step": 24565
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.547489949693851e-07,
+      "loss": 0.1621,
+      "step": 24566
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.543927499773426e-07,
+      "loss": 0.1849,
+      "step": 24567
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.540366413350339e-07,
+      "loss": 0.1884,
+      "step": 24568
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5368066904754816e-07,
+      "loss": 0.1939,
+      "step": 24569
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.533248331199702e-07,
+      "loss": 0.1805,
+      "step": 24570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.529691335573794e-07,
+      "loss": 0.2002,
+      "step": 24571
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.526135703648604e-07,
+      "loss": 0.1694,
+      "step": 24572
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5225814354749043e-07,
+      "loss": 0.1694,
+      "step": 24573
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.519028531103464e-07,
+      "loss": 0.1817,
+      "step": 24574
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5154769905850436e-07,
+      "loss": 0.1848,
+      "step": 24575
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5119268139703467e-07,
+      "loss": 0.1712,
+      "step": 24576
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5083780013101206e-07,
+      "loss": 0.1873,
+      "step": 24577
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.504830552655037e-07,
+      "loss": 0.1651,
+      "step": 24578
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5012844680557666e-07,
+      "loss": 0.1807,
+      "step": 24579
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.497739747562946e-07,
+      "loss": 0.1689,
+      "step": 24580
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4941963912272456e-07,
+      "loss": 0.179,
+      "step": 24581
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.490654399099259e-07,
+      "loss": 0.1796,
+      "step": 24582
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4871137712295675e-07,
+      "loss": 0.1757,
+      "step": 24583
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.483574507668764e-07,
+      "loss": 0.184,
+      "step": 24584
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4800366084673863e-07,
+      "loss": 0.1778,
+      "step": 24585
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4765000736760043e-07,
+      "loss": 0.1587,
+      "step": 24586
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4729649033451005e-07,
+      "loss": 0.1773,
+      "step": 24587
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.469431097525179e-07,
+      "loss": 0.1559,
+      "step": 24588
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4658986562667205e-07,
+      "loss": 0.1811,
+      "step": 24589
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.462367579620186e-07,
+      "loss": 0.1824,
+      "step": 24590
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.458837867636012e-07,
+      "loss": 0.1628,
+      "step": 24591
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4553095203646144e-07,
+      "loss": 0.1839,
+      "step": 24592
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.451782537856375e-07,
+      "loss": 0.1728,
+      "step": 24593
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4482569201617085e-07,
+      "loss": 0.1798,
+      "step": 24594
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.444732667330942e-07,
+      "loss": 0.1722,
+      "step": 24595
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.441209779414446e-07,
+      "loss": 0.176,
+      "step": 24596
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4376882564625245e-07,
+      "loss": 0.1633,
+      "step": 24597
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4341680985254597e-07,
+      "loss": 0.1678,
+      "step": 24598
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4306493056535785e-07,
+      "loss": 0.1714,
+      "step": 24599
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.427131877897106e-07,
+      "loss": 0.1759,
+      "step": 24600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4236158153063037e-07,
+      "loss": 0.1771,
+      "step": 24601
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.420101117931397e-07,
+      "loss": 0.1838,
+      "step": 24602
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.416587785822568e-07,
+      "loss": 0.1694,
+      "step": 24603
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.413075819030044e-07,
+      "loss": 0.1736,
+      "step": 24604
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4095652176039395e-07,
+      "loss": 0.1748,
+      "step": 24605
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4060559815944373e-07,
+      "loss": 0.1939,
+      "step": 24606
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.40254811105163e-07,
+      "loss": 0.1812,
+      "step": 24607
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.399041606025656e-07,
+      "loss": 0.183,
+      "step": 24608
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.395536466566597e-07,
+      "loss": 0.186,
+      "step": 24609
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3920326927244906e-07,
+      "loss": 0.1829,
+      "step": 24610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3885302845494194e-07,
+      "loss": 0.1629,
+      "step": 24611
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.385029242091399e-07,
+      "loss": 0.1884,
+      "step": 24612
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.381529565400444e-07,
+      "loss": 0.1702,
+      "step": 24613
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.378031254526538e-07,
+      "loss": 0.1799,
+      "step": 24614
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3745343095196513e-07,
+      "loss": 0.1609,
+      "step": 24615
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.371038730429733e-07,
+      "loss": 0.198,
+      "step": 24616
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.367544517306721e-07,
+      "loss": 0.1555,
+      "step": 24617
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.364051670200531e-07,
+      "loss": 0.1735,
+      "step": 24618
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.360560189161023e-07,
+      "loss": 0.1814,
+      "step": 24619
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.357070074238112e-07,
+      "loss": 0.1847,
+      "step": 24620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.353581325481626e-07,
+      "loss": 0.152,
+      "step": 24621
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3500939429413913e-07,
+      "loss": 0.1803,
+      "step": 24622
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.346607926667246e-07,
+      "loss": 0.1896,
+      "step": 24623
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3431232767089716e-07,
+      "loss": 0.1727,
+      "step": 24624
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.33963999311634e-07,
+      "loss": 0.1902,
+      "step": 24625
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.336158075939112e-07,
+      "loss": 0.1846,
+      "step": 24626
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3326775252270134e-07,
+      "loss": 0.1853,
+      "step": 24627
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.329198341029783e-07,
+      "loss": 0.1809,
+      "step": 24628
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3257205233971145e-07,
+      "loss": 0.1692,
+      "step": 24629
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3222440723786563e-07,
+      "loss": 0.1913,
+      "step": 24630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3187689880240913e-07,
+      "loss": 0.1788,
+      "step": 24631
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3152952703830355e-07,
+      "loss": 0.173,
+      "step": 24632
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3118229195051377e-07,
+      "loss": 0.1733,
+      "step": 24633
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3083519354399695e-07,
+      "loss": 0.1663,
+      "step": 24634
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.304882318237136e-07,
+      "loss": 0.1521,
+      "step": 24635
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.301414067946186e-07,
+      "loss": 0.1864,
+      "step": 24636
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2979471846166467e-07,
+      "loss": 0.1745,
+      "step": 24637
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.294481668298056e-07,
+      "loss": 0.1799,
+      "step": 24638
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2910175190399083e-07,
+      "loss": 0.1489,
+      "step": 24639
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.287554736891686e-07,
+      "loss": 0.1644,
+      "step": 24640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.28409332190286e-07,
+      "loss": 0.1743,
+      "step": 24641
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2806332741228586e-07,
+      "loss": 0.167,
+      "step": 24642
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.27717459360113e-07,
+      "loss": 0.1658,
+      "step": 24643
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2737172803870466e-07,
+      "loss": 0.1562,
+      "step": 24644
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2702613345300125e-07,
+      "loss": 0.1642,
+      "step": 24645
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.266806756079378e-07,
+      "loss": 0.1889,
+      "step": 24646
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.263353545084503e-07,
+      "loss": 0.1569,
+      "step": 24647
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2599017015947153e-07,
+      "loss": 0.1715,
+      "step": 24648
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.256451225659297e-07,
+      "loss": 0.1754,
+      "step": 24649
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.253002117327565e-07,
+      "loss": 0.1827,
+      "step": 24650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.249554376648768e-07,
+      "loss": 0.1611,
+      "step": 24651
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.246108003672156e-07,
+      "loss": 0.1642,
+      "step": 24652
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.242662998446978e-07,
+      "loss": 0.1683,
+      "step": 24653
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.239219361022395e-07,
+      "loss": 0.1754,
+      "step": 24654
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.235777091447646e-07,
+      "loss": 0.1755,
+      "step": 24655
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2323361897718793e-07,
+      "loss": 0.1734,
+      "step": 24656
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2288966560442455e-07,
+      "loss": 0.1808,
+      "step": 24657
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.225458490313883e-07,
+      "loss": 0.1592,
+      "step": 24658
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.222021692629874e-07,
+      "loss": 0.1667,
+      "step": 24659
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2185862630413357e-07,
+      "loss": 0.1704,
+      "step": 24660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.215152201597328e-07,
+      "loss": 0.1587,
+      "step": 24661
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.211719508346923e-07,
+      "loss": 0.19,
+      "step": 24662
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.208288183339149e-07,
+      "loss": 0.1917,
+      "step": 24663
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.204858226622988e-07,
+      "loss": 0.1598,
+      "step": 24664
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.201429638247467e-07,
+      "loss": 0.1649,
+      "step": 24665
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1980024182615376e-07,
+      "loss": 0.1687,
+      "step": 24666
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1945765667141813e-07,
+      "loss": 0.1985,
+      "step": 24667
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.191152083654315e-07,
+      "loss": 0.1783,
+      "step": 24668
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.187728969130844e-07,
+      "loss": 0.1665,
+      "step": 24669
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1843072231927075e-07,
+      "loss": 0.1636,
+      "step": 24670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1808868458887207e-07,
+      "loss": 0.168,
+      "step": 24671
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.177467837267779e-07,
+      "loss": 0.1768,
+      "step": 24672
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1740501973786984e-07,
+      "loss": 0.1851,
+      "step": 24673
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.170633926270329e-07,
+      "loss": 0.1808,
+      "step": 24674
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1672190239914533e-07,
+      "loss": 0.1652,
+      "step": 24675
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.163805490590822e-07,
+      "loss": 0.1823,
+      "step": 24676
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1603933261172404e-07,
+      "loss": 0.1816,
+      "step": 24677
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1569825306194025e-07,
+      "loss": 0.159,
+      "step": 24678
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.153573104146069e-07,
+      "loss": 0.1756,
+      "step": 24679
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1501650467459244e-07,
+      "loss": 0.1984,
+      "step": 24680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1467583584676395e-07,
+      "loss": 0.1711,
+      "step": 24681
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1433530393598877e-07,
+      "loss": 0.1623,
+      "step": 24682
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.139949089471307e-07,
+      "loss": 0.1625,
+      "step": 24683
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1365465088505253e-07,
+      "loss": 0.1544,
+      "step": 24684
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.133145297546115e-07,
+      "loss": 0.1629,
+      "step": 24685
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.129745455606704e-07,
+      "loss": 0.1702,
+      "step": 24686
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.126346983080842e-07,
+      "loss": 0.1674,
+      "step": 24687
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1229498800170464e-07,
+      "loss": 0.2016,
+      "step": 24688
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1195541464638665e-07,
+      "loss": 0.1764,
+      "step": 24689
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.116159782469797e-07,
+      "loss": 0.1755,
+      "step": 24690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1127667880833443e-07,
+      "loss": 0.1829,
+      "step": 24691
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1093751633529465e-07,
+      "loss": 0.1528,
+      "step": 24692
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.105984908327054e-07,
+      "loss": 0.174,
+      "step": 24693
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.102596023054106e-07,
+      "loss": 0.1791,
+      "step": 24694
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0992085075825084e-07,
+      "loss": 0.1914,
+      "step": 24695
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.095822361960644e-07,
+      "loss": 0.1544,
+      "step": 24696
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.092437586236886e-07,
+      "loss": 0.1609,
+      "step": 24697
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0890541804595506e-07,
+      "loss": 0.1888,
+      "step": 24698
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0856721446770106e-07,
+      "loss": 0.1593,
+      "step": 24699
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.08229147893755e-07,
+      "loss": 0.1617,
+      "step": 24700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.078912183289474e-07,
+      "loss": 0.1795,
+      "step": 24701
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0755342577810443e-07,
+      "loss": 0.1718,
+      "step": 24702
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0721577024604885e-07,
+      "loss": 0.1758,
+      "step": 24703
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.06878251737608e-07,
+      "loss": 0.1693,
+      "step": 24704
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0654087025760016e-07,
+      "loss": 0.1848,
+      "step": 24705
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0620362581084594e-07,
+      "loss": 0.169,
+      "step": 24706
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.058665184021626e-07,
+      "loss": 0.1827,
+      "step": 24707
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.05529548036363e-07,
+      "loss": 0.185,
+      "step": 24708
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.051927147182644e-07,
+      "loss": 0.184,
+      "step": 24709
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.04856018452674e-07,
+      "loss": 0.1829,
+      "step": 24710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.045194592444035e-07,
+      "loss": 0.1985,
+      "step": 24711
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.041830370982591e-07,
+      "loss": 0.1629,
+      "step": 24712
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.038467520190492e-07,
+      "loss": 0.1867,
+      "step": 24713
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.035106040115744e-07,
+      "loss": 0.1825,
+      "step": 24714
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.031745930806352e-07,
+      "loss": 0.1712,
+      "step": 24715
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.028387192310346e-07,
+      "loss": 0.1758,
+      "step": 24716
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0250298246756746e-07,
+      "loss": 0.1732,
+      "step": 24717
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.021673827950323e-07,
+      "loss": 0.1674,
+      "step": 24718
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.018319202182208e-07,
+      "loss": 0.157,
+      "step": 24719
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.014965947419236e-07,
+      "loss": 0.1666,
+      "step": 24720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.011614063709335e-07,
+      "loss": 0.1714,
+      "step": 24721
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0082635511003666e-07,
+      "loss": 0.167,
+      "step": 24722
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0049144096401925e-07,
+      "loss": 0.2023,
+      "step": 24723
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0015666393766417e-07,
+      "loss": 0.1757,
+      "step": 24724
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.998220240357553e-07,
+      "loss": 0.164,
+      "step": 24725
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.994875212630711e-07,
+      "loss": 0.176,
+      "step": 24726
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9915315562438884e-07,
+      "loss": 0.1716,
+      "step": 24727
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.98818927124488e-07,
+      "loss": 0.1774,
+      "step": 24728
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.984848357681403e-07,
+      "loss": 0.1645,
+      "step": 24729
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.981508815601165e-07,
+      "loss": 0.1813,
+      "step": 24730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.978170645051904e-07,
+      "loss": 0.1613,
+      "step": 24731
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.974833846081272e-07,
+      "loss": 0.178,
+      "step": 24732
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.971498418736952e-07,
+      "loss": 0.1824,
+      "step": 24733
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9681643630665734e-07,
+      "loss": 0.1888,
+      "step": 24734
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9648316791177645e-07,
+      "loss": 0.1704,
+      "step": 24735
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9615003669381536e-07,
+      "loss": 0.1675,
+      "step": 24736
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.95817042657527e-07,
+      "loss": 0.1785,
+      "step": 24737
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.95484185807673e-07,
+      "loss": 0.1762,
+      "step": 24738
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9515146614900523e-07,
+      "loss": 0.1931,
+      "step": 24739
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9481888368627764e-07,
+      "loss": 0.1804,
+      "step": 24740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.944864384242397e-07,
+      "loss": 0.1675,
+      "step": 24741
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9415413036764106e-07,
+      "loss": 0.1755,
+      "step": 24742
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9382195952122784e-07,
+      "loss": 0.1814,
+      "step": 24743
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.93489925889744e-07,
+      "loss": 0.1658,
+      "step": 24744
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.931580294779347e-07,
+      "loss": 0.1909,
+      "step": 24745
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.928262702905383e-07,
+      "loss": 0.1838,
+      "step": 24746
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9249464833229443e-07,
+      "loss": 0.1592,
+      "step": 24747
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.921631636079415e-07,
+      "loss": 0.1657,
+      "step": 24748
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.918318161222123e-07,
+      "loss": 0.1985,
+      "step": 24749
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9150060587984097e-07,
+      "loss": 0.1723,
+      "step": 24750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9116953288555804e-07,
+      "loss": 0.1715,
+      "step": 24751
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9083859714409313e-07,
+      "loss": 0.1938,
+      "step": 24752
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.905077986601724e-07,
+      "loss": 0.1985,
+      "step": 24753
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9017713743852103e-07,
+      "loss": 0.2137,
+      "step": 24754
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.898466134838641e-07,
+      "loss": 0.1891,
+      "step": 24755
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.895162268009189e-07,
+      "loss": 0.1859,
+      "step": 24756
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8918597739440953e-07,
+      "loss": 0.1621,
+      "step": 24757
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8885586526905105e-07,
+      "loss": 0.1802,
+      "step": 24758
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.885258904295575e-07,
+      "loss": 0.1958,
+      "step": 24759
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8819605288064387e-07,
+      "loss": 0.1714,
+      "step": 24760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.878663526270221e-07,
+      "loss": 0.1682,
+      "step": 24761
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.875367896734006e-07,
+      "loss": 0.1768,
+      "step": 24762
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8720736402448444e-07,
+      "loss": 0.1987,
+      "step": 24763
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8687807568498436e-07,
+      "loss": 0.1678,
+      "step": 24764
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.86548924659601e-07,
+      "loss": 0.1641,
+      "step": 24765
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.86219910953034e-07,
+      "loss": 0.1632,
+      "step": 24766
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8589103456998733e-07,
+      "loss": 0.1866,
+      "step": 24767
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.855622955151561e-07,
+      "loss": 0.1698,
+      "step": 24768
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.852336937932355e-07,
+      "loss": 0.1704,
+      "step": 24769
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.849052294089206e-07,
+      "loss": 0.1755,
+      "step": 24770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.845769023669033e-07,
+      "loss": 0.1806,
+      "step": 24771
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8424871267187305e-07,
+      "loss": 0.1834,
+      "step": 24772
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8392066032851725e-07,
+      "loss": 0.1751,
+      "step": 24773
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8359274534152116e-07,
+      "loss": 0.1756,
+      "step": 24774
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.832649677155731e-07,
+      "loss": 0.1551,
+      "step": 24775
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.829373274553483e-07,
+      "loss": 0.185,
+      "step": 24776
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.826098245655319e-07,
+      "loss": 0.1835,
+      "step": 24777
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.82282459050799e-07,
+      "loss": 0.1807,
+      "step": 24778
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.819552309158281e-07,
+      "loss": 0.1617,
+      "step": 24779
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8162814016529217e-07,
+      "loss": 0.1765,
+      "step": 24780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8130118680386187e-07,
+      "loss": 0.1746,
+      "step": 24781
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8097437083621123e-07,
+      "loss": 0.185,
+      "step": 24782
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.806476922670044e-07,
+      "loss": 0.177,
+      "step": 24783
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8032115110090973e-07,
+      "loss": 0.1947,
+      "step": 24784
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7999474734259245e-07,
+      "loss": 0.1809,
+      "step": 24785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7966848099671214e-07,
+      "loss": 0.1791,
+      "step": 24786
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.793423520679329e-07,
+      "loss": 0.177,
+      "step": 24787
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7901636056090987e-07,
+      "loss": 0.1671,
+      "step": 24788
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.786905064803015e-07,
+      "loss": 0.1683,
+      "step": 24789
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7836478983075966e-07,
+      "loss": 0.1901,
+      "step": 24790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.780392106169406e-07,
+      "loss": 0.1663,
+      "step": 24791
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7771376884349285e-07,
+      "loss": 0.1714,
+      "step": 24792
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7738846451506385e-07,
+      "loss": 0.1611,
+      "step": 24793
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.770632976363031e-07,
+      "loss": 0.149,
+      "step": 24794
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7673826821185256e-07,
+      "loss": 0.1781,
+      "step": 24795
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.764133762463584e-07,
+      "loss": 0.1789,
+      "step": 24796
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7608862174445816e-07,
+      "loss": 0.1619,
+      "step": 24797
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7576400471079023e-07,
+      "loss": 0.1874,
+      "step": 24798
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7543952514999535e-07,
+      "loss": 0.1664,
+      "step": 24799
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7511518306670547e-07,
+      "loss": 0.1813,
+      "step": 24800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7479097846555343e-07,
+      "loss": 0.1788,
+      "step": 24801
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.744669113511712e-07,
+      "loss": 0.17,
+      "step": 24802
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7414298172818607e-07,
+      "loss": 0.1986,
+      "step": 24803
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7381918960122776e-07,
+      "loss": 0.1756,
+      "step": 24804
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.734955349749192e-07,
+      "loss": 0.1634,
+      "step": 24805
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7317201785388445e-07,
+      "loss": 0.1618,
+      "step": 24806
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7284863824274433e-07,
+      "loss": 0.1754,
+      "step": 24807
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7252539614611725e-07,
+      "loss": 0.1633,
+      "step": 24808
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7220229156862297e-07,
+      "loss": 0.1611,
+      "step": 24809
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.718793245148733e-07,
+      "loss": 0.1773,
+      "step": 24810
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.715564949894834e-07,
+      "loss": 0.1656,
+      "step": 24811
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7123380299706524e-07,
+      "loss": 0.1759,
+      "step": 24812
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7091124854222613e-07,
+      "loss": 0.1757,
+      "step": 24813
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.705888316295758e-07,
+      "loss": 0.1699,
+      "step": 24814
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7026655226371834e-07,
+      "loss": 0.1661,
+      "step": 24815
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6994441044925664e-07,
+      "loss": 0.1862,
+      "step": 24816
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.696224061907927e-07,
+      "loss": 0.1541,
+      "step": 24817
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.693005394929272e-07,
+      "loss": 0.1744,
+      "step": 24818
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.689788103602565e-07,
+      "loss": 0.1902,
+      "step": 24819
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6865721879737473e-07,
+      "loss": 0.1491,
+      "step": 24820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.683357648088792e-07,
+      "loss": 0.1798,
+      "step": 24821
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.680144483993575e-07,
+      "loss": 0.165,
+      "step": 24822
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.676932695734026e-07,
+      "loss": 0.1841,
+      "step": 24823
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6737222833560183e-07,
+      "loss": 0.1876,
+      "step": 24824
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.670513246905383e-07,
+      "loss": 0.1763,
+      "step": 24825
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.667305586427983e-07,
+      "loss": 0.1775,
+      "step": 24826
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.664099301969626e-07,
+      "loss": 0.1771,
+      "step": 24827
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.660894393576131e-07,
+      "loss": 0.1857,
+      "step": 24828
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.657690861293239e-07,
+      "loss": 0.1739,
+      "step": 24829
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6544887051667476e-07,
+      "loss": 0.1775,
+      "step": 24830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.651287925242375e-07,
+      "loss": 0.1704,
+      "step": 24831
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.648088521565829e-07,
+      "loss": 0.1646,
+      "step": 24832
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.644890494182851e-07,
+      "loss": 0.1981,
+      "step": 24833
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6416938431390824e-07,
+      "loss": 0.1638,
+      "step": 24834
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6384985684801977e-07,
+      "loss": 0.1687,
+      "step": 24835
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.635304670251849e-07,
+      "loss": 0.1529,
+      "step": 24836
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6321121484996447e-07,
+      "loss": 0.1623,
+      "step": 24837
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6289210032691923e-07,
+      "loss": 0.172,
+      "step": 24838
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.625731234606078e-07,
+      "loss": 0.1656,
+      "step": 24839
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.622542842555843e-07,
+      "loss": 0.1684,
+      "step": 24840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6193558271640727e-07,
+      "loss": 0.17,
+      "step": 24841
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6161701884762424e-07,
+      "loss": 0.1607,
+      "step": 24842
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6129859265378933e-07,
+      "loss": 0.1643,
+      "step": 24843
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6098030413944776e-07,
+      "loss": 0.1806,
+      "step": 24844
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.606621533091492e-07,
+      "loss": 0.1796,
+      "step": 24845
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.603441401674368e-07,
+      "loss": 0.1828,
+      "step": 24846
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6002626471885127e-07,
+      "loss": 0.1928,
+      "step": 24847
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5970852696793566e-07,
+      "loss": 0.1725,
+      "step": 24848
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.593909269192264e-07,
+      "loss": 0.181,
+      "step": 24849
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.590734645772631e-07,
+      "loss": 0.1658,
+      "step": 24850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.587561399465778e-07,
+      "loss": 0.1606,
+      "step": 24851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5843895303170337e-07,
+      "loss": 0.1648,
+      "step": 24852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5812190383717306e-07,
+      "loss": 0.1739,
+      "step": 24853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5780499236751197e-07,
+      "loss": 0.1918,
+      "step": 24854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.574882186272499e-07,
+      "loss": 0.1801,
+      "step": 24855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.571715826209088e-07,
+      "loss": 0.1662,
+      "step": 24856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5685508435301383e-07,
+      "loss": 0.1657,
+      "step": 24857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5653872382808595e-07,
+      "loss": 0.1732,
+      "step": 24858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5622250105064147e-07,
+      "loss": 0.1806,
+      "step": 24859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5590641602520017e-07,
+      "loss": 0.1895,
+      "step": 24860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5559046875627503e-07,
+      "loss": 0.1759,
+      "step": 24861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5527465924838134e-07,
+      "loss": 0.1594,
+      "step": 24862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5495898750602773e-07,
+      "loss": 0.19,
+      "step": 24863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5464345353372397e-07,
+      "loss": 0.1707,
+      "step": 24864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5432805733597865e-07,
+      "loss": 0.1698,
+      "step": 24865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5401279891729366e-07,
+      "loss": 0.1744,
+      "step": 24866
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5369767828217663e-07,
+      "loss": 0.1736,
+      "step": 24867
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5338269543512494e-07,
+      "loss": 0.1501,
+      "step": 24868
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.530678503806395e-07,
+      "loss": 0.1737,
+      "step": 24869
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5275314312321783e-07,
+      "loss": 0.1612,
+      "step": 24870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.52438573667353e-07,
+      "loss": 0.1671,
+      "step": 24871
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.521241420175414e-07,
+      "loss": 0.1832,
+      "step": 24872
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.518098481782728e-07,
+      "loss": 0.1568,
+      "step": 24873
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.514956921540358e-07,
+      "loss": 0.1702,
+      "step": 24874
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5118167394931904e-07,
+      "loss": 0.172,
+      "step": 24875
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5086779356860777e-07,
+      "loss": 0.166,
+      "step": 24876
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.505540510163863e-07,
+      "loss": 0.1802,
+      "step": 24877
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5024044629713426e-07,
+      "loss": 0.1629,
+      "step": 24878
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.499269794153315e-07,
+      "loss": 0.1677,
+      "step": 24879
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.496136503754588e-07,
+      "loss": 0.1518,
+      "step": 24880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4930045918198705e-07,
+      "loss": 0.1536,
+      "step": 24881
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.489874058393927e-07,
+      "loss": 0.1573,
+      "step": 24882
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.486744903521455e-07,
+      "loss": 0.1768,
+      "step": 24883
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.483617127247174e-07,
+      "loss": 0.1679,
+      "step": 24884
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4804907296157596e-07,
+      "loss": 0.1796,
+      "step": 24885
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.477365710671843e-07,
+      "loss": 0.1833,
+      "step": 24886
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4742420704600875e-07,
+      "loss": 0.165,
+      "step": 24887
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.471119809025092e-07,
+      "loss": 0.1635,
+      "step": 24888
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4679989264114866e-07,
+      "loss": 0.1728,
+      "step": 24889
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.464879422663825e-07,
+      "loss": 0.1791,
+      "step": 24890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.46176129782666e-07,
+      "loss": 0.1766,
+      "step": 24891
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4586445519445565e-07,
+      "loss": 0.1729,
+      "step": 24892
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.455529185062001e-07,
+      "loss": 0.1781,
+      "step": 24893
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4524151972235466e-07,
+      "loss": 0.1811,
+      "step": 24894
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.449302588473613e-07,
+      "loss": 0.179,
+      "step": 24895
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.446191358856688e-07,
+      "loss": 0.177,
+      "step": 24896
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.443081508417223e-07,
+      "loss": 0.1556,
+      "step": 24897
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4399730371996177e-07,
+      "loss": 0.1856,
+      "step": 24898
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.436865945248291e-07,
+      "loss": 0.1787,
+      "step": 24899
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.433760232607608e-07,
+      "loss": 0.1771,
+      "step": 24900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4306558993219443e-07,
+      "loss": 0.1742,
+      "step": 24901
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4275529454356413e-07,
+      "loss": 0.1731,
+      "step": 24902
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4244513709930094e-07,
+      "loss": 0.168,
+      "step": 24903
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4213511760383786e-07,
+      "loss": 0.1684,
+      "step": 24904
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4182523606160033e-07,
+      "loss": 0.1654,
+      "step": 24905
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.415154924770159e-07,
+      "loss": 0.1848,
+      "step": 24906
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4120588685450984e-07,
+      "loss": 0.1698,
+      "step": 24907
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4089641919850203e-07,
+      "loss": 0.1742,
+      "step": 24908
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4058708951341556e-07,
+      "loss": 0.1739,
+      "step": 24909
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4027789780366695e-07,
+      "loss": 0.1958,
+      "step": 24910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.399688440736748e-07,
+      "loss": 0.182,
+      "step": 24911
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.396599283278523e-07,
+      "loss": 0.1873,
+      "step": 24912
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3935115057061044e-07,
+      "loss": 0.193,
+      "step": 24913
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.390425108063633e-07,
+      "loss": 0.1725,
+      "step": 24914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3873400903951636e-07,
+      "loss": 0.1808,
+      "step": 24915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.384256452744783e-07,
+      "loss": 0.1746,
+      "step": 24916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3811741951565335e-07,
+      "loss": 0.1797,
+      "step": 24917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3780933176744355e-07,
+      "loss": 0.1754,
+      "step": 24918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3750138203424987e-07,
+      "loss": 0.1863,
+      "step": 24919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3719357032047205e-07,
+      "loss": 0.1796,
+      "step": 24920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3688589663050664e-07,
+      "loss": 0.1677,
+      "step": 24921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3657836096874565e-07,
+      "loss": 0.1564,
+      "step": 24922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3627096333958666e-07,
+      "loss": 0.1826,
+      "step": 24923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3596370374741725e-07,
+      "loss": 0.167,
+      "step": 24924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.356565821966262e-07,
+      "loss": 0.168,
+      "step": 24925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.353495986916022e-07,
+      "loss": 0.1716,
+      "step": 24926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.350427532367284e-07,
+      "loss": 0.1775,
+      "step": 24927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.347360458363902e-07,
+      "loss": 0.1712,
+      "step": 24928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3442947649496737e-07,
+      "loss": 0.1656,
+      "step": 24929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.341230452168376e-07,
+      "loss": 0.1731,
+      "step": 24930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.338167520063795e-07,
+      "loss": 0.1772,
+      "step": 24931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.335105968679675e-07,
+      "loss": 0.1711,
+      "step": 24932
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.33204579805978e-07,
+      "loss": 0.1807,
+      "step": 24933
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.328987008247764e-07,
+      "loss": 0.1648,
+      "step": 24934
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3259295992873476e-07,
+      "loss": 0.1766,
+      "step": 24935
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3228735712222184e-07,
+      "loss": 0.1713,
+      "step": 24936
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.319818924095985e-07,
+      "loss": 0.1688,
+      "step": 24937
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3167656579523365e-07,
+      "loss": 0.1951,
+      "step": 24938
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.313713772834848e-07,
+      "loss": 0.1719,
+      "step": 24939
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3106632687871063e-07,
+      "loss": 0.1611,
+      "step": 24940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.307614145852722e-07,
+      "loss": 0.1811,
+      "step": 24941
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.304566404075216e-07,
+      "loss": 0.1734,
+      "step": 24942
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.301520043498141e-07,
+      "loss": 0.1732,
+      "step": 24943
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2984750641649966e-07,
+      "loss": 0.1764,
+      "step": 24944
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.295431466119281e-07,
+      "loss": 0.1673,
+      "step": 24945
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.292389249404493e-07,
+      "loss": 0.17,
+      "step": 24946
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.289348414064053e-07,
+      "loss": 0.1788,
+      "step": 24947
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2863089601414155e-07,
+      "loss": 0.177,
+      "step": 24948
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.283270887679979e-07,
+      "loss": 0.1693,
+      "step": 24949
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.280234196723164e-07,
+      "loss": 0.1683,
+      "step": 24950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.277198887314337e-07,
+      "loss": 0.1806,
+      "step": 24951
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.274164959496851e-07,
+      "loss": 0.1817,
+      "step": 24952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.271132413314038e-07,
+      "loss": 0.1792,
+      "step": 24953
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.268101248809219e-07,
+      "loss": 0.1799,
+      "step": 24954
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2650714660257045e-07,
+      "loss": 0.1687,
+      "step": 24955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.262043065006748e-07,
+      "loss": 0.184,
+      "step": 24956
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.259016045795627e-07,
+      "loss": 0.1636,
+      "step": 24957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.255990408435572e-07,
+      "loss": 0.1537,
+      "step": 24958
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2529661529698054e-07,
+      "loss": 0.1601,
+      "step": 24959
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2499432794415255e-07,
+      "loss": 0.1656,
+      "step": 24960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2469217878938864e-07,
+      "loss": 0.1819,
+      "step": 24961
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.243901678370087e-07,
+      "loss": 0.1746,
+      "step": 24962
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2408829509132486e-07,
+      "loss": 0.1931,
+      "step": 24963
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2378656055664817e-07,
+      "loss": 0.1789,
+      "step": 24964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.234849642372895e-07,
+      "loss": 0.1778,
+      "step": 24965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2318350613755653e-07,
+      "loss": 0.1856,
+      "step": 24966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2288218626175704e-07,
+      "loss": 0.1506,
+      "step": 24967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.225810046141931e-07,
+      "loss": 0.1619,
+      "step": 24968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.222799611991656e-07,
+      "loss": 0.1687,
+      "step": 24969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.21979056020979e-07,
+      "loss": 0.1651,
+      "step": 24970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.216782890839265e-07,
+      "loss": 0.177,
+      "step": 24971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.21377660392308e-07,
+      "loss": 0.1672,
+      "step": 24972
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.210771699504167e-07,
+      "loss": 0.1636,
+      "step": 24973
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.207768177625448e-07,
+      "loss": 0.1698,
+      "step": 24974
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2047660383298097e-07,
+      "loss": 0.1835,
+      "step": 24975
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2017652816601407e-07,
+      "loss": 0.1604,
+      "step": 24976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.198765907659329e-07,
+      "loss": 0.177,
+      "step": 24977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.195767916370185e-07,
+      "loss": 0.1829,
+      "step": 24978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.19277130783554e-07,
+      "loss": 0.1792,
+      "step": 24979
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1897760820982173e-07,
+      "loss": 0.1784,
+      "step": 24980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1867822392009696e-07,
+      "loss": 0.1765,
+      "step": 24981
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1837897791865966e-07,
+      "loss": 0.1783,
+      "step": 24982
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1807987020978206e-07,
+      "loss": 0.1861,
+      "step": 24983
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1778090079773617e-07,
+      "loss": 0.163,
+      "step": 24984
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1748206968679416e-07,
+      "loss": 0.1737,
+      "step": 24985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.171833768812227e-07,
+      "loss": 0.1761,
+      "step": 24986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.168848223852905e-07,
+      "loss": 0.1694,
+      "step": 24987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1658640620325973e-07,
+      "loss": 0.1747,
+      "step": 24988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.162881283393948e-07,
+      "loss": 0.1807,
+      "step": 24989
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1598998879795673e-07,
+      "loss": 0.166,
+      "step": 24990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.15691987583201e-07,
+      "loss": 0.1793,
+      "step": 24991
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.153941246993875e-07,
+      "loss": 0.1737,
+      "step": 24992
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1509640015076946e-07,
+      "loss": 0.1779,
+      "step": 24993
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.147988139416003e-07,
+      "loss": 0.1752,
+      "step": 24994
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.145013660761298e-07,
+      "loss": 0.1778,
+      "step": 24995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1420405655860795e-07,
+      "loss": 0.1845,
+      "step": 24996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1390688539328027e-07,
+      "loss": 0.1692,
+      "step": 24997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1360985258439223e-07,
+      "loss": 0.1765,
+      "step": 24998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.133129581361882e-07,
+      "loss": 0.1644,
+      "step": 24999
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1301620205290485e-07,
+      "loss": 0.1744,
+      "step": 25000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1271958433878537e-07,
+      "loss": 0.1773,
+      "step": 25001
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.124231049980653e-07,
+      "loss": 0.1711,
+      "step": 25002
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.121267640349768e-07,
+      "loss": 0.1849,
+      "step": 25003
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1183056145375754e-07,
+      "loss": 0.1614,
+      "step": 25004
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1153449725863425e-07,
+      "loss": 0.1897,
+      "step": 25005
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1123857145383907e-07,
+      "loss": 0.1614,
+      "step": 25006
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1094278404359747e-07,
+      "loss": 0.1807,
+      "step": 25007
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1064713503213386e-07,
+      "loss": 0.1689,
+      "step": 25008
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1035162442367374e-07,
+      "loss": 0.1755,
+      "step": 25009
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1005625222243596e-07,
+      "loss": 0.202,
+      "step": 25010
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.097610184326394e-07,
+      "loss": 0.1805,
+      "step": 25011
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.094659230585051e-07,
+      "loss": 0.1838,
+      "step": 25012
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0917096610424193e-07,
+      "loss": 0.1762,
+      "step": 25013
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.088761475740676e-07,
+      "loss": 0.1758,
+      "step": 25014
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0858146747219096e-07,
+      "loss": 0.1628,
+      "step": 25015
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0828692580282427e-07,
+      "loss": 0.1616,
+      "step": 25016
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0799252257017185e-07,
+      "loss": 0.1887,
+      "step": 25017
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0769825777843933e-07,
+      "loss": 0.1674,
+      "step": 25018
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.07404131431831e-07,
+      "loss": 0.1697,
+      "step": 25019
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.071101435345469e-07,
+      "loss": 0.1581,
+      "step": 25020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0681629409078927e-07,
+      "loss": 0.175,
+      "step": 25021
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.065225831047525e-07,
+      "loss": 0.1931,
+      "step": 25022
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0622901058063205e-07,
+      "loss": 0.1827,
+      "step": 25023
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.059355765226224e-07,
+      "loss": 0.177,
+      "step": 25024
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.056422809349158e-07,
+      "loss": 0.1594,
+      "step": 25025
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.05349123821701e-07,
+      "loss": 0.1787,
+      "step": 25026
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0505610518716253e-07,
+      "loss": 0.1794,
+      "step": 25027
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.047632250354904e-07,
+      "loss": 0.1553,
+      "step": 25028
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.044704833708667e-07,
+      "loss": 0.1727,
+      "step": 25029
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0417788019747043e-07,
+      "loss": 0.1868,
+      "step": 25030
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.038854155194848e-07,
+      "loss": 0.1679,
+      "step": 25031
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0359308934108435e-07,
+      "loss": 0.1895,
+      "step": 25032
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0330090166644675e-07,
+      "loss": 0.1621,
+      "step": 25033
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0300885249974544e-07,
+      "loss": 0.1595,
+      "step": 25034
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0271694184515033e-07,
+      "loss": 0.1715,
+      "step": 25035
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0242516970683257e-07,
+      "loss": 0.1862,
+      "step": 25036
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.021335360889599e-07,
+      "loss": 0.1562,
+      "step": 25037
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0184204099569903e-07,
+      "loss": 0.176,
+      "step": 25038
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0155068443120996e-07,
+      "loss": 0.167,
+      "step": 25039
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.012594663996582e-07,
+      "loss": 0.1783,
+      "step": 25040
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.009683869052016e-07,
+      "loss": 0.1711,
+      "step": 25041
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0067744595199786e-07,
+      "loss": 0.1625,
+      "step": 25042
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0038664354420486e-07,
+      "loss": 0.177,
+      "step": 25043
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0009597968597484e-07,
+      "loss": 0.1613,
+      "step": 25044
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.998054543814577e-07,
+      "loss": 0.1816,
+      "step": 25045
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9951506763480685e-07,
+      "loss": 0.1753,
+      "step": 25046
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.992248194501679e-07,
+      "loss": 0.184,
+      "step": 25047
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.989347098316897e-07,
+      "loss": 0.1919,
+      "step": 25048
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.986447387835123e-07,
+      "loss": 0.1741,
+      "step": 25049
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9835490630977904e-07,
+      "loss": 0.1785,
+      "step": 25050
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.980652124146322e-07,
+      "loss": 0.1873,
+      "step": 25051
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9777565710220615e-07,
+      "loss": 0.1537,
+      "step": 25052
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.974862403766399e-07,
+      "loss": 0.2035,
+      "step": 25053
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.971969622420645e-07,
+      "loss": 0.1763,
+      "step": 25054
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9690782270261565e-07,
+      "loss": 0.1733,
+      "step": 25055
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9661882176241994e-07,
+      "loss": 0.1791,
+      "step": 25056
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.963299594256075e-07,
+      "loss": 0.1878,
+      "step": 25057
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9604123569630383e-07,
+      "loss": 0.1784,
+      "step": 25058
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.957526505786335e-07,
+      "loss": 0.1652,
+      "step": 25059
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.954642040767175e-07,
+      "loss": 0.1623,
+      "step": 25060
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9517589619467826e-07,
+      "loss": 0.1679,
+      "step": 25061
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9488772693663127e-07,
+      "loss": 0.1732,
+      "step": 25062
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.945996963066955e-07,
+      "loss": 0.1744,
+      "step": 25063
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.943118043089821e-07,
+      "loss": 0.183,
+      "step": 25064
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.940240509476067e-07,
+      "loss": 0.1821,
+      "step": 25065
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.93736436226677e-07,
+      "loss": 0.178,
+      "step": 25066
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.934489601503021e-07,
+      "loss": 0.1763,
+      "step": 25067
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9316162272258863e-07,
+      "loss": 0.1649,
+      "step": 25068
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9287442394764e-07,
+      "loss": 0.1755,
+      "step": 25069
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.925873638295595e-07,
+      "loss": 0.1835,
+      "step": 25070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.923004423724474e-07,
+      "loss": 0.191,
+      "step": 25071
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.920136595804024e-07,
+      "loss": 0.1741,
+      "step": 25072
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9172701545752024e-07,
+      "loss": 0.1643,
+      "step": 25073
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.914405100078943e-07,
+      "loss": 0.1671,
+      "step": 25074
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9115414323562016e-07,
+      "loss": 0.1761,
+      "step": 25075
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9086791514478463e-07,
+      "loss": 0.1605,
+      "step": 25076
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 0.191,
+      "step": 25077
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.902958750237894e-07,
+      "loss": 0.1673,
+      "step": 25078
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.900100630017999e-07,
+      "loss": 0.1754,
+      "step": 25079
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8972438967759255e-07,
+      "loss": 0.1774,
+      "step": 25080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8943885505524625e-07,
+      "loss": 0.1648,
+      "step": 25081
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.891534591388445e-07,
+      "loss": 0.1904,
+      "step": 25082
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8886820193245956e-07,
+      "loss": 0.1905,
+      "step": 25083
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8858308344016704e-07,
+      "loss": 0.178,
+      "step": 25084
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.882981036660404e-07,
+      "loss": 0.174,
+      "step": 25085
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.880132626141496e-07,
+      "loss": 0.1688,
+      "step": 25086
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8772856028856376e-07,
+      "loss": 0.1821,
+      "step": 25087
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8744399669335065e-07,
+      "loss": 0.1822,
+      "step": 25088
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.871595718325726e-07,
+      "loss": 0.1659,
+      "step": 25089
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.868752857102941e-07,
+      "loss": 0.1594,
+      "step": 25090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.865911383305753e-07,
+      "loss": 0.1773,
+      "step": 25091
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.863071296974751e-07,
+      "loss": 0.1929,
+      "step": 25092
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.860232598150492e-07,
+      "loss": 0.1833,
+      "step": 25093
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.857395286873543e-07,
+      "loss": 0.1782,
+      "step": 25094
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.854559363184417e-07,
+      "loss": 0.1662,
+      "step": 25095
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8517248271236255e-07,
+      "loss": 0.1826,
+      "step": 25096
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.848891678731669e-07,
+      "loss": 0.1683,
+      "step": 25097
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8460599180490044e-07,
+      "loss": 0.1752,
+      "step": 25098
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8432295451160887e-07,
+      "loss": 0.174,
+      "step": 25099
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.840400559973344e-07,
+      "loss": 0.1743,
+      "step": 25100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8375729626611724e-07,
+      "loss": 0.1695,
+      "step": 25101
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8347467532199857e-07,
+      "loss": 0.1527,
+      "step": 25102
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8319219316901293e-07,
+      "loss": 0.1744,
+      "step": 25103
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.829098498111982e-07,
+      "loss": 0.2029,
+      "step": 25104
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8262764525258334e-07,
+      "loss": 0.1739,
+      "step": 25105
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.82345579497203e-07,
+      "loss": 0.1702,
+      "step": 25106
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8206365254908496e-07,
+      "loss": 0.1848,
+      "step": 25107
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8178186441225497e-07,
+      "loss": 0.1628,
+      "step": 25108
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.815002150907409e-07,
+      "loss": 0.1738,
+      "step": 25109
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8121870458856284e-07,
+      "loss": 0.1978,
+      "step": 25110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8093733290974423e-07,
+      "loss": 0.1612,
+      "step": 25111
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8065610005830304e-07,
+      "loss": 0.1735,
+      "step": 25112
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8037500603825594e-07,
+      "loss": 0.1785,
+      "step": 25113
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.800940508536187e-07,
+      "loss": 0.1737,
+      "step": 25114
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.798132345084059e-07,
+      "loss": 0.1581,
+      "step": 25115
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.795325570066254e-07,
+      "loss": 0.1712,
+      "step": 25116
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7925201835229067e-07,
+      "loss": 0.1779,
+      "step": 25117
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.789716185494051e-07,
+      "loss": 0.184,
+      "step": 25118
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7869135760197675e-07,
+      "loss": 0.1664,
+      "step": 25119
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7841123551400563e-07,
+      "loss": 0.1707,
+      "step": 25120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7813125228949744e-07,
+      "loss": 0.1851,
+      "step": 25121
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.778514079324479e-07,
+      "loss": 0.1869,
+      "step": 25122
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.775717024468538e-07,
+      "loss": 0.1693,
+      "step": 25123
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.772921358367142e-07,
+      "loss": 0.1669,
+      "step": 25124
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.770127081060192e-07,
+      "loss": 0.1807,
+      "step": 25125
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.767334192587623e-07,
+      "loss": 0.1762,
+      "step": 25126
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7645426929893137e-07,
+      "loss": 0.1596,
+      "step": 25127
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7617525823051327e-07,
+      "loss": 0.1801,
+      "step": 25128
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.758963860574959e-07,
+      "loss": 0.16,
+      "step": 25129
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.756176527838594e-07,
+      "loss": 0.1457,
+      "step": 25130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7533905841358957e-07,
+      "loss": 0.1825,
+      "step": 25131
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7506060295066196e-07,
+      "loss": 0.1798,
+      "step": 25132
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.747822863990557e-07,
+      "loss": 0.1653,
+      "step": 25133
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7450410876274534e-07,
+      "loss": 0.1555,
+      "step": 25134
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7422607004570444e-07,
+      "loss": 0.1906,
+      "step": 25135
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7394817025190645e-07,
+      "loss": 0.1683,
+      "step": 25136
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.736704093853182e-07,
+      "loss": 0.1768,
+      "step": 25137
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7339278744990985e-07,
+      "loss": 0.1768,
+      "step": 25138
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7311530444964486e-07,
+      "loss": 0.1778,
+      "step": 25139
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7283796038848675e-07,
+      "loss": 0.1839,
+      "step": 25140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7256075527039905e-07,
+      "loss": 0.1758,
+      "step": 25141
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7228368909933964e-07,
+      "loss": 0.1789,
+      "step": 25142
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7200676187926765e-07,
+      "loss": 0.1812,
+      "step": 25143
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7172997361413766e-07,
+      "loss": 0.1796,
+      "step": 25144
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7145332430790426e-07,
+      "loss": 0.1734,
+      "step": 25145
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.711768139645177e-07,
+      "loss": 0.181,
+      "step": 25146
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7090044258792693e-07,
+      "loss": 0.1637,
+      "step": 25147
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7062421018208217e-07,
+      "loss": 0.1636,
+      "step": 25148
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.703481167509281e-07,
+      "loss": 0.156,
+      "step": 25149
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7007216229840705e-07,
+      "loss": 0.187,
+      "step": 25150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.697963468284637e-07,
+      "loss": 0.1705,
+      "step": 25151
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6952067034503484e-07,
+      "loss": 0.1775,
+      "step": 25152
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6924513285206065e-07,
+      "loss": 0.1923,
+      "step": 25153
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6896973435347584e-07,
+      "loss": 0.1811,
+      "step": 25154
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.686944748532128e-07,
+      "loss": 0.1692,
+      "step": 25155
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.684193543552061e-07,
+      "loss": 0.174,
+      "step": 25156
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.681443728633837e-07,
+      "loss": 0.1809,
+      "step": 25157
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6786953038167476e-07,
+      "loss": 0.1764,
+      "step": 25158
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.675948269140027e-07,
+      "loss": 0.1743,
+      "step": 25159
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.673202624642934e-07,
+      "loss": 0.1758,
+      "step": 25160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6704583703646925e-07,
+      "loss": 0.1614,
+      "step": 25161
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.667715506344482e-07,
+      "loss": 0.1689,
+      "step": 25162
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6649740326215036e-07,
+      "loss": 0.1712,
+      "step": 25163
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6622339492348936e-07,
+      "loss": 0.1897,
+      "step": 25164
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6594952562238096e-07,
+      "loss": 0.1803,
+      "step": 25165
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.656757953627376e-07,
+      "loss": 0.1677,
+      "step": 25166
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6540220414846606e-07,
+      "loss": 0.1763,
+      "step": 25167
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6512875198347774e-07,
+      "loss": 0.1602,
+      "step": 25168
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.648554388716762e-07,
+      "loss": 0.1679,
+      "step": 25169
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6458226481696827e-07,
+      "loss": 0.1886,
+      "step": 25170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.64309229823253e-07,
+      "loss": 0.1753,
+      "step": 25171
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6403633389443187e-07,
+      "loss": 0.1716,
+      "step": 25172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.637635770344027e-07,
+      "loss": 0.1746,
+      "step": 25173
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.634909592470614e-07,
+      "loss": 0.169,
+      "step": 25174
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.632184805363025e-07,
+      "loss": 0.1735,
+      "step": 25175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6294614090601745e-07,
+      "loss": 0.1726,
+      "step": 25176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6267394036009754e-07,
+      "loss": 0.1707,
+      "step": 25177
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6240187890242965e-07,
+      "loss": 0.1844,
+      "step": 25178
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6212995653689954e-07,
+      "loss": 0.1705,
+      "step": 25179
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6185817326739304e-07,
+      "loss": 0.1781,
+      "step": 25180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.615865290977904e-07,
+      "loss": 0.1794,
+      "step": 25181
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.61315024031974e-07,
+      "loss": 0.161,
+      "step": 25182
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6104365807382183e-07,
+      "loss": 0.1726,
+      "step": 25183
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6077243122720754e-07,
+      "loss": 0.1812,
+      "step": 25184
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6050134349600797e-07,
+      "loss": 0.1589,
+      "step": 25185
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.602303948840934e-07,
+      "loss": 0.1652,
+      "step": 25186
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5995958539533514e-07,
+      "loss": 0.1913,
+      "step": 25187
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.596889150336024e-07,
+      "loss": 0.1765,
+      "step": 25188
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.594183838027575e-07,
+      "loss": 0.1748,
+      "step": 25189
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.591479917066697e-07,
+      "loss": 0.1918,
+      "step": 25190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.588777387491981e-07,
+      "loss": 0.1783,
+      "step": 25191
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.586076249342051e-07,
+      "loss": 0.1708,
+      "step": 25192
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.583376502655477e-07,
+      "loss": 0.1752,
+      "step": 25193
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.580678147470816e-07,
+      "loss": 0.1852,
+      "step": 25194
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.577981183826639e-07,
+      "loss": 0.195,
+      "step": 25195
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5752856117614243e-07,
+      "loss": 0.1698,
+      "step": 25196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5725914313137315e-07,
+      "loss": 0.173,
+      "step": 25197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.569898642521995e-07,
+      "loss": 0.1654,
+      "step": 25198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5672072454246967e-07,
+      "loss": 0.1758,
+      "step": 25199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5645172400602934e-07,
+      "loss": 0.1881,
+      "step": 25200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.561828626467189e-07,
+      "loss": 0.1837,
+      "step": 25201
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.559141404683796e-07,
+      "loss": 0.1768,
+      "step": 25202
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5564555747484955e-07,
+      "loss": 0.1798,
+      "step": 25203
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5537711366996676e-07,
+      "loss": 0.2012,
+      "step": 25204
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5510880905756487e-07,
+      "loss": 0.1736,
+      "step": 25205
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.548406436414741e-07,
+      "loss": 0.1852,
+      "step": 25206
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5457261742552807e-07,
+      "loss": 0.1933,
+      "step": 25207
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.543047304135537e-07,
+      "loss": 0.2011,
+      "step": 25208
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5403698260937803e-07,
+      "loss": 0.1899,
+      "step": 25209
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5376937401682566e-07,
+      "loss": 0.1654,
+      "step": 25210
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5350190463971916e-07,
+      "loss": 0.1683,
+      "step": 25211
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.532345744818787e-07,
+      "loss": 0.1728,
+      "step": 25212
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5296738354712135e-07,
+      "loss": 0.1651,
+      "step": 25213
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5270033183926737e-07,
+      "loss": 0.1913,
+      "step": 25214
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5243341936212697e-07,
+      "loss": 0.1941,
+      "step": 25215
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.52166646119516e-07,
+      "loss": 0.1634,
+      "step": 25216
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5190001211524483e-07,
+      "loss": 0.1776,
+      "step": 25217
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5163351735311923e-07,
+      "loss": 0.1743,
+      "step": 25218
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5136716183694953e-07,
+      "loss": 0.1695,
+      "step": 25219
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.511009455705382e-07,
+      "loss": 0.1905,
+      "step": 25220
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5083486855768777e-07,
+      "loss": 0.1776,
+      "step": 25221
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.505689308021997e-07,
+      "loss": 0.1855,
+      "step": 25222
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5030313230787307e-07,
+      "loss": 0.1751,
+      "step": 25223
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.500374730785038e-07,
+      "loss": 0.1679,
+      "step": 25224
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.497719531178855e-07,
+      "loss": 0.1818,
+      "step": 25225
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.49506572429814e-07,
+      "loss": 0.1704,
+      "step": 25226
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4924133101807636e-07,
+      "loss": 0.1613,
+      "step": 25227
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.489762288864628e-07,
+      "loss": 0.1663,
+      "step": 25228
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.487112660387614e-07,
+      "loss": 0.1785,
+      "step": 25229
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.484464424787547e-07,
+      "loss": 0.1581,
+      "step": 25230
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.481817582102275e-07,
+      "loss": 0.1859,
+      "step": 25231
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4791721323695896e-07,
+      "loss": 0.1733,
+      "step": 25232
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4765280756272823e-07,
+      "loss": 0.1774,
+      "step": 25233
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.473885411913124e-07,
+      "loss": 0.1999,
+      "step": 25234
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4712441412648614e-07,
+      "loss": 0.1878,
+      "step": 25235
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.468604263720231e-07,
+      "loss": 0.1609,
+      "step": 25236
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.465965779316914e-07,
+      "loss": 0.1707,
+      "step": 25237
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.463328688092637e-07,
+      "loss": 0.1651,
+      "step": 25238
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.460692990085045e-07,
+      "loss": 0.1737,
+      "step": 25239
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.458058685331777e-07,
+      "loss": 0.1713,
+      "step": 25240
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.455425773870479e-07,
+      "loss": 0.1925,
+      "step": 25241
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.452794255738744e-07,
+      "loss": 0.1718,
+      "step": 25242
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4501641309741865e-07,
+      "loss": 0.1757,
+      "step": 25243
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.447535399614365e-07,
+      "loss": 0.1852,
+      "step": 25244
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4449080616968045e-07,
+      "loss": 0.1607,
+      "step": 25245
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.442282117259065e-07,
+      "loss": 0.17,
+      "step": 25246
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.439657566338638e-07,
+      "loss": 0.1937,
+      "step": 25247
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.437034408973027e-07,
+      "loss": 0.1881,
+      "step": 25248
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4344126451996795e-07,
+      "loss": 0.1858,
+      "step": 25249
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.431792275056077e-07,
+      "loss": 0.1687,
+      "step": 25250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.429173298579612e-07,
+      "loss": 0.16,
+      "step": 25251
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.426555715807721e-07,
+      "loss": 0.1706,
+      "step": 25252
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.423939526777774e-07,
+      "loss": 0.1813,
+      "step": 25253
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4213247315271526e-07,
+      "loss": 0.1899,
+      "step": 25254
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.418711330093215e-07,
+      "loss": 0.1859,
+      "step": 25255
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4160993225132767e-07,
+      "loss": 0.1853,
+      "step": 25256
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.413488708824652e-07,
+      "loss": 0.1694,
+      "step": 25257
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4108794890646323e-07,
+      "loss": 0.159,
+      "step": 25258
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4082716632704896e-07,
+      "loss": 0.1685,
+      "step": 25259
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4056652314794593e-07,
+      "loss": 0.1755,
+      "step": 25260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.40306019372879e-07,
+      "loss": 0.163,
+      "step": 25261
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4004565500556966e-07,
+      "loss": 0.1629,
+      "step": 25262
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3978543004973486e-07,
+      "loss": 0.1642,
+      "step": 25263
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.395253445090928e-07,
+      "loss": 0.1734,
+      "step": 25264
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.392653983873583e-07,
+      "loss": 0.1752,
+      "step": 25265
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.39005591688245e-07,
+      "loss": 0.175,
+      "step": 25266
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3874592441546217e-07,
+      "loss": 0.1601,
+      "step": 25267
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3848639657272134e-07,
+      "loss": 0.1732,
+      "step": 25268
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3822700816372835e-07,
+      "loss": 0.185,
+      "step": 25269
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3796775919218918e-07,
+      "loss": 0.1698,
+      "step": 25270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.377086496618064e-07,
+      "loss": 0.1754,
+      "step": 25271
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.374496795762793e-07,
+      "loss": 0.1704,
+      "step": 25272
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3719084893931044e-07,
+      "loss": 0.1842,
+      "step": 25273
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3693215775459355e-07,
+      "loss": 0.1673,
+      "step": 25274
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.366736060258268e-07,
+      "loss": 0.1809,
+      "step": 25275
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3641519375670273e-07,
+      "loss": 0.1798,
+      "step": 25276
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.361569209509118e-07,
+      "loss": 0.1655,
+      "step": 25277
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3589878761214214e-07,
+      "loss": 0.1805,
+      "step": 25278
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.356407937440819e-07,
+      "loss": 0.2003,
+      "step": 25279
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.35382939350417e-07,
+      "loss": 0.1764,
+      "step": 25280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.35125224434829e-07,
+      "loss": 0.1633,
+      "step": 25281
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3486764900100157e-07,
+      "loss": 0.1535,
+      "step": 25282
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3461021305261177e-07,
+      "loss": 0.1877,
+      "step": 25283
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.343529165933378e-07,
+      "loss": 0.1667,
+      "step": 25284
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.340957596268545e-07,
+      "loss": 0.1746,
+      "step": 25285
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3383874215683445e-07,
+      "loss": 0.1842,
+      "step": 25286
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3358186418695138e-07,
+      "loss": 0.1631,
+      "step": 25287
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.333251257208724e-07,
+      "loss": 0.1821,
+      "step": 25288
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3306852676226455e-07,
+      "loss": 0.1703,
+      "step": 25289
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3281206731479377e-07,
+      "loss": 0.1935,
+      "step": 25290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3255574738212384e-07,
+      "loss": 0.1566,
+      "step": 25291
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.322995669679151e-07,
+      "loss": 0.1719,
+      "step": 25292
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3204352607582804e-07,
+      "loss": 0.18,
+      "step": 25293
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3178762470951744e-07,
+      "loss": 0.1719,
+      "step": 25294
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3153186287264152e-07,
+      "loss": 0.1732,
+      "step": 25295
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3127624056885177e-07,
+      "loss": 0.1851,
+      "step": 25296
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.310207578018009e-07,
+      "loss": 0.1723,
+      "step": 25297
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3076541457513813e-07,
+      "loss": 0.1886,
+      "step": 25298
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3051021089250836e-07,
+      "loss": 0.1694,
+      "step": 25299
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3025514675755976e-07,
+      "loss": 0.1598,
+      "step": 25300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.300002221739339e-07,
+      "loss": 0.1811,
+      "step": 25301
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2974543714527565e-07,
+      "loss": 0.1735,
+      "step": 25302
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.294907916752187e-07,
+      "loss": 0.1922,
+      "step": 25303
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2923628576740354e-07,
+      "loss": 0.175,
+      "step": 25304
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.289819194254661e-07,
+      "loss": 0.1846,
+      "step": 25305
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2872769265303795e-07,
+      "loss": 0.1499,
+      "step": 25306
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2847360545375286e-07,
+      "loss": 0.1894,
+      "step": 25307
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.282196578312379e-07,
+      "loss": 0.1911,
+      "step": 25308
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2796584978912128e-07,
+      "loss": 0.2131,
+      "step": 25309
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2771218133102901e-07,
+      "loss": 0.1672,
+      "step": 25310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2745865246058373e-07,
+      "loss": 0.1645,
+      "step": 25311
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.27205263181407e-07,
+      "loss": 0.1731,
+      "step": 25312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2695201349711816e-07,
+      "loss": 0.1851,
+      "step": 25313
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2669890341133537e-07,
+      "loss": 0.1659,
+      "step": 25314
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.264459329276736e-07,
+      "loss": 0.1875,
+      "step": 25315
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2619310204974653e-07,
+      "loss": 0.1584,
+      "step": 25316
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2594041078116468e-07,
+      "loss": 0.1853,
+      "step": 25317
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2568785912553848e-07,
+      "loss": 0.1862,
+      "step": 25318
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2543544708647504e-07,
+      "loss": 0.167,
+      "step": 25319
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2518317466757812e-07,
+      "loss": 0.1718,
+      "step": 25320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2493104187245485e-07,
+      "loss": 0.1818,
+      "step": 25321
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2467904870470458e-07,
+      "loss": 0.1619,
+      "step": 25322
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2442719516792555e-07,
+      "loss": 0.1778,
+      "step": 25323
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2417548126571818e-07,
+      "loss": 0.1681,
+      "step": 25324
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2392390700167633e-07,
+      "loss": 0.1843,
+      "step": 25325
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.236724723793915e-07,
+      "loss": 0.1743,
+      "step": 25326
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2342117740245972e-07,
+      "loss": 0.159,
+      "step": 25327
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2317002207446704e-07,
+      "loss": 0.1737,
+      "step": 25328
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2291900639900277e-07,
+      "loss": 0.1614,
+      "step": 25329
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2266813037964963e-07,
+      "loss": 0.1743,
+      "step": 25330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2241739401999474e-07,
+      "loss": 0.1706,
+      "step": 25331
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2216679732361745e-07,
+      "loss": 0.1773,
+      "step": 25332
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2191634029409826e-07,
+      "loss": 0.16,
+      "step": 25333
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2166602293501428e-07,
+      "loss": 0.1545,
+      "step": 25334
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.214158452499393e-07,
+      "loss": 0.1777,
+      "step": 25335
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2116580724245053e-07,
+      "loss": 0.1635,
+      "step": 25336
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2091590891611836e-07,
+      "loss": 0.1829,
+      "step": 25337
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2066615027450889e-07,
+      "loss": 0.1756,
+      "step": 25338
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.204165313211948e-07,
+      "loss": 0.1597,
+      "step": 25339
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2016705205973765e-07,
+      "loss": 0.1532,
+      "step": 25340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1991771249370463e-07,
+      "loss": 0.1785,
+      "step": 25341
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.196685126266529e-07,
+      "loss": 0.165,
+      "step": 25342
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1941945246214624e-07,
+      "loss": 0.1694,
+      "step": 25343
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1917053200374073e-07,
+      "loss": 0.1729,
+      "step": 25344
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1892175125498905e-07,
+      "loss": 0.1638,
+      "step": 25345
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1867311021944947e-07,
+      "loss": 0.1477,
+      "step": 25346
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.184246089006703e-07,
+      "loss": 0.1693,
+      "step": 25347
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.181762473022042e-07,
+      "loss": 0.1883,
+      "step": 25348
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1792802542759616e-07,
+      "loss": 0.1754,
+      "step": 25349
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.176799432803911e-07,
+      "loss": 0.1925,
+      "step": 25350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1743200086413508e-07,
+      "loss": 0.1705,
+      "step": 25351
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1718419818236857e-07,
+      "loss": 0.1839,
+      "step": 25352
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1693653523863212e-07,
+      "loss": 0.1774,
+      "step": 25353
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1668901203646176e-07,
+      "loss": 0.1728,
+      "step": 25354
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1644162857939465e-07,
+      "loss": 0.1769,
+      "step": 25355
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1619438487096356e-07,
+      "loss": 0.1846,
+      "step": 25356
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1594728091470008e-07,
+      "loss": 0.1704,
+      "step": 25357
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.157003167141336e-07,
+      "loss": 0.1909,
+      "step": 25358
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1545349227279244e-07,
+      "loss": 0.1709,
+      "step": 25359
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1520680759420266e-07,
+      "loss": 0.1829,
+      "step": 25360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1496026268188697e-07,
+      "loss": 0.1778,
+      "step": 25361
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1471385753936592e-07,
+      "loss": 0.1818,
+      "step": 25362
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.144675921701622e-07,
+      "loss": 0.1778,
+      "step": 25363
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1422146657779198e-07,
+      "loss": 0.1846,
+      "step": 25364
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1397548076577012e-07,
+      "loss": 0.1695,
+      "step": 25365
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1372963473761055e-07,
+      "loss": 0.1843,
+      "step": 25366
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1348392849682597e-07,
+      "loss": 0.1453,
+      "step": 25367
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1323836204692582e-07,
+      "loss": 0.1831,
+      "step": 25368
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1299293539141618e-07,
+      "loss": 0.1772,
+      "step": 25369
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1274764853380537e-07,
+      "loss": 0.1785,
+      "step": 25370
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.12502501477595e-07,
+      "loss": 0.1909,
+      "step": 25371
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1225749422628672e-07,
+      "loss": 0.1659,
+      "step": 25372
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1201262678338218e-07,
+      "loss": 0.1826,
+      "step": 25373
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1176789915237638e-07,
+      "loss": 0.1904,
+      "step": 25374
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1152331133676762e-07,
+      "loss": 0.1577,
+      "step": 25375
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1127886334004866e-07,
+      "loss": 0.1743,
+      "step": 25376
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1103455516570893e-07,
+      "loss": 0.1737,
+      "step": 25377
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.107903868172423e-07,
+      "loss": 0.1744,
+      "step": 25378
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1054635829813263e-07,
+      "loss": 0.1875,
+      "step": 25379
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1030246961186829e-07,
+      "loss": 0.1802,
+      "step": 25380
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1005872076193311e-07,
+      "loss": 0.1717,
+      "step": 25381
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0981511175180657e-07,
+      "loss": 0.186,
+      "step": 25382
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0957164258497031e-07,
+      "loss": 0.169,
+      "step": 25383
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0932831326489934e-07,
+      "loss": 0.1683,
+      "step": 25384
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0908512379507305e-07,
+      "loss": 0.168,
+      "step": 25385
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0884207417896206e-07,
+      "loss": 0.1894,
+      "step": 25386
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0859916442004135e-07,
+      "loss": 0.1677,
+      "step": 25387
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0835639452177813e-07,
+      "loss": 0.1879,
+      "step": 25388
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0811376448763964e-07,
+      "loss": 0.1618,
+      "step": 25389
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.078712743210942e-07,
+      "loss": 0.1585,
+      "step": 25390
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0762892402560353e-07,
+      "loss": 0.1912,
+      "step": 25391
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0738671360463038e-07,
+      "loss": 0.1685,
+      "step": 25392
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.071446430616342e-07,
+      "loss": 0.172,
+      "step": 25393
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0690271240007221e-07,
+      "loss": 0.1782,
+      "step": 25394
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0666092162340056e-07,
+      "loss": 0.1739,
+      "step": 25395
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0641927073507207e-07,
+      "loss": 0.1869,
+      "step": 25396
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.061777597385406e-07,
+      "loss": 0.1643,
+      "step": 25397
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0593638863725563e-07,
+      "loss": 0.1727,
+      "step": 25398
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.056951574346622e-07,
+      "loss": 0.1718,
+      "step": 25399
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.054540661342086e-07,
+      "loss": 0.1677,
+      "step": 25400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.052131147393366e-07,
+      "loss": 0.1607,
+      "step": 25401
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0497230325349005e-07,
+      "loss": 0.183,
+      "step": 25402
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0473163168010845e-07,
+      "loss": 0.1776,
+      "step": 25403
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0449110002262795e-07,
+      "loss": 0.1841,
+      "step": 25404
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.042507082844869e-07,
+      "loss": 0.1715,
+      "step": 25405
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0401045646911588e-07,
+      "loss": 0.162,
+      "step": 25406
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0377034457994994e-07,
+      "loss": 0.166,
+      "step": 25407
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.035303726204152e-07,
+      "loss": 0.1772,
+      "step": 25408
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0329054059394227e-07,
+      "loss": 0.1588,
+      "step": 25409
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.030508485039562e-07,
+      "loss": 0.1741,
+      "step": 25410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0281129635387974e-07,
+      "loss": 0.1827,
+      "step": 25411
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.025718841471369e-07,
+      "loss": 0.1663,
+      "step": 25412
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0233261188714491e-07,
+      "loss": 0.1576,
+      "step": 25413
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0209347957732328e-07,
+      "loss": 0.1767,
+      "step": 25414
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0185448722108703e-07,
+      "loss": 0.1712,
+      "step": 25415
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.016156348218501e-07,
+      "loss": 0.1635,
+      "step": 25416
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.013769223830242e-07,
+      "loss": 0.1867,
+      "step": 25417
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0113834990801885e-07,
+      "loss": 0.1829,
+      "step": 25418
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0089991740024239e-07,
+      "loss": 0.1882,
+      "step": 25419
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0066162486310104e-07,
+      "loss": 0.1774,
+      "step": 25420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.004234722999976e-07,
+      "loss": 0.2066,
+      "step": 25421
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.001854597143349e-07,
+      "loss": 0.162,
+      "step": 25422
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9994758710951024e-07,
+      "loss": 0.2003,
+      "step": 25423
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.997098544889242e-07,
+      "loss": 0.1736,
+      "step": 25424
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.994722618559708e-07,
+      "loss": 0.1903,
+      "step": 25425
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9923480921404503e-07,
+      "loss": 0.1627,
+      "step": 25426
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9899749656653755e-07,
+      "loss": 0.1914,
+      "step": 25427
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9876032391683897e-07,
+      "loss": 0.1789,
+      "step": 25428
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.985232912683377e-07,
+      "loss": 0.1627,
+      "step": 25429
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9828639862441767e-07,
+      "loss": 0.1863,
+      "step": 25430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9804964598846398e-07,
+      "loss": 0.1859,
+      "step": 25431
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9781303336385837e-07,
+      "loss": 0.1867,
+      "step": 25432
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9757656075397925e-07,
+      "loss": 0.1524,
+      "step": 25433
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.973402281622072e-07,
+      "loss": 0.1845,
+      "step": 25434
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9710403559191516e-07,
+      "loss": 0.1827,
+      "step": 25435
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9686798304647813e-07,
+      "loss": 0.1758,
+      "step": 25436
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9663207052926793e-07,
+      "loss": 0.1717,
+      "step": 25437
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9639629804365402e-07,
+      "loss": 0.1926,
+      "step": 25438
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9616066559300484e-07,
+      "loss": 0.1793,
+      "step": 25439
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9592517318068438e-07,
+      "loss": 0.1761,
+      "step": 25440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9568982081005994e-07,
+      "loss": 0.1751,
+      "step": 25441
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9545460848448993e-07,
+      "loss": 0.1785,
+      "step": 25442
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.95219536207335e-07,
+      "loss": 0.1757,
+      "step": 25443
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9498460398195363e-07,
+      "loss": 0.1718,
+      "step": 25444
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.947498118116997e-07,
+      "loss": 0.1713,
+      "step": 25445
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9451515969993063e-07,
+      "loss": 0.1989,
+      "step": 25446
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9428064764999589e-07,
+      "loss": 0.1747,
+      "step": 25447
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9404627566524502e-07,
+      "loss": 0.1583,
+      "step": 25448
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.938120437490265e-07,
+      "loss": 0.1634,
+      "step": 25449
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.935779519046843e-07,
+      "loss": 0.1938,
+      "step": 25450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9334400013556465e-07,
+      "loss": 0.1791,
+      "step": 25451
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9311018844500817e-07,
+      "loss": 0.155,
+      "step": 25452
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9287651683635555e-07,
+      "loss": 0.1741,
+      "step": 25453
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9264298531294412e-07,
+      "loss": 0.1522,
+      "step": 25454
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9240959387810788e-07,
+      "loss": 0.1713,
+      "step": 25455
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9217634253518303e-07,
+      "loss": 0.1917,
+      "step": 25456
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9194323128749913e-07,
+      "loss": 0.1893,
+      "step": 25457
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9171026013838912e-07,
+      "loss": 0.1748,
+      "step": 25458
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9147742909117805e-07,
+      "loss": 0.1794,
+      "step": 25459
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.912447381491933e-07,
+      "loss": 0.1696,
+      "step": 25460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9101218731575777e-07,
+      "loss": 0.1764,
+      "step": 25461
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.907797765941921e-07,
+      "loss": 0.1782,
+      "step": 25462
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9054750598781702e-07,
+      "loss": 0.1705,
+      "step": 25463
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9031537549995093e-07,
+      "loss": 0.1776,
+      "step": 25464
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.900833851339101e-07,
+      "loss": 0.1733,
+      "step": 25465
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8985153489300746e-07,
+      "loss": 0.1792,
+      "step": 25466
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8961982478055253e-07,
+      "loss": 0.1731,
+      "step": 25467
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8938825479985935e-07,
+      "loss": 0.1891,
+      "step": 25468
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8915682495423304e-07,
+      "loss": 0.1794,
+      "step": 25469
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8892553524697987e-07,
+      "loss": 0.1641,
+      "step": 25470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8869438568140275e-07,
+      "loss": 0.1874,
+      "step": 25471
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8846337626080457e-07,
+      "loss": 0.1821,
+      "step": 25472
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8823250698848604e-07,
+      "loss": 0.1868,
+      "step": 25473
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8800177786774232e-07,
+      "loss": 0.1632,
+      "step": 25474
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8777118890187075e-07,
+      "loss": 0.1828,
+      "step": 25475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8754074009416535e-07,
+      "loss": 0.1817,
+      "step": 25476
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.873104314479157e-07,
+      "loss": 0.1784,
+      "step": 25477
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.870802629664137e-07,
+      "loss": 0.1654,
+      "step": 25478
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.868502346529466e-07,
+      "loss": 0.164,
+      "step": 25479
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8662034651080075e-07,
+      "loss": 0.1833,
+      "step": 25480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8639059854325902e-07,
+      "loss": 0.1731,
+      "step": 25481
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8616099075360215e-07,
+      "loss": 0.169,
+      "step": 25482
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8593152314511198e-07,
+      "loss": 0.1722,
+      "step": 25483
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8570219572106363e-07,
+      "loss": 0.1742,
+      "step": 25484
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.854730084847356e-07,
+      "loss": 0.1637,
+      "step": 25485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.852439614394008e-07,
+      "loss": 0.1831,
+      "step": 25486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8501505458833114e-07,
+      "loss": 0.1634,
+      "step": 25487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8478628793479502e-07,
+      "loss": 0.1686,
+      "step": 25488
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.845576614820599e-07,
+      "loss": 0.1906,
+      "step": 25489
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8432917523339422e-07,
+      "loss": 0.164,
+      "step": 25490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8410082919205763e-07,
+      "loss": 0.1817,
+      "step": 25491
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8387262336131639e-07,
+      "loss": 0.1785,
+      "step": 25492
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.836445577444268e-07,
+      "loss": 0.1777,
+      "step": 25493
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8341663234464735e-07,
+      "loss": 0.1568,
+      "step": 25494
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8318884716523544e-07,
+      "loss": 0.1846,
+      "step": 25495
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.829612022094418e-07,
+      "loss": 0.1731,
+      "step": 25496
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8273369748052162e-07,
+      "loss": 0.183,
+      "step": 25497
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8250633298172116e-07,
+      "loss": 0.1749,
+      "step": 25498
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8227910871629117e-07,
+      "loss": 0.1716,
+      "step": 25499
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8205202468747463e-07,
+      "loss": 0.1639,
+      "step": 25500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8182508089851558e-07,
+      "loss": 0.1687,
+      "step": 25501
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8159827735265813e-07,
+      "loss": 0.1812,
+      "step": 25502
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8137161405313964e-07,
+      "loss": 0.1789,
+      "step": 25503
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8114509100319756e-07,
+      "loss": 0.1637,
+      "step": 25504
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.809187082060704e-07,
+      "loss": 0.1705,
+      "step": 25505
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8069246566498776e-07,
+      "loss": 0.1794,
+      "step": 25506
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.804663633831849e-07,
+      "loss": 0.1615,
+      "step": 25507
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.802404013638892e-07,
+      "loss": 0.2007,
+      "step": 25508
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.800145796103292e-07,
+      "loss": 0.1823,
+      "step": 25509
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7978889812573007e-07,
+      "loss": 0.1861,
+      "step": 25510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7956335691331595e-07,
+      "loss": 0.1564,
+      "step": 25511
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7933795597630866e-07,
+      "loss": 0.174,
+      "step": 25512
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7911269531792786e-07,
+      "loss": 0.1774,
+      "step": 25513
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7888757494139098e-07,
+      "loss": 0.1801,
+      "step": 25514
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7866259484991212e-07,
+      "loss": 0.1839,
+      "step": 25515
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7843775504670647e-07,
+      "loss": 0.1753,
+      "step": 25516
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.78213055534987e-07,
+      "loss": 0.1764,
+      "step": 25517
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7798849631796012e-07,
+      "loss": 0.1776,
+      "step": 25518
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7776407739883538e-07,
+      "loss": 0.1781,
+      "step": 25519
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7753979878081916e-07,
+      "loss": 0.1729,
+      "step": 25520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7731566046711224e-07,
+      "loss": 0.1758,
+      "step": 25521
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.770916624609187e-07,
+      "loss": 0.1881,
+      "step": 25522
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7686780476543707e-07,
+      "loss": 0.1903,
+      "step": 25523
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7664408738386707e-07,
+      "loss": 0.1682,
+      "step": 25524
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7642051031940054e-07,
+      "loss": 0.1852,
+      "step": 25525
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7619707357523498e-07,
+      "loss": 0.1788,
+      "step": 25526
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7597377715455887e-07,
+      "loss": 0.1737,
+      "step": 25527
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7575062106056195e-07,
+      "loss": 0.1712,
+      "step": 25528
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7552760529643388e-07,
+      "loss": 0.1886,
+      "step": 25529
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7530472986535873e-07,
+      "loss": 0.1758,
+      "step": 25530
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7508199477052068e-07,
+      "loss": 0.1737,
+      "step": 25531
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.748594000151005e-07,
+      "loss": 0.2004,
+      "step": 25532
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7463694560227784e-07,
+      "loss": 0.1653,
+      "step": 25533
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7441463153523242e-07,
+      "loss": 0.1776,
+      "step": 25534
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7419245781713723e-07,
+      "loss": 0.1672,
+      "step": 25535
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.739704244511653e-07,
+      "loss": 0.1718,
+      "step": 25536
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7374853144049074e-07,
+      "loss": 0.1964,
+      "step": 25537
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7352677878828105e-07,
+      "loss": 0.17,
+      "step": 25538
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.73305166497707e-07,
+      "loss": 0.1551,
+      "step": 25539
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.730836945719283e-07,
+      "loss": 0.1601,
+      "step": 25540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7286236301411352e-07,
+      "loss": 0.1617,
+      "step": 25541
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7264117182742234e-07,
+      "loss": 0.1903,
+      "step": 25542
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7242012101501447e-07,
+      "loss": 0.1754,
+      "step": 25543
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.721992105800463e-07,
+      "loss": 0.1638,
+      "step": 25544
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7197844052567525e-07,
+      "loss": 0.1955,
+      "step": 25545
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7175781085505328e-07,
+      "loss": 0.1735,
+      "step": 25546
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.715373215713334e-07,
+      "loss": 0.1795,
+      "step": 25547
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7131697267766313e-07,
+      "loss": 0.166,
+      "step": 25548
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.71096764177191e-07,
+      "loss": 0.1871,
+      "step": 25549
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7087669607306235e-07,
+      "loss": 0.199,
+      "step": 25550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7065676836842126e-07,
+      "loss": 0.1712,
+      "step": 25551
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.704369810664086e-07,
+      "loss": 0.1755,
+      "step": 25552
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7021733417016406e-07,
+      "loss": 0.1929,
+      "step": 25553
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6999782768282404e-07,
+      "loss": 0.1567,
+      "step": 25554
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6977846160752486e-07,
+      "loss": 0.1769,
+      "step": 25555
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6955923594740076e-07,
+      "loss": 0.1633,
+      "step": 25556
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.693401507055814e-07,
+      "loss": 0.168,
+      "step": 25557
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6912120588519766e-07,
+      "loss": 0.1798,
+      "step": 25558
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6890240148937586e-07,
+      "loss": 0.1495,
+      "step": 25559
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6868373752124245e-07,
+      "loss": 0.2006,
+      "step": 25560
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6846521398392046e-07,
+      "loss": 0.1696,
+      "step": 25561
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6824683088053072e-07,
+      "loss": 0.1583,
+      "step": 25562
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.68028588214193e-07,
+      "loss": 0.1839,
+      "step": 25563
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6781048598802474e-07,
+      "loss": 0.1986,
+      "step": 25564
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6759252420514238e-07,
+      "loss": 0.1608,
+      "step": 25565
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6737470286865786e-07,
+      "loss": 0.1571,
+      "step": 25566
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6715702198168203e-07,
+      "loss": 0.1797,
+      "step": 25567
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6693948154732575e-07,
+      "loss": 0.1518,
+      "step": 25568
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.667220815686954e-07,
+      "loss": 0.1687,
+      "step": 25569
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6650482204889628e-07,
+      "loss": 0.1904,
+      "step": 25570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.662877029910337e-07,
+      "loss": 0.1926,
+      "step": 25571
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6607072439820627e-07,
+      "loss": 0.1757,
+      "step": 25572
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6585388627351483e-07,
+      "loss": 0.1601,
+      "step": 25573
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6563718862005584e-07,
+      "loss": 0.1774,
+      "step": 25574
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.654206314409257e-07,
+      "loss": 0.1651,
+      "step": 25575
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6520421473921744e-07,
+      "loss": 0.1695,
+      "step": 25576
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6498793851802087e-07,
+      "loss": 0.154,
+      "step": 25577
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6477180278042793e-07,
+      "loss": 0.1691,
+      "step": 25578
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6455580752952394e-07,
+      "loss": 0.1697,
+      "step": 25579
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.643399527683942e-07,
+      "loss": 0.1713,
+      "step": 25580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6412423850012294e-07,
+      "loss": 0.1785,
+      "step": 25581
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.63908664727791e-07,
+      "loss": 0.1873,
+      "step": 25582
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6369323145447812e-07,
+      "loss": 0.1782,
+      "step": 25583
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6347793868325967e-07,
+      "loss": 0.1825,
+      "step": 25584
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6326278641721316e-07,
+      "loss": 0.1699,
+      "step": 25585
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6304777465941167e-07,
+      "loss": 0.165,
+      "step": 25586
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.628329034129239e-07,
+      "loss": 0.1899,
+      "step": 25587
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6261817268082292e-07,
+      "loss": 0.1976,
+      "step": 25588
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6240358246617183e-07,
+      "loss": 0.1678,
+      "step": 25589
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.621891327720393e-07,
+      "loss": 0.1847,
+      "step": 25590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6197482360148732e-07,
+      "loss": 0.1806,
+      "step": 25591
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6176065495757676e-07,
+      "loss": 0.1631,
+      "step": 25592
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6154662684336742e-07,
+      "loss": 0.1511,
+      "step": 25593
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6133273926191573e-07,
+      "loss": 0.1771,
+      "step": 25594
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6111899221627702e-07,
+      "loss": 0.1907,
+      "step": 25595
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6090538570950443e-07,
+      "loss": 0.1593,
+      "step": 25596
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6069191974464992e-07,
+      "loss": 0.1655,
+      "step": 25597
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.604785943247611e-07,
+      "loss": 0.1784,
+      "step": 25598
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.602654094528866e-07,
+      "loss": 0.1851,
+      "step": 25599
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.600523651320718e-07,
+      "loss": 0.1785,
+      "step": 25600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5983946136535756e-07,
+      "loss": 0.1738,
+      "step": 25601
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5962669815578812e-07,
+      "loss": 0.1922,
+      "step": 25602
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5941407550639998e-07,
+      "loss": 0.1763,
+      "step": 25603
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5920159342023068e-07,
+      "loss": 0.1733,
+      "step": 25604
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5898925190031778e-07,
+      "loss": 0.1698,
+      "step": 25605
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5877705094969108e-07,
+      "loss": 0.1891,
+      "step": 25606
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5856499057138374e-07,
+      "loss": 0.1801,
+      "step": 25607
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5835307076842443e-07,
+      "loss": 0.1838,
+      "step": 25608
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.581412915438385e-07,
+      "loss": 0.1695,
+      "step": 25609
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5792965290065354e-07,
+      "loss": 0.1813,
+      "step": 25610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5771815484189158e-07,
+      "loss": 0.1616,
+      "step": 25611
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5750679737057351e-07,
+      "loss": 0.1577,
+      "step": 25612
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5729558048971805e-07,
+      "loss": 0.1707,
+      "step": 25613
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5708450420234278e-07,
+      "loss": 0.168,
+      "step": 25614
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5687356851146306e-07,
+      "loss": 0.1864,
+      "step": 25615
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5666277342009094e-07,
+      "loss": 0.1901,
+      "step": 25616
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5645211893123846e-07,
+      "loss": 0.1768,
+      "step": 25617
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5624160504791319e-07,
+      "loss": 0.1743,
+      "step": 25618
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5603123177312384e-07,
+      "loss": 0.1694,
+      "step": 25619
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.558209991098747e-07,
+      "loss": 0.1657,
+      "step": 25620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5561090706116778e-07,
+      "loss": 0.1777,
+      "step": 25621
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5540095563000513e-07,
+      "loss": 0.1796,
+      "step": 25622
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5519114481938547e-07,
+      "loss": 0.1913,
+      "step": 25623
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5498147463230528e-07,
+      "loss": 0.1801,
+      "step": 25624
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5477194507176107e-07,
+      "loss": 0.1706,
+      "step": 25625
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5456255614074377e-07,
+      "loss": 0.1784,
+      "step": 25626
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5435330784224545e-07,
+      "loss": 0.202,
+      "step": 25627
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5414420017925368e-07,
+      "loss": 0.1682,
+      "step": 25628
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5393523315475721e-07,
+      "loss": 0.1646,
+      "step": 25629
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5372640677174032e-07,
+      "loss": 0.1759,
+      "step": 25630
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5351772103318617e-07,
+      "loss": 0.1562,
+      "step": 25631
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.533091759420746e-07,
+      "loss": 0.1991,
+      "step": 25632
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5310077150138435e-07,
+      "loss": 0.156,
+      "step": 25633
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5289250771409303e-07,
+      "loss": 0.186,
+      "step": 25634
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5268438458317603e-07,
+      "loss": 0.1672,
+      "step": 25635
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.524764021116054e-07,
+      "loss": 0.183,
+      "step": 25636
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5226856030235216e-07,
+      "loss": 0.1982,
+      "step": 25637
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.52060859158385e-07,
+      "loss": 0.1908,
+      "step": 25638
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.518532986826704e-07,
+      "loss": 0.1695,
+      "step": 25639
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5164587887817384e-07,
+      "loss": 0.1643,
+      "step": 25640
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5143859974785734e-07,
+      "loss": 0.1826,
+      "step": 25641
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5123146129468302e-07,
+      "loss": 0.1553,
+      "step": 25642
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5102446352160848e-07,
+      "loss": 0.1552,
+      "step": 25643
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5081760643159248e-07,
+      "loss": 0.1674,
+      "step": 25644
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.50610890027586e-07,
+      "loss": 0.1774,
+      "step": 25645
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5040431431254444e-07,
+      "loss": 0.1631,
+      "step": 25646
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5019787928941876e-07,
+      "loss": 0.1671,
+      "step": 25647
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.499915849611555e-07,
+      "loss": 0.1636,
+      "step": 25648
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4978543133070344e-07,
+      "loss": 0.1966,
+      "step": 25649
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4957941840100575e-07,
+      "loss": 0.1703,
+      "step": 25650
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4937354617500788e-07,
+      "loss": 0.1609,
+      "step": 25651
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4916781465564746e-07,
+      "loss": 0.1601,
+      "step": 25652
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4896222384586433e-07,
+      "loss": 0.1751,
+      "step": 25653
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.487567737485951e-07,
+      "loss": 0.1862,
+      "step": 25654
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4855146436677515e-07,
+      "loss": 0.1941,
+      "step": 25655
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4834629570333548e-07,
+      "loss": 0.1776,
+      "step": 25656
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4814126776120928e-07,
+      "loss": 0.1769,
+      "step": 25657
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4793638054332315e-07,
+      "loss": 0.171,
+      "step": 25658
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.477316340526036e-07,
+      "loss": 0.1702,
+      "step": 25659
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4752702829197497e-07,
+      "loss": 0.1797,
+      "step": 25660
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4732256326436267e-07,
+      "loss": 0.1695,
+      "step": 25661
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.471182389726833e-07,
+      "loss": 0.1753,
+      "step": 25662
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4691405541985893e-07,
+      "loss": 0.1674,
+      "step": 25663
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.467100126088039e-07,
+      "loss": 0.1615,
+      "step": 25664
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4650611054243368e-07,
+      "loss": 0.1641,
+      "step": 25665
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4630234922366038e-07,
+      "loss": 0.1609,
+      "step": 25666
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4609872865539498e-07,
+      "loss": 0.1649,
+      "step": 25667
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4589524884054629e-07,
+      "loss": 0.1742,
+      "step": 25668
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4569190978201975e-07,
+      "loss": 0.1769,
+      "step": 25669
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4548871148272082e-07,
+      "loss": 0.1797,
+      "step": 25670
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4528565394555162e-07,
+      "loss": 0.1887,
+      "step": 25671
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4508273717341202e-07,
+      "loss": 0.204,
+      "step": 25672
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.44879961169202e-07,
+      "loss": 0.1677,
+      "step": 25673
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4467732593581585e-07,
+      "loss": 0.1858,
+      "step": 25674
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4447483147615126e-07,
+      "loss": 0.176,
+      "step": 25675
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4427247779309706e-07,
+      "loss": 0.1742,
+      "step": 25676
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4407026488954534e-07,
+      "loss": 0.1852,
+      "step": 25677
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4386819276838494e-07,
+      "loss": 0.1834,
+      "step": 25678
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4366626143250239e-07,
+      "loss": 0.1689,
+      "step": 25679
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.434644708847799e-07,
+      "loss": 0.1755,
+      "step": 25680
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4326282112810287e-07,
+      "loss": 0.1863,
+      "step": 25681
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4306131216534902e-07,
+      "loss": 0.1912,
+      "step": 25682
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.428599439993994e-07,
+      "loss": 0.1795,
+      "step": 25683
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4265871663312836e-07,
+      "loss": 0.1667,
+      "step": 25684
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4245763006941138e-07,
+      "loss": 0.1703,
+      "step": 25685
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.422566843111195e-07,
+      "loss": 0.1785,
+      "step": 25686
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4205587936112265e-07,
+      "loss": 0.1902,
+      "step": 25687
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4185521522229184e-07,
+      "loss": 0.1727,
+      "step": 25688
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4165469189749038e-07,
+      "loss": 0.1718,
+      "step": 25689
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.414543093895848e-07,
+      "loss": 0.1617,
+      "step": 25690
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4125406770143625e-07,
+      "loss": 0.1725,
+      "step": 25691
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4105396683590455e-07,
+      "loss": 0.1738,
+      "step": 25692
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.408540067958497e-07,
+      "loss": 0.1622,
+      "step": 25693
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4065418758412607e-07,
+      "loss": 0.1825,
+      "step": 25694
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4045450920358917e-07,
+      "loss": 0.1775,
+      "step": 25695
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.402549716570911e-07,
+      "loss": 0.1588,
+      "step": 25696
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.400555749474808e-07,
+      "loss": 0.1742,
+      "step": 25697
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.398563190776081e-07,
+      "loss": 0.1666,
+      "step": 25698
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3965720405031745e-07,
+      "loss": 0.1697,
+      "step": 25699
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3945822986845437e-07,
+      "loss": 0.1756,
+      "step": 25700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3925939653485988e-07,
+      "loss": 0.1808,
+      "step": 25701
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3906070405237614e-07,
+      "loss": 0.1804,
+      "step": 25702
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3886215242383981e-07,
+      "loss": 0.1574,
+      "step": 25703
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.386637416520864e-07,
+      "loss": 0.1634,
+      "step": 25704
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.384654717399514e-07,
+      "loss": 0.1634,
+      "step": 25705
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3826734269026475e-07,
+      "loss": 0.1909,
+      "step": 25706
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3806935450585978e-07,
+      "loss": 0.186,
+      "step": 25707
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.378715071895631e-07,
+      "loss": 0.1821,
+      "step": 25708
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.376738007441991e-07,
+      "loss": 0.1825,
+      "step": 25709
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3747623517259446e-07,
+      "loss": 0.1935,
+      "step": 25710
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3727881047756908e-07,
+      "loss": 0.1745,
+      "step": 25711
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.370815266619452e-07,
+      "loss": 0.1613,
+      "step": 25712
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.368843837285383e-07,
+      "loss": 0.1663,
+      "step": 25713
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3668738168016504e-07,
+      "loss": 0.1732,
+      "step": 25714
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3649052051964207e-07,
+      "loss": 0.1808,
+      "step": 25715
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3629380024977713e-07,
+      "loss": 0.1663,
+      "step": 25716
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3609722087338352e-07,
+      "loss": 0.165,
+      "step": 25717
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3590078239326787e-07,
+      "loss": 0.1737,
+      "step": 25718
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3570448481223574e-07,
+      "loss": 0.1756,
+      "step": 25719
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3550832813309156e-07,
+      "loss": 0.1733,
+      "step": 25720
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3531231235863752e-07,
+      "loss": 0.1642,
+      "step": 25721
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3511643749167248e-07,
+      "loss": 0.1719,
+      "step": 25722
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3492070353499641e-07,
+      "loss": 0.1546,
+      "step": 25723
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3472511049140268e-07,
+      "loss": 0.1578,
+      "step": 25724
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.345296583636868e-07,
+      "loss": 0.1748,
+      "step": 25725
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.343343471546399e-07,
+      "loss": 0.1654,
+      "step": 25726
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3413917686705192e-07,
+      "loss": 0.1556,
+      "step": 25727
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3394414750371064e-07,
+      "loss": 0.1815,
+      "step": 25728
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3374925906740278e-07,
+      "loss": 0.177,
+      "step": 25729
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.335545115609105e-07,
+      "loss": 0.1771,
+      "step": 25730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3335990498701601e-07,
+      "loss": 0.1677,
+      "step": 25731
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3316543934850045e-07,
+      "loss": 0.16,
+      "step": 25732
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.329711146481394e-07,
+      "loss": 0.1777,
+      "step": 25733
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.327769308887117e-07,
+      "loss": 0.1807,
+      "step": 25734
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3258288807298847e-07,
+      "loss": 0.1657,
+      "step": 25735
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3238898620374195e-07,
+      "loss": 0.1732,
+      "step": 25736
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3219522528374108e-07,
+      "loss": 0.1778,
+      "step": 25737
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.320016053157558e-07,
+      "loss": 0.1952,
+      "step": 25738
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3180812630254947e-07,
+      "loss": 0.1842,
+      "step": 25739
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3161478824688655e-07,
+      "loss": 0.1791,
+      "step": 25740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.314215911515304e-07,
+      "loss": 0.1576,
+      "step": 25741
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.312285350192377e-07,
+      "loss": 0.1568,
+      "step": 25742
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3103561985276848e-07,
+      "loss": 0.1643,
+      "step": 25743
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3084284565487714e-07,
+      "loss": 0.1595,
+      "step": 25744
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.306502124283171e-07,
+      "loss": 0.1716,
+      "step": 25745
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3045772017584058e-07,
+      "loss": 0.1734,
+      "step": 25746
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.302653689001965e-07,
+      "loss": 0.1695,
+      "step": 25747
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3007315860413372e-07,
+      "loss": 0.1617,
+      "step": 25748
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2988108929039567e-07,
+      "loss": 0.1793,
+      "step": 25749
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.29689160961729e-07,
+      "loss": 0.1764,
+      "step": 25750
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2949737362087156e-07,
+      "loss": 0.1947,
+      "step": 25751
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2930572727056556e-07,
+      "loss": 0.1826,
+      "step": 25752
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2911422191354661e-07,
+      "loss": 0.156,
+      "step": 25753
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2892285755255141e-07,
+      "loss": 0.162,
+      "step": 25754
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.287316341903133e-07,
+      "loss": 0.1684,
+      "step": 25755
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.285405518295635e-07,
+      "loss": 0.188,
+      "step": 25756
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2834961047303195e-07,
+      "loss": 0.1533,
+      "step": 25757
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2815881012344433e-07,
+      "loss": 0.1789,
+      "step": 25758
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.279681507835284e-07,
+      "loss": 0.1508,
+      "step": 25759
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2777763245600538e-07,
+      "loss": 0.1936,
+      "step": 25760
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.275872551435986e-07,
+      "loss": 0.1821,
+      "step": 25761
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2739701884902588e-07,
+      "loss": 0.1654,
+      "step": 25762
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2720692357500508e-07,
+      "loss": 0.184,
+      "step": 25763
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.270169693242529e-07,
+      "loss": 0.178,
+      "step": 25764
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2682715609947937e-07,
+      "loss": 0.1642,
+      "step": 25765
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2663748390339902e-07,
+      "loss": 0.1807,
+      "step": 25766
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2644795273871967e-07,
+      "loss": 0.1485,
+      "step": 25767
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2625856260814805e-07,
+      "loss": 0.1754,
+      "step": 25768
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.26069313514392e-07,
+      "loss": 0.1726,
+      "step": 25769
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2588020546015045e-07,
+      "loss": 0.1741,
+      "step": 25770
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2569123844812903e-07,
+      "loss": 0.1763,
+      "step": 25771
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2550241248102447e-07,
+      "loss": 0.1619,
+      "step": 25772
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2531372756153458e-07,
+      "loss": 0.1775,
+      "step": 25773
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2512518369235394e-07,
+      "loss": 0.173,
+      "step": 25774
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.24936780876177e-07,
+      "loss": 0.1793,
+      "step": 25775
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2474851911569496e-07,
+      "loss": 0.1808,
+      "step": 25776
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2456039841359458e-07,
+      "loss": 0.1777,
+      "step": 25777
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2437241877256478e-07,
+      "loss": 0.1897,
+      "step": 25778
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2418458019529012e-07,
+      "loss": 0.1697,
+      "step": 25779
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.239968826844551e-07,
+      "loss": 0.1781,
+      "step": 25780
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2380932624273978e-07,
+      "loss": 0.183,
+      "step": 25781
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2362191087282204e-07,
+      "loss": 0.1654,
+      "step": 25782
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2343463657738085e-07,
+      "loss": 0.1877,
+      "step": 25783
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2324750335909074e-07,
+      "loss": 0.1729,
+      "step": 25784
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.230605112206229e-07,
+      "loss": 0.1833,
+      "step": 25785
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2287366016465185e-07,
+      "loss": 0.1813,
+      "step": 25786
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2268695019384325e-07,
+      "loss": 0.1666,
+      "step": 25787
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2250038131086607e-07,
+      "loss": 0.1672,
+      "step": 25788
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2231395351838372e-07,
+      "loss": 0.1453,
+      "step": 25789
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2212766681906074e-07,
+      "loss": 0.1722,
+      "step": 25790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2194152121555835e-07,
+      "loss": 0.1796,
+      "step": 25791
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.217555167105322e-07,
+      "loss": 0.1719,
+      "step": 25792
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.215696533066435e-07,
+      "loss": 0.1819,
+      "step": 25793
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2138393100654345e-07,
+      "loss": 0.169,
+      "step": 25794
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.211983498128877e-07,
+      "loss": 0.1681,
+      "step": 25795
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2101290972832524e-07,
+      "loss": 0.1711,
+      "step": 25796
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.208276107555051e-07,
+      "loss": 0.1721,
+      "step": 25797
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.206424528970751e-07,
+      "loss": 0.1629,
+      "step": 25798
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2045743615567984e-07,
+      "loss": 0.1576,
+      "step": 25799
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2027256053396163e-07,
+      "loss": 0.1762,
+      "step": 25800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.200878260345617e-07,
+      "loss": 0.1778,
+      "step": 25801
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1990323266011793e-07,
+      "loss": 0.1779,
+      "step": 25802
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1971878041326823e-07,
+      "loss": 0.1585,
+      "step": 25803
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1953446929664602e-07,
+      "loss": 0.1894,
+      "step": 25804
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1935029931288478e-07,
+      "loss": 0.1858,
+      "step": 25805
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1916627046461571e-07,
+      "loss": 0.1764,
+      "step": 25806
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1898238275446672e-07,
+      "loss": 0.1678,
+      "step": 25807
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.187986361850646e-07,
+      "loss": 0.187,
+      "step": 25808
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.186150307590328e-07,
+      "loss": 0.1567,
+      "step": 25809
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1843156647899701e-07,
+      "loss": 0.1685,
+      "step": 25810
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1824824334757512e-07,
+      "loss": 0.1651,
+      "step": 25811
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1806506136738616e-07,
+      "loss": 0.1703,
+      "step": 25812
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1788202054104803e-07,
+      "loss": 0.189,
+      "step": 25813
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1769912087117419e-07,
+      "loss": 0.194,
+      "step": 25814
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1751636236037812e-07,
+      "loss": 0.189,
+      "step": 25815
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1733374501126771e-07,
+      "loss": 0.1946,
+      "step": 25816
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1715126882645533e-07,
+      "loss": 0.1525,
+      "step": 25817
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1696893380854446e-07,
+      "loss": 0.1533,
+      "step": 25818
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1678673996014078e-07,
+      "loss": 0.1704,
+      "step": 25819
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1660468728384666e-07,
+      "loss": 0.1765,
+      "step": 25820
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1642277578226224e-07,
+      "loss": 0.1693,
+      "step": 25821
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1624100545798766e-07,
+      "loss": 0.1703,
+      "step": 25822
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1605937631361642e-07,
+      "loss": 0.1732,
+      "step": 25823
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1587788835174419e-07,
+      "loss": 0.1795,
+      "step": 25824
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1569654157496447e-07,
+      "loss": 0.1606,
+      "step": 25825
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1551533598586629e-07,
+      "loss": 0.1569,
+      "step": 25826
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.153342715870387e-07,
+      "loss": 0.1782,
+      "step": 25827
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1515334838106851e-07,
+      "loss": 0.1684,
+      "step": 25828
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.149725663705381e-07,
+      "loss": 0.1749,
+      "step": 25829
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1479192555803209e-07,
+      "loss": 0.1628,
+      "step": 25830
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1461142594612729e-07,
+      "loss": 0.1772,
+      "step": 25831
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1443106753740607e-07,
+      "loss": 0.172,
+      "step": 25832
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1425085033444194e-07,
+      "loss": 0.1835,
+      "step": 25833
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1407077433981173e-07,
+      "loss": 0.1607,
+      "step": 25834
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1389083955608448e-07,
+      "loss": 0.1608,
+      "step": 25835
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.137110459858326e-07,
+      "loss": 0.1703,
+      "step": 25836
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1353139363162291e-07,
+      "loss": 0.1546,
+      "step": 25837
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1335188249602225e-07,
+      "loss": 0.1821,
+      "step": 25838
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1317251258159524e-07,
+      "loss": 0.1694,
+      "step": 25839
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1299328389090424e-07,
+      "loss": 0.1922,
+      "step": 25840
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1281419642650726e-07,
+      "loss": 0.1687,
+      "step": 25841
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1263525019096443e-07,
+      "loss": 0.1614,
+      "step": 25842
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1245644518683041e-07,
+      "loss": 0.1909,
+      "step": 25843
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.122777814166609e-07,
+      "loss": 0.1999,
+      "step": 25844
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.120992588830061e-07,
+      "loss": 0.1775,
+      "step": 25845
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.119208775884173e-07,
+      "loss": 0.1692,
+      "step": 25846
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1174263753544246e-07,
+      "loss": 0.1822,
+      "step": 25847
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1156453872662732e-07,
+      "loss": 0.1939,
+      "step": 25848
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1138658116451539e-07,
+      "loss": 0.173,
+      "step": 25849
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1120876485164911e-07,
+      "loss": 0.163,
+      "step": 25850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1103108979056865e-07,
+      "loss": 0.168,
+      "step": 25851
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1085355598381087e-07,
+      "loss": 0.1567,
+      "step": 25852
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1067616343391263e-07,
+      "loss": 0.1759,
+      "step": 25853
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1049891214340858e-07,
+      "loss": 0.1806,
+      "step": 25854
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.103218021148289e-07,
+      "loss": 0.1802,
+      "step": 25855
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.101448333507038e-07,
+      "loss": 0.1782,
+      "step": 25856
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0996800585356238e-07,
+      "loss": 0.1774,
+      "step": 25857
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0979131962592815e-07,
+      "loss": 0.178,
+      "step": 25858
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0961477467032689e-07,
+      "loss": 0.1644,
+      "step": 25859
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.094383709892799e-07,
+      "loss": 0.1747,
+      "step": 25860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0926210858530628e-07,
+      "loss": 0.1602,
+      "step": 25861
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0908598746092514e-07,
+      "loss": 0.1558,
+      "step": 25862
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0891000761865001e-07,
+      "loss": 0.1712,
+      "step": 25863
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0873416906099665e-07,
+      "loss": 0.1908,
+      "step": 25864
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0855847179047641e-07,
+      "loss": 0.1889,
+      "step": 25865
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0838291580959836e-07,
+      "loss": 0.1616,
+      "step": 25866
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0820750112086942e-07,
+      "loss": 0.1434,
+      "step": 25867
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0803222772679756e-07,
+      "loss": 0.2001,
+      "step": 25868
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0785709562988411e-07,
+      "loss": 0.1913,
+      "step": 25869
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0768210483263153e-07,
+      "loss": 0.1779,
+      "step": 25870
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0750725533753892e-07,
+      "loss": 0.165,
+      "step": 25871
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0733254714710429e-07,
+      "loss": 0.1725,
+      "step": 25872
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.071579802638234e-07,
+      "loss": 0.1587,
+      "step": 25873
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0698355469018984e-07,
+      "loss": 0.1766,
+      "step": 25874
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0680927042869493e-07,
+      "loss": 0.1858,
+      "step": 25875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0663512748182781e-07,
+      "loss": 0.1739,
+      "step": 25876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0646112585207535e-07,
+      "loss": 0.1651,
+      "step": 25877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0628726554192448e-07,
+      "loss": 0.1678,
+      "step": 25878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0611354655385875e-07,
+      "loss": 0.1546,
+      "step": 25879
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0593996889035729e-07,
+      "loss": 0.1884,
+      "step": 25880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0576653255390256e-07,
+      "loss": 0.162,
+      "step": 25881
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0559323754696926e-07,
+      "loss": 0.1772,
+      "step": 25882
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0542008387203429e-07,
+      "loss": 0.1684,
+      "step": 25883
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0524707153157009e-07,
+      "loss": 0.1646,
+      "step": 25884
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0507420052804807e-07,
+      "loss": 0.1854,
+      "step": 25885
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0490147086393842e-07,
+      "loss": 0.1648,
+      "step": 25886
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.047288825417081e-07,
+      "loss": 0.1753,
+      "step": 25887
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0455643556382178e-07,
+      "loss": 0.1886,
+      "step": 25888
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0438412993274305e-07,
+      "loss": 0.1739,
+      "step": 25889
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0421196565093217e-07,
+      "loss": 0.1707,
+      "step": 25890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.040399427208505e-07,
+      "loss": 0.1781,
+      "step": 25891
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0386806114495384e-07,
+      "loss": 0.1707,
+      "step": 25892
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0369632092569692e-07,
+      "loss": 0.1792,
+      "step": 25893
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0352472206553443e-07,
+      "loss": 0.1789,
+      "step": 25894
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0335326456691553e-07,
+      "loss": 0.1638,
+      "step": 25895
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0318194843229156e-07,
+      "loss": 0.1718,
+      "step": 25896
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0301077366410728e-07,
+      "loss": 0.1714,
+      "step": 25897
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0283974026480848e-07,
+      "loss": 0.1801,
+      "step": 25898
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0266884823683876e-07,
+      "loss": 0.1736,
+      "step": 25899
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.024980975826395e-07,
+      "loss": 0.163,
+      "step": 25900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0232748830464989e-07,
+      "loss": 0.1723,
+      "step": 25901
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0215702040530462e-07,
+      "loss": 0.1726,
+      "step": 25902
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0198669388704064e-07,
+      "loss": 0.1718,
+      "step": 25903
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0181650875229044e-07,
+      "loss": 0.1519,
+      "step": 25904
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0164646500348652e-07,
+      "loss": 0.1751,
+      "step": 25905
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0147656264305472e-07,
+      "loss": 0.1829,
+      "step": 25906
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.013068016734231e-07,
+      "loss": 0.1742,
+      "step": 25907
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0113718209701861e-07,
+      "loss": 0.168,
+      "step": 25908
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0096770391626153e-07,
+      "loss": 0.1687,
+      "step": 25909
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0079836713357327e-07,
+      "loss": 0.1752,
+      "step": 25910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.00629171751373e-07,
+      "loss": 0.1833,
+      "step": 25911
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0046011777207765e-07,
+      "loss": 0.1692,
+      "step": 25912
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0029120519810198e-07,
+      "loss": 0.1808,
+      "step": 25913
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0012243403185851e-07,
+      "loss": 0.1677,
+      "step": 25914
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.995380427575752e-08,
+      "loss": 0.1726,
+      "step": 25915
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.978531593220819e-08,
+      "loss": 0.1677,
+      "step": 25916
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.96169690036175e-08,
+      "loss": 0.1689,
+      "step": 25917
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.944876349239018e-08,
+      "loss": 0.1754,
+      "step": 25918
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.928069940092766e-08,
+      "loss": 0.1818,
+      "step": 25919
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.911277673163244e-08,
+      "loss": 0.1716,
+      "step": 25920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.894499548690262e-08,
+      "loss": 0.1663,
+      "step": 25921
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.877735566913294e-08,
+      "loss": 0.1663,
+      "step": 25922
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.860985728072037e-08,
+      "loss": 0.1637,
+      "step": 25923
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.844250032405633e-08,
+      "loss": 0.1937,
+      "step": 25924
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.827528480153114e-08,
+      "loss": 0.1846,
+      "step": 25925
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.81082107155329e-08,
+      "loss": 0.1672,
+      "step": 25926
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.794127806844966e-08,
+      "loss": 0.1749,
+      "step": 25927
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.777448686266621e-08,
+      "loss": 0.1764,
+      "step": 25928
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.760783710056176e-08,
+      "loss": 0.1905,
+      "step": 25929
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.744132878451995e-08,
+      "loss": 0.1778,
+      "step": 25930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.727496191691776e-08,
+      "loss": 0.1655,
+      "step": 25931
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.71087365001322e-08,
+      "loss": 0.1664,
+      "step": 25932
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.6942652536538e-08,
+      "loss": 0.1739,
+      "step": 25933
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.677671002850664e-08,
+      "loss": 0.1756,
+      "step": 25934
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.661090897840841e-08,
+      "loss": 0.1791,
+      "step": 25935
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.644524938861144e-08,
+      "loss": 0.1947,
+      "step": 25936
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.627973126148271e-08,
+      "loss": 0.1564,
+      "step": 25937
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.61143545993859e-08,
+      "loss": 0.1621,
+      "step": 25938
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.594911940468355e-08,
+      "loss": 0.1754,
+      "step": 25939
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.57840256797371e-08,
+      "loss": 0.1589,
+      "step": 25940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.561907342690247e-08,
+      "loss": 0.1699,
+      "step": 25941
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.545426264853664e-08,
+      "loss": 0.1786,
+      "step": 25942
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.528959334699439e-08,
+      "loss": 0.177,
+      "step": 25943
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.51250655246283e-08,
+      "loss": 0.1629,
+      "step": 25944
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.49606791837876e-08,
+      "loss": 0.1711,
+      "step": 25945
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.479643432681928e-08,
+      "loss": 0.1741,
+      "step": 25946
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.46323309560726e-08,
+      "loss": 0.1806,
+      "step": 25947
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.446836907388901e-08,
+      "loss": 0.171,
+      "step": 25948
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.430454868261108e-08,
+      "loss": 0.1599,
+      "step": 25949
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.414086978457804e-08,
+      "loss": 0.1603,
+      "step": 25950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.397733238213025e-08,
+      "loss": 0.1815,
+      "step": 25951
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.38139364776014e-08,
+      "loss": 0.1755,
+      "step": 25952
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.365068207332629e-08,
+      "loss": 0.1776,
+      "step": 25953
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.348756917163637e-08,
+      "loss": 0.1592,
+      "step": 25954
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.332459777486314e-08,
+      "loss": 0.177,
+      "step": 25955
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.31617678853325e-08,
+      "loss": 0.1559,
+      "step": 25956
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.299907950537145e-08,
+      "loss": 0.1732,
+      "step": 25957
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.283653263730375e-08,
+      "loss": 0.1805,
+      "step": 25958
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.267412728345082e-08,
+      "loss": 0.1539,
+      "step": 25959
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.251186344613305e-08,
+      "loss": 0.1717,
+      "step": 25960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.23497411276686e-08,
+      "loss": 0.1706,
+      "step": 25961
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.218776033037114e-08,
+      "loss": 0.1895,
+      "step": 25962
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.202592105655661e-08,
+      "loss": 0.1722,
+      "step": 25963
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.18642233085354e-08,
+      "loss": 0.1813,
+      "step": 25964
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.170266708861786e-08,
+      "loss": 0.1649,
+      "step": 25965
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.154125239911105e-08,
+      "loss": 0.1816,
+      "step": 25966
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.137997924232201e-08,
+      "loss": 0.1771,
+      "step": 25967
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.121884762055222e-08,
+      "loss": 0.1574,
+      "step": 25968
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.105785753610541e-08,
+      "loss": 0.1962,
+      "step": 25969
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.089700899127973e-08,
+      "loss": 0.1794,
+      "step": 25970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.073630198837336e-08,
+      "loss": 0.1695,
+      "step": 25971
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.057573652968222e-08,
+      "loss": 0.1688,
+      "step": 25972
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.041531261749892e-08,
+      "loss": 0.1625,
+      "step": 25973
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.025503025411608e-08,
+      "loss": 0.1825,
+      "step": 25974
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.009488944182187e-08,
+      "loss": 0.152,
+      "step": 25975
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.993489018290447e-08,
+      "loss": 0.1616,
+      "step": 25976
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.977503247964869e-08,
+      "loss": 0.1574,
+      "step": 25977
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.961531633433828e-08,
+      "loss": 0.1686,
+      "step": 25978
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.945574174925586e-08,
+      "loss": 0.1815,
+      "step": 25979
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.92963087266796e-08,
+      "loss": 0.1686,
+      "step": 25980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.913701726888657e-08,
+      "loss": 0.1696,
+      "step": 25981
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.897786737815272e-08,
+      "loss": 0.1854,
+      "step": 25982
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.881885905675069e-08,
+      "loss": 0.1659,
+      "step": 25983
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.86599923069531e-08,
+      "loss": 0.1601,
+      "step": 25984
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.850126713102813e-08,
+      "loss": 0.1826,
+      "step": 25985
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.834268353124287e-08,
+      "loss": 0.1845,
+      "step": 25986
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.818424150986327e-08,
+      "loss": 0.1763,
+      "step": 25987
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.802594106915196e-08,
+      "loss": 0.1696,
+      "step": 25988
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.786778221137049e-08,
+      "loss": 0.1855,
+      "step": 25989
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.770976493877703e-08,
+      "loss": 0.1637,
+      "step": 25990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.75518892536309e-08,
+      "loss": 0.1634,
+      "step": 25991
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.739415515818473e-08,
+      "loss": 0.1858,
+      "step": 25992
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.723656265469227e-08,
+      "loss": 0.1488,
+      "step": 25993
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.707911174540617e-08,
+      "loss": 0.1661,
+      "step": 25994
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.692180243257242e-08,
+      "loss": 0.1957,
+      "step": 25995
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.676463471844143e-08,
+      "loss": 0.1806,
+      "step": 25996
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.660760860525586e-08,
+      "loss": 0.1528,
+      "step": 25997
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.645072409526056e-08,
+      "loss": 0.1648,
+      "step": 25998
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.629398119069376e-08,
+      "loss": 0.1631,
+      "step": 25999
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.6137379893797e-08,
+      "loss": 0.1966,
+      "step": 26000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.598092020680627e-08,
+      "loss": 0.1675,
+      "step": 26001
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.582460213195532e-08,
+      "loss": 0.1835,
+      "step": 26002
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.566842567147903e-08,
+      "loss": 0.1742,
+      "step": 26003
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.551239082760677e-08,
+      "loss": 0.1906,
+      "step": 26004
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.535649760256781e-08,
+      "loss": 0.1911,
+      "step": 26005
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.52007459985893e-08,
+      "loss": 0.1725,
+      "step": 26006
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.504513601789388e-08,
+      "loss": 0.1791,
+      "step": 26007
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.488966766270756e-08,
+      "loss": 0.1869,
+      "step": 26008
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.473434093524857e-08,
+      "loss": 0.177,
+      "step": 26009
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.457915583773734e-08,
+      "loss": 0.1691,
+      "step": 26010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.44241123723899e-08,
+      "loss": 0.1642,
+      "step": 26011
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.426921054142001e-08,
+      "loss": 0.1845,
+      "step": 26012
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.411445034704258e-08,
+      "loss": 0.1732,
+      "step": 26013
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.395983179146472e-08,
+      "loss": 0.1755,
+      "step": 26014
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.380535487689801e-08,
+      "loss": 0.1751,
+      "step": 26015
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.365101960554622e-08,
+      "loss": 0.1798,
+      "step": 26016
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.349682597961761e-08,
+      "loss": 0.1582,
+      "step": 26017
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.334277400131152e-08,
+      "loss": 0.1682,
+      "step": 26018
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.318886367282953e-08,
+      "loss": 0.1653,
+      "step": 26019
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.303509499637097e-08,
+      "loss": 0.1591,
+      "step": 26020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.288146797413077e-08,
+      "loss": 0.1737,
+      "step": 26021
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.272798260830383e-08,
+      "loss": 0.1736,
+      "step": 26022
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.257463890108397e-08,
+      "loss": 0.176,
+      "step": 26023
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.242143685465831e-08,
+      "loss": 0.1787,
+      "step": 26024
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.226837647121844e-08,
+      "loss": 0.1698,
+      "step": 26025
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.211545775295037e-08,
+      "loss": 0.184,
+      "step": 26026
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.196268070203572e-08,
+      "loss": 0.1794,
+      "step": 26027
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.181004532065939e-08,
+      "loss": 0.1786,
+      "step": 26028
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.165755161100075e-08,
+      "loss": 0.1561,
+      "step": 26029
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.150519957523806e-08,
+      "loss": 0.1758,
+      "step": 26030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.135298921554623e-08,
+      "loss": 0.181,
+      "step": 26031
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.120092053410134e-08,
+      "loss": 0.1669,
+      "step": 26032
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.104899353307604e-08,
+      "loss": 0.1885,
+      "step": 26033
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.089720821463754e-08,
+      "loss": 0.1732,
+      "step": 26034
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.074556458095739e-08,
+      "loss": 0.1842,
+      "step": 26035
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.059406263419833e-08,
+      "loss": 0.1657,
+      "step": 26036
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.044270237652751e-08,
+      "loss": 0.1725,
+      "step": 26037
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.029148381010432e-08,
+      "loss": 0.1704,
+      "step": 26038
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.014040693709147e-08,
+      "loss": 0.1771,
+      "step": 26039
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.998947175964389e-08,
+      "loss": 0.1866,
+      "step": 26040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.983867827991987e-08,
+      "loss": 0.1665,
+      "step": 26041
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.968802650007213e-08,
+      "loss": 0.1881,
+      "step": 26042
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.953751642225337e-08,
+      "loss": 0.1917,
+      "step": 26043
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.938714804861302e-08,
+      "loss": 0.1615,
+      "step": 26044
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.923692138129935e-08,
+      "loss": 0.1645,
+      "step": 26045
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.90868364224584e-08,
+      "loss": 0.1772,
+      "step": 26046
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.893689317423293e-08,
+      "loss": 0.176,
+      "step": 26047
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.878709163876453e-08,
+      "loss": 0.173,
+      "step": 26048
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.863743181819483e-08,
+      "loss": 0.1795,
+      "step": 26049
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.848791371465992e-08,
+      "loss": 0.1799,
+      "step": 26050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.833853733029695e-08,
+      "loss": 0.1751,
+      "step": 26051
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.818930266723756e-08,
+      "loss": 0.1749,
+      "step": 26052
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.804020972761562e-08,
+      "loss": 0.1851,
+      "step": 26053
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.78912585135605e-08,
+      "loss": 0.1801,
+      "step": 26054
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.774244902719719e-08,
+      "loss": 0.1718,
+      "step": 26055
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.759378127065397e-08,
+      "loss": 0.2028,
+      "step": 26056
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.744525524605473e-08,
+      "loss": 0.1751,
+      "step": 26057
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.729687095551885e-08,
+      "loss": 0.177,
+      "step": 26058
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.714862840116689e-08,
+      "loss": 0.1688,
+      "step": 26059
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.700052758511601e-08,
+      "loss": 0.1917,
+      "step": 26060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.685256850948341e-08,
+      "loss": 0.1933,
+      "step": 26061
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.670475117638076e-08,
+      "loss": 0.1668,
+      "step": 26062
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.655707558791969e-08,
+      "loss": 0.1676,
+      "step": 26063
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.640954174620962e-08,
+      "loss": 0.1983,
+      "step": 26064
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.626214965335887e-08,
+      "loss": 0.1675,
+      "step": 26065
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.611489931147131e-08,
+      "loss": 0.1723,
+      "step": 26066
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.596779072265192e-08,
+      "loss": 0.1691,
+      "step": 26067
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.582082388900014e-08,
+      "loss": 0.1828,
+      "step": 26068
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.567399881261761e-08,
+      "loss": 0.1593,
+      "step": 26069
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.552731549559932e-08,
+      "loss": 0.1793,
+      "step": 26070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.53807739400414e-08,
+      "loss": 0.156,
+      "step": 26071
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.523437414803769e-08,
+      "loss": 0.1919,
+      "step": 26072
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.508811612167766e-08,
+      "loss": 0.1777,
+      "step": 26073
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.494199986305184e-08,
+      "loss": 0.1796,
+      "step": 26074
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.479602537424747e-08,
+      "loss": 0.1831,
+      "step": 26075
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.465019265734841e-08,
+      "loss": 0.1779,
+      "step": 26076
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.450450171443967e-08,
+      "loss": 0.1725,
+      "step": 26077
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.435895254760072e-08,
+      "loss": 0.1902,
+      "step": 26078
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.421354515891099e-08,
+      "loss": 0.1712,
+      "step": 26079
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.406827955044771e-08,
+      "loss": 0.1806,
+      "step": 26080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.392315572428588e-08,
+      "loss": 0.1716,
+      "step": 26081
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.377817368249829e-08,
+      "loss": 0.1695,
+      "step": 26082
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.363333342715551e-08,
+      "loss": 0.1732,
+      "step": 26083
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.348863496032699e-08,
+      "loss": 0.1676,
+      "step": 26084
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.334407828407885e-08,
+      "loss": 0.1534,
+      "step": 26085
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.319966340047723e-08,
+      "loss": 0.1726,
+      "step": 26086
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.305539031158382e-08,
+      "loss": 0.168,
+      "step": 26087
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 0.1747,
+      "step": 26088
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.276726952616497e-08,
+      "loss": 0.173,
+      "step": 26089
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.262342183375403e-08,
+      "loss": 0.1912,
+      "step": 26090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.247971594428249e-08,
+      "loss": 0.1848,
+      "step": 26091
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.233615185980424e-08,
+      "loss": 0.1608,
+      "step": 26092
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.219272958236878e-08,
+      "loss": 0.1842,
+      "step": 26093
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.204944911402445e-08,
+      "loss": 0.1513,
+      "step": 26094
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.190631045681962e-08,
+      "loss": 0.1778,
+      "step": 26095
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.176331361279709e-08,
+      "loss": 0.1658,
+      "step": 26096
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.162045858399969e-08,
+      "loss": 0.174,
+      "step": 26097
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.147774537246798e-08,
+      "loss": 0.187,
+      "step": 26098
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.133517398024259e-08,
+      "loss": 0.171,
+      "step": 26099
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.119274440935631e-08,
+      "loss": 0.1813,
+      "step": 26100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.10504566618464e-08,
+      "loss": 0.199,
+      "step": 26101
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.090831073974347e-08,
+      "loss": 0.1762,
+      "step": 26102
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.076630664508033e-08,
+      "loss": 0.1667,
+      "step": 26103
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.062444437988314e-08,
+      "loss": 0.1636,
+      "step": 26104
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.048272394618028e-08,
+      "loss": 0.1581,
+      "step": 26105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.034114534599345e-08,
+      "loss": 0.163,
+      "step": 26106
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.019970858134662e-08,
+      "loss": 0.2039,
+      "step": 26107
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.005841365426037e-08,
+      "loss": 0.178,
+      "step": 26108
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.991726056675196e-08,
+      "loss": 0.1887,
+      "step": 26109
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.977624932083871e-08,
+      "loss": 0.1793,
+      "step": 26110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.963537991853342e-08,
+      "loss": 0.1881,
+      "step": 26111
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.949465236185005e-08,
+      "loss": 0.1644,
+      "step": 26112
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.935406665279698e-08,
+      "loss": 0.1809,
+      "step": 26113
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.921362279338261e-08,
+      "loss": 0.1766,
+      "step": 26114
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.907332078561535e-08,
+      "loss": 0.1795,
+      "step": 26115
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.89331606314958e-08,
+      "loss": 0.1772,
+      "step": 26116
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.879314233302903e-08,
+      "loss": 0.1602,
+      "step": 26117
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.865326589221344e-08,
+      "loss": 0.1757,
+      "step": 26118
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.851353131104632e-08,
+      "loss": 0.1702,
+      "step": 26119
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.837393859152608e-08,
+      "loss": 0.177,
+      "step": 26120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.823448773564446e-08,
+      "loss": 0.1581,
+      "step": 26121
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.809517874539429e-08,
+      "loss": 0.1606,
+      "step": 26122
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.795601162276511e-08,
+      "loss": 0.1608,
+      "step": 26123
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.781698636974532e-08,
+      "loss": 0.1568,
+      "step": 26124
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.767810298832112e-08,
+      "loss": 0.1586,
+      "step": 26125
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.753936148047424e-08,
+      "loss": 0.1893,
+      "step": 26126
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.740076184818867e-08,
+      "loss": 0.174,
+      "step": 26127
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.726230409344392e-08,
+      "loss": 0.1605,
+      "step": 26128
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.71239882182162e-08,
+      "loss": 0.1848,
+      "step": 26129
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.69858142244828e-08,
+      "loss": 0.1517,
+      "step": 26130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.68477821142166e-08,
+      "loss": 0.1812,
+      "step": 26131
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.670989188939048e-08,
+      "loss": 0.1655,
+      "step": 26132
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.657214355197172e-08,
+      "loss": 0.1758,
+      "step": 26133
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.643453710392988e-08,
+      "loss": 0.1969,
+      "step": 26134
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.629707254723005e-08,
+      "loss": 0.1782,
+      "step": 26135
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.615974988383511e-08,
+      "loss": 0.1861,
+      "step": 26136
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.602256911570792e-08,
+      "loss": 0.1822,
+      "step": 26137
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.588553024480692e-08,
+      "loss": 0.1838,
+      "step": 26138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.574863327308834e-08,
+      "loss": 0.1858,
+      "step": 26139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.561187820251058e-08,
+      "loss": 0.1684,
+      "step": 26140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.547526503502544e-08,
+      "loss": 0.1674,
+      "step": 26141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.533879377258467e-08,
+      "loss": 0.1811,
+      "step": 26142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.520246441713785e-08,
+      "loss": 0.1694,
+      "step": 26143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.506627697063007e-08,
+      "loss": 0.1672,
+      "step": 26144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.493023143501087e-08,
+      "loss": 0.188,
+      "step": 26145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.479432781221873e-08,
+      "loss": 0.1745,
+      "step": 26146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.465856610419873e-08,
+      "loss": 0.1652,
+      "step": 26147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.452294631288824e-08,
+      "loss": 0.2058,
+      "step": 26148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.438746844022459e-08,
+      "loss": 0.1654,
+      "step": 26149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.425213248814399e-08,
+      "loss": 0.1723,
+      "step": 26150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.411693845857714e-08,
+      "loss": 0.1822,
+      "step": 26151
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.398188635345914e-08,
+      "loss": 0.1715,
+      "step": 26152
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.384697617471514e-08,
+      "loss": 0.1874,
+      "step": 26153
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.371220792427357e-08,
+      "loss": 0.1738,
+      "step": 26154
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.357758160406069e-08,
+      "loss": 0.1938,
+      "step": 26155
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.344309721599829e-08,
+      "loss": 0.1726,
+      "step": 26156
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.330875476200703e-08,
+      "loss": 0.1856,
+      "step": 26157
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.317455424400764e-08,
+      "loss": 0.1699,
+      "step": 26158
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.304049566391523e-08,
+      "loss": 0.1957,
+      "step": 26159
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.290657902364494e-08,
+      "loss": 0.1798,
+      "step": 26160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.277280432511079e-08,
+      "loss": 0.1583,
+      "step": 26161
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.263917157022236e-08,
+      "loss": 0.1417,
+      "step": 26162
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.250568076088814e-08,
+      "loss": 0.1745,
+      "step": 26163
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.23723318990166e-08,
+      "loss": 0.1693,
+      "step": 26164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.223912498651064e-08,
+      "loss": 0.1692,
+      "step": 26165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.210606002527542e-08,
+      "loss": 0.1835,
+      "step": 26166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.197313701720942e-08,
+      "loss": 0.1796,
+      "step": 26167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.184035596421223e-08,
+      "loss": 0.1622,
+      "step": 26168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.170771686818012e-08,
+      "loss": 0.1779,
+      "step": 26169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.157521973100822e-08,
+      "loss": 0.1517,
+      "step": 26170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.144286455458836e-08,
+      "loss": 0.1699,
+      "step": 26171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.131065134081238e-08,
+      "loss": 0.1599,
+      "step": 26172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.117858009156763e-08,
+      "loss": 0.1606,
+      "step": 26173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.104665080874039e-08,
+      "loss": 0.196,
+      "step": 26174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.091486349421583e-08,
+      "loss": 0.1614,
+      "step": 26175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.078321814987686e-08,
+      "loss": 0.1651,
+      "step": 26176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.065171477760423e-08,
+      "loss": 0.1866,
+      "step": 26177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.05203533792742e-08,
+      "loss": 0.1757,
+      "step": 26178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.038913395676415e-08,
+      "loss": 0.1649,
+      "step": 26179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.025805651194927e-08,
+      "loss": 0.1691,
+      "step": 26180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.012712104670138e-08,
+      "loss": 0.1644,
+      "step": 26181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.99963275628912e-08,
+      "loss": 0.1846,
+      "step": 26182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.986567606238614e-08,
+      "loss": 0.1628,
+      "step": 26183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.973516654705248e-08,
+      "loss": 0.1744,
+      "step": 26184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.960479901875538e-08,
+      "loss": 0.1766,
+      "step": 26185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.947457347935559e-08,
+      "loss": 0.1822,
+      "step": 26186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9344489930713846e-08,
+      "loss": 0.1857,
+      "step": 26187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.921454837468865e-08,
+      "loss": 0.1603,
+      "step": 26188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9084748813136305e-08,
+      "loss": 0.1792,
+      "step": 26189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.895509124790977e-08,
+      "loss": 0.1697,
+      "step": 26190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8825575680862004e-08,
+      "loss": 0.1718,
+      "step": 26191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8696202113842634e-08,
+      "loss": 0.1922,
+      "step": 26192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8566970548700195e-08,
+      "loss": 0.1848,
+      "step": 26193
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.843788098727987e-08,
+      "loss": 0.2095,
+      "step": 26194
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.830893343142574e-08,
+      "loss": 0.1543,
+      "step": 26195
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8180127882980774e-08,
+      "loss": 0.1668,
+      "step": 26196
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8051464343783506e-08,
+      "loss": 0.1709,
+      "step": 26197
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7922942815672456e-08,
+      "loss": 0.1561,
+      "step": 26198
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.779456330048172e-08,
+      "loss": 0.1465,
+      "step": 26199
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.766632580004761e-08,
+      "loss": 0.1761,
+      "step": 26200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.753823031620087e-08,
+      "loss": 0.1843,
+      "step": 26201
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7410276850770055e-08,
+      "loss": 0.1963,
+      "step": 26202
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.728246540558591e-08,
+      "loss": 0.1466,
+      "step": 26203
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7154795982470315e-08,
+      "loss": 0.1626,
+      "step": 26204
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.702726858324959e-08,
+      "loss": 0.167,
+      "step": 26205
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.689988320974449e-08,
+      "loss": 0.1838,
+      "step": 26206
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.677263986377357e-08,
+      "loss": 0.1852,
+      "step": 26207
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6645538547156485e-08,
+      "loss": 0.1859,
+      "step": 26208
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651857926170734e-08,
+      "loss": 0.1888,
+      "step": 26209
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.639176200923913e-08,
+      "loss": 0.1821,
+      "step": 26210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.626508679156484e-08,
+      "loss": 0.1845,
+      "step": 26211
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.613855361049303e-08,
+      "loss": 0.1916,
+      "step": 26212
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6012162467831144e-08,
+      "loss": 0.1712,
+      "step": 26213
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.58859133653844e-08,
+      "loss": 0.1682,
+      "step": 26214
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.575980630495692e-08,
+      "loss": 0.2001,
+      "step": 26215
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5633841288349477e-08,
+      "loss": 0.1741,
+      "step": 26216
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5508018317360636e-08,
+      "loss": 0.1667,
+      "step": 26217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.538233739379006e-08,
+      "loss": 0.1725,
+      "step": 26218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5256798519429665e-08,
+      "loss": 0.1921,
+      "step": 26219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5131401696074674e-08,
+      "loss": 0.143,
+      "step": 26220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.500614692551587e-08,
+      "loss": 0.1796,
+      "step": 26221
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.488103420954294e-08,
+      "loss": 0.1638,
+      "step": 26222
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.475606354994223e-08,
+      "loss": 0.17,
+      "step": 26223
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.463123494849898e-08,
+      "loss": 0.1745,
+      "step": 26224
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.450654840699621e-08,
+      "loss": 0.1746,
+      "step": 26225
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.438200392721471e-08,
+      "loss": 0.1926,
+      "step": 26226
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.425760151093529e-08,
+      "loss": 0.1584,
+      "step": 26227
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4133341159932075e-08,
+      "loss": 0.1764,
+      "step": 26228
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4009222875981425e-08,
+      "loss": 0.1616,
+      "step": 26229
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.388524666085637e-08,
+      "loss": 0.1661,
+      "step": 26230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.376141251632771e-08,
+      "loss": 0.1818,
+      "step": 26231
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3637720444165155e-08,
+      "loss": 0.1607,
+      "step": 26232
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.351417044613394e-08,
+      "loss": 0.1808,
+      "step": 26233
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.339076252399822e-08,
+      "loss": 0.1661,
+      "step": 26234
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3267496679523244e-08,
+      "loss": 0.1873,
+      "step": 26235
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.314437291446872e-08,
+      "loss": 0.1809,
+      "step": 26236
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.302139123059324e-08,
+      "loss": 0.1504,
+      "step": 26237
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2898551629653186e-08,
+      "loss": 0.1809,
+      "step": 26238
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2775854113403804e-08,
+      "loss": 0.1665,
+      "step": 26239
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.265329868359703e-08,
+      "loss": 0.1941,
+      "step": 26240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2530885341982586e-08,
+      "loss": 0.1858,
+      "step": 26241
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.240861409031128e-08,
+      "loss": 0.176,
+      "step": 26242
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2286484930329503e-08,
+      "loss": 0.16,
+      "step": 26243
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.21644978637792e-08,
+      "loss": 0.148,
+      "step": 26244
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2042652892405623e-08,
+      "loss": 0.159,
+      "step": 26245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.192095001794628e-08,
+      "loss": 0.1815,
+      "step": 26246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1799389242143116e-08,
+      "loss": 0.1624,
+      "step": 26247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1677970566730295e-08,
+      "loss": 0.1892,
+      "step": 26248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1556693993441984e-08,
+      "loss": 0.1725,
+      "step": 26249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.143555952401236e-08,
+      "loss": 0.1792,
+      "step": 26250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.131456716016892e-08,
+      "loss": 0.1872,
+      "step": 26251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.11937169036425e-08,
+      "loss": 0.1711,
+      "step": 26252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.107300875615728e-08,
+      "loss": 0.1605,
+      "step": 26253
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.095244271943966e-08,
+      "loss": 0.1673,
+      "step": 26254
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.083201879521049e-08,
+      "loss": 0.1926,
+      "step": 26255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.07117369851895e-08,
+      "loss": 0.1424,
+      "step": 26256
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.059159729109531e-08,
+      "loss": 0.165,
+      "step": 26257
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.047159971464322e-08,
+      "loss": 0.1579,
+      "step": 26258
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.035174425754963e-08,
+      "loss": 0.1862,
+      "step": 26259
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0232030921524286e-08,
+      "loss": 0.1657,
+      "step": 26260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.011245970827805e-08,
+      "loss": 0.1617,
+      "step": 26261
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9993030619518435e-08,
+      "loss": 0.1649,
+      "step": 26262
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9873743656952965e-08,
+      "loss": 0.1625,
+      "step": 26263
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.975459882228251e-08,
+      "loss": 0.171,
+      "step": 26264
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.963559611721125e-08,
+      "loss": 0.1697,
+      "step": 26265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9516735543438946e-08,
+      "loss": 0.1782,
+      "step": 26266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.939801710266312e-08,
+      "loss": 0.1829,
+      "step": 26267
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.927944079657798e-08,
+      "loss": 0.1827,
+      "step": 26268
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.916100662687995e-08,
+      "loss": 0.1638,
+      "step": 26269
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9042714595258776e-08,
+      "loss": 0.1652,
+      "step": 26270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8924564703404234e-08,
+      "loss": 0.1537,
+      "step": 26271
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8806556953006065e-08,
+      "loss": 0.1868,
+      "step": 26272
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8688691345746273e-08,
+      "loss": 0.1993,
+      "step": 26273
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.857096788331239e-08,
+      "loss": 0.1926,
+      "step": 26274
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.845338656738419e-08,
+      "loss": 0.1785,
+      "step": 26275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.833594739964031e-08,
+      "loss": 0.1739,
+      "step": 26276
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8218650381759434e-08,
+      "loss": 0.1849,
+      "step": 26277
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.810149551541576e-08,
+      "loss": 0.1649,
+      "step": 26278
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.798448280228463e-08,
+      "loss": 0.1816,
+      "step": 26279
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7867612244036906e-08,
+      "loss": 0.1688,
+      "step": 26280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.775088384234017e-08,
+      "loss": 0.176,
+      "step": 26281
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.763429759886418e-08,
+      "loss": 0.1788,
+      "step": 26282
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.751785351527316e-08,
+      "loss": 0.1527,
+      "step": 26283
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7401551593230234e-08,
+      "loss": 0.174,
+      "step": 26284
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.72853918343974e-08,
+      "loss": 0.1774,
+      "step": 26285
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.716937424043333e-08,
+      "loss": 0.1716,
+      "step": 26286
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.705349881299448e-08,
+      "loss": 0.1718,
+      "step": 26287
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.69377655537373e-08,
+      "loss": 0.1793,
+      "step": 26288
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.682217446431492e-08,
+      "loss": 0.1757,
+      "step": 26289
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.670672554637823e-08,
+      "loss": 0.1571,
+      "step": 26290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6591418801574804e-08,
+      "loss": 0.1849,
+      "step": 26291
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.647625423155444e-08,
+      "loss": 0.1853,
+      "step": 26292
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.636123183795915e-08,
+      "loss": 0.1745,
+      "step": 26293
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.624635162243429e-08,
+      "loss": 0.1703,
+      "step": 26294
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.613161358661966e-08,
+      "loss": 0.1871,
+      "step": 26295
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.601701773215505e-08,
+      "loss": 0.1892,
+      "step": 26296
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.590256406067583e-08,
+      "loss": 0.1532,
+      "step": 26297
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.578825257381847e-08,
+      "loss": 0.2045,
+      "step": 26298
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.567408327321499e-08,
+      "loss": 0.1877,
+      "step": 26299
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.55600561604963e-08,
+      "loss": 0.1664,
+      "step": 26300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.544617123729111e-08,
+      "loss": 0.1866,
+      "step": 26301
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.53324285052259e-08,
+      "loss": 0.1895,
+      "step": 26302
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5218827965927134e-08,
+      "loss": 0.161,
+      "step": 26303
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.510536962101464e-08,
+      "loss": 0.1695,
+      "step": 26304
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.499205347211155e-08,
+      "loss": 0.1818,
+      "step": 26305
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4878879520834364e-08,
+      "loss": 0.178,
+      "step": 26306
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4765847768800666e-08,
+      "loss": 0.1721,
+      "step": 26307
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.465295821762583e-08,
+      "loss": 0.1771,
+      "step": 26308
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.454021086892191e-08,
+      "loss": 0.1746,
+      "step": 26309
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4427605724297604e-08,
+      "loss": 0.1772,
+      "step": 26310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.431514278536386e-08,
+      "loss": 0.1749,
+      "step": 26311
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.420282205372606e-08,
+      "loss": 0.1648,
+      "step": 26312
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.409064353098847e-08,
+      "loss": 0.1617,
+      "step": 26313
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.397860721875313e-08,
+      "loss": 0.1504,
+      "step": 26314
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3866713118621005e-08,
+      "loss": 0.1611,
+      "step": 26315
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3754961232191915e-08,
+      "loss": 0.1859,
+      "step": 26316
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.364335156105792e-08,
+      "loss": 0.1913,
+      "step": 26317
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.353188410681775e-08,
+      "loss": 0.1686,
+      "step": 26318
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3420558871060116e-08,
+      "loss": 0.175,
+      "step": 26319
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.33093758553782e-08,
+      "loss": 0.18,
+      "step": 26320
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.31983350613574e-08,
+      "loss": 0.1836,
+      "step": 26321
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.308743649058533e-08,
+      "loss": 0.1647,
+      "step": 26322
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2976680144646287e-08,
+      "loss": 0.1602,
+      "step": 26323
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.286606602512122e-08,
+      "loss": 0.1656,
+      "step": 26324
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.27555941335911e-08,
+      "loss": 0.1872,
+      "step": 26325
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.264526447163353e-08,
+      "loss": 0.1681,
+      "step": 26326
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.253507704082505e-08,
+      "loss": 0.1721,
+      "step": 26327
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.242503184273772e-08,
+      "loss": 0.1797,
+      "step": 26328
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2315128878945846e-08,
+      "loss": 0.1749,
+      "step": 26329
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2205368151018164e-08,
+      "loss": 0.1844,
+      "step": 26330
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2095749660522325e-08,
+      "loss": 0.1741,
+      "step": 26331
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.198627340902484e-08,
+      "loss": 0.1674,
+      "step": 26332
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.187693939808779e-08,
+      "loss": 0.1828,
+      "step": 26333
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.176774762927438e-08,
+      "loss": 0.1842,
+      "step": 26334
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.165869810414558e-08,
+      "loss": 0.166,
+      "step": 26335
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.15497908242557e-08,
+      "loss": 0.1838,
+      "step": 26336
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.144102579116238e-08,
+      "loss": 0.1845,
+      "step": 26337
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.133240300641994e-08,
+      "loss": 0.1793,
+      "step": 26338
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.122392247157825e-08,
+      "loss": 0.1874,
+      "step": 26339
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.111558418818717e-08,
+      "loss": 0.1689,
+      "step": 26340
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.100738815779548e-08,
+      "loss": 0.1793,
+      "step": 26341
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0899334381947486e-08,
+      "loss": 0.2022,
+      "step": 26342
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.079142286218751e-08,
+      "loss": 0.1655,
+      "step": 26343
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.068365360005544e-08,
+      "loss": 0.1592,
+      "step": 26344
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.057602659709225e-08,
+      "loss": 0.1881,
+      "step": 26345
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.046854185483451e-08,
+      "loss": 0.1703,
+      "step": 26346
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.036119937481764e-08,
+      "loss": 0.1741,
+      "step": 26347
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.025399915857598e-08,
+      "loss": 0.2033,
+      "step": 26348
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.014694120763829e-08,
+      "loss": 0.166,
+      "step": 26349
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0040025523535584e-08,
+      "loss": 0.1617,
+      "step": 26350
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.993325210779553e-08,
+      "loss": 0.1692,
+      "step": 26351
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9826620961941345e-08,
+      "loss": 0.184,
+      "step": 26352
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9720132087498474e-08,
+      "loss": 0.1867,
+      "step": 26353
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.961378548598571e-08,
+      "loss": 0.1771,
+      "step": 26354
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9507581158925166e-08,
+      "loss": 0.1859,
+      "step": 26355
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.940151910783008e-08,
+      "loss": 0.2039,
+      "step": 26356
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9295599334218116e-08,
+      "loss": 0.1835,
+      "step": 26357
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9189821839600294e-08,
+      "loss": 0.1779,
+      "step": 26358
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.908418662548985e-08,
+      "loss": 0.1796,
+      "step": 26359
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8978693693394464e-08,
+      "loss": 0.1672,
+      "step": 26360
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.887334304482071e-08,
+      "loss": 0.1579,
+      "step": 26361
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.876813468127405e-08,
+      "loss": 0.185,
+      "step": 26362
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8663068604255506e-08,
+      "loss": 0.174,
+      "step": 26363
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.855814481526943e-08,
+      "loss": 0.1786,
+      "step": 26364
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8453363315811285e-08,
+      "loss": 0.197,
+      "step": 26365
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8348724107378777e-08,
+      "loss": 0.1558,
+      "step": 26366
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.824422719146736e-08,
+      "loss": 0.176,
+      "step": 26367
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8139872569569195e-08,
+      "loss": 0.1785,
+      "step": 26368
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8035660243174174e-08,
+      "loss": 0.178,
+      "step": 26369
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.793159021377224e-08,
+      "loss": 0.1745,
+      "step": 26370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.782766248284775e-08,
+      "loss": 0.1738,
+      "step": 26371
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.77238770518884e-08,
+      "loss": 0.1738,
+      "step": 26372
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7620233922373014e-08,
+      "loss": 0.1647,
+      "step": 26373
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.751673309578485e-08,
+      "loss": 0.1827,
+      "step": 26374
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.74133745736005e-08,
+      "loss": 0.1788,
+      "step": 26375
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7310158357298785e-08,
+      "loss": 0.1602,
+      "step": 26376
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.720708444835075e-08,
+      "loss": 0.1712,
+      "step": 26377
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.710415284823188e-08,
+      "loss": 0.1673,
+      "step": 26378
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7001363558409887e-08,
+      "loss": 0.1762,
+      "step": 26379
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.689871658035471e-08,
+      "loss": 0.1896,
+      "step": 26380
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6796211915530733e-08,
+      "loss": 0.1593,
+      "step": 26381
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.669384956540567e-08,
+      "loss": 0.1886,
+      "step": 26382
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6591629531437244e-08,
+      "loss": 0.1801,
+      "step": 26383
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.648955181508873e-08,
+      "loss": 0.1764,
+      "step": 26384
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.638761641781674e-08,
+      "loss": 0.1667,
+      "step": 26385
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6285823341077885e-08,
+      "loss": 0.1888,
+      "step": 26386
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.618417258632545e-08,
+      "loss": 0.1793,
+      "step": 26387
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.608266415501271e-08,
+      "loss": 0.1788,
+      "step": 26388
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.598129804858852e-08,
+      "loss": 0.1786,
+      "step": 26389
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5880074268501707e-08,
+      "loss": 0.1869,
+      "step": 26390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5778992816196676e-08,
+      "loss": 0.1817,
+      "step": 26391
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.567805369311894e-08,
+      "loss": 0.1682,
+      "step": 26392
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.557725690070957e-08,
+      "loss": 0.1772,
+      "step": 26393
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5476602440408514e-08,
+      "loss": 0.1728,
+      "step": 26394
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.537609031365352e-08,
+      "loss": 0.1928,
+      "step": 26395
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.527572052188011e-08,
+      "loss": 0.1668,
+      "step": 26396
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.517549306652157e-08,
+      "loss": 0.1862,
+      "step": 26397
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.50754079490101e-08,
+      "loss": 0.1669,
+      "step": 26398
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.497546517077566e-08,
+      "loss": 0.1701,
+      "step": 26399
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.487566473324488e-08,
+      "loss": 0.1726,
+      "step": 26400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4776006637843307e-08,
+      "loss": 0.189,
+      "step": 26401
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.467649088599645e-08,
+      "loss": 0.1821,
+      "step": 26402
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.457711747912318e-08,
+      "loss": 0.1711,
+      "step": 26403
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.447788641864458e-08,
+      "loss": 0.2049,
+      "step": 26404
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4378797705977296e-08,
+      "loss": 0.1743,
+      "step": 26405
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4279851342537975e-08,
+      "loss": 0.179,
+      "step": 26406
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.41810473297377e-08,
+      "loss": 0.18,
+      "step": 26407
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.408238566899091e-08,
+      "loss": 0.1865,
+      "step": 26408
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3983866361703145e-08,
+      "loss": 0.178,
+      "step": 26409
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.388548940928438e-08,
+      "loss": 0.1716,
+      "step": 26410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.378725481314016e-08,
+      "loss": 0.1827,
+      "step": 26411
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.368916257467047e-08,
+      "loss": 0.1772,
+      "step": 26412
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.359121269527976e-08,
+      "loss": 0.1562,
+      "step": 26413
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3493405176365787e-08,
+      "loss": 0.1683,
+      "step": 26414
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.339574001932633e-08,
+      "loss": 0.1686,
+      "step": 26415
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.329821722555471e-08,
+      "loss": 0.1853,
+      "step": 26416
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.320083679644648e-08,
+      "loss": 0.1791,
+      "step": 26417
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.310359873339053e-08,
+      "loss": 0.1625,
+      "step": 26418
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.300650303777686e-08,
+      "loss": 0.1724,
+      "step": 26419
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.290954971099103e-08,
+      "loss": 0.1694,
+      "step": 26420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.281273875442081e-08,
+      "loss": 0.151,
+      "step": 26421
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2716070169446224e-08,
+      "loss": 0.1719,
+      "step": 26422
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.261954395744948e-08,
+      "loss": 0.1646,
+      "step": 26423
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.252316011980838e-08,
+      "loss": 0.1748,
+      "step": 26424
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.242691865790071e-08,
+      "loss": 0.181,
+      "step": 26425
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2330819573099804e-08,
+      "loss": 0.1692,
+      "step": 26426
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2234862866780124e-08,
+      "loss": 0.2017,
+      "step": 26427
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.213904854031169e-08,
+      "loss": 0.1835,
+      "step": 26428
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.204337659506229e-08,
+      "loss": 0.1752,
+      "step": 26429
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.194784703239973e-08,
+      "loss": 0.157,
+      "step": 26430
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.185245985368845e-08,
+      "loss": 0.166,
+      "step": 26431
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1757215060288507e-08,
+      "loss": 0.1735,
+      "step": 26432
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1662112653564334e-08,
+      "loss": 0.1582,
+      "step": 26433
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.156715263487153e-08,
+      "loss": 0.1867,
+      "step": 26434
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.14723350055679e-08,
+      "loss": 0.1745,
+      "step": 26435
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1377659767006795e-08,
+      "loss": 0.1819,
+      "step": 26436
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1283126920540475e-08,
+      "loss": 0.1575,
+      "step": 26437
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.118873646752118e-08,
+      "loss": 0.1563,
+      "step": 26438
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.109448840929452e-08,
+      "loss": 0.1659,
+      "step": 26439
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.100038274720829e-08,
+      "loss": 0.1608,
+      "step": 26440
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.090641948260698e-08,
+      "loss": 0.1576,
+      "step": 26441
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.081259861683172e-08,
+      "loss": 0.1658,
+      "step": 26442
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0718920151223684e-08,
+      "loss": 0.1767,
+      "step": 26443
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.062538408711957e-08,
+      "loss": 0.1735,
+      "step": 26444
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.053199042585719e-08,
+      "loss": 0.1905,
+      "step": 26445
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0438739168769937e-08,
+      "loss": 0.1802,
+      "step": 26446
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.034563031718896e-08,
+      "loss": 0.1779,
+      "step": 26447
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0252663872445406e-08,
+      "loss": 0.1612,
+      "step": 26448
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0159839835866014e-08,
+      "loss": 0.1911,
+      "step": 26449
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.006715820877859e-08,
+      "loss": 0.1745,
+      "step": 26450
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.997461899250431e-08,
+      "loss": 0.1845,
+      "step": 26451
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.988222218836656e-08,
+      "loss": 0.1684,
+      "step": 26452
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9789967797686502e-08,
+      "loss": 0.1728,
+      "step": 26453
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9697855821778642e-08,
+      "loss": 0.1687,
+      "step": 26454
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9605886261961923e-08,
+      "loss": 0.1707,
+      "step": 26455
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.951405911954752e-08,
+      "loss": 0.1815,
+      "step": 26456
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.942237439584883e-08,
+      "loss": 0.1788,
+      "step": 26457
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.93308320921748e-08,
+      "loss": 0.1755,
+      "step": 26458
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9239432209833275e-08,
+      "loss": 0.1682,
+      "step": 26459
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9148174750128765e-08,
+      "loss": 0.1814,
+      "step": 26460
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9057059714366897e-08,
+      "loss": 0.1931,
+      "step": 26461
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.896608710384663e-08,
+      "loss": 0.1786,
+      "step": 26462
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8875256919869143e-08,
+      "loss": 0.1783,
+      "step": 26463
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.878456916373118e-08,
+      "loss": 0.1651,
+      "step": 26464
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.869402383672948e-08,
+      "loss": 0.1833,
+      "step": 26465
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.860362094015523e-08,
+      "loss": 0.1863,
+      "step": 26466
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8513360475300734e-08,
+      "loss": 0.1619,
+      "step": 26467
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8423242443456066e-08,
+      "loss": 0.1779,
+      "step": 26468
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.833326684590687e-08,
+      "loss": 0.1816,
+      "step": 26469
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8243433683939892e-08,
+      "loss": 0.1836,
+      "step": 26470
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8153742958837437e-08,
+      "loss": 0.1794,
+      "step": 26471
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8064194671880706e-08,
+      "loss": 0.175,
+      "step": 26472
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7974788824348676e-08,
+      "loss": 0.1648,
+      "step": 26473
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.788552541751921e-08,
+      "loss": 0.1727,
+      "step": 26474
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7796404452666847e-08,
+      "loss": 0.1732,
+      "step": 26475
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7707425931065012e-08,
+      "loss": 0.1754,
+      "step": 26476
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.76185898539838e-08,
+      "loss": 0.1654,
+      "step": 26477
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7529896222692197e-08,
+      "loss": 0.1975,
+      "step": 26478
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7441345038458077e-08,
+      "loss": 0.1823,
+      "step": 26479
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7352936302544875e-08,
+      "loss": 0.1561,
+      "step": 26480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.726467001621713e-08,
+      "loss": 0.1608,
+      "step": 26481
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7176546180734954e-08,
+      "loss": 0.1933,
+      "step": 26482
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.708856479735622e-08,
+      "loss": 0.17,
+      "step": 26483
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7000725867338818e-08,
+      "loss": 0.1978,
+      "step": 26484
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6913029391937295e-08,
+      "loss": 0.1917,
+      "step": 26485
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6825475372403987e-08,
+      "loss": 0.1726,
+      "step": 26486
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6738063809989e-08,
+      "loss": 0.161,
+      "step": 26487
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.665079470594245e-08,
+      "loss": 0.161,
+      "step": 26488
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6563668061508894e-08,
+      "loss": 0.1756,
+      "step": 26489
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6476683877934007e-08,
+      "loss": 0.1692,
+      "step": 26490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6389842156461232e-08,
+      "loss": 0.1667,
+      "step": 26491
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.630314289832847e-08,
+      "loss": 0.1621,
+      "step": 26492
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6216586104776953e-08,
+      "loss": 0.1611,
+      "step": 26493
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6130171777040138e-08,
+      "loss": 0.1583,
+      "step": 26494
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6043899916354808e-08,
+      "loss": 0.1734,
+      "step": 26495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.595777052395332e-08,
+      "loss": 0.1763,
+      "step": 26496
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.587178360106357e-08,
+      "loss": 0.179,
+      "step": 26497
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5785939148915695e-08,
+      "loss": 0.1754,
+      "step": 26498
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5700237168736485e-08,
+      "loss": 0.1735,
+      "step": 26499
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.56146776617483e-08,
+      "loss": 0.1692,
+      "step": 26500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.55292606291746e-08,
+      "loss": 0.1728,
+      "step": 26501
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5443986072233307e-08,
+      "loss": 0.1912,
+      "step": 26502
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.535885399214566e-08,
+      "loss": 0.1803,
+      "step": 26503
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5273864390126244e-08,
+      "loss": 0.1706,
+      "step": 26504
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.518901726738854e-08,
+      "loss": 0.167,
+      "step": 26505
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.51043126251449e-08,
+      "loss": 0.1784,
+      "step": 26506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5019750464604366e-08,
+      "loss": 0.1657,
+      "step": 26507
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4935330786977074e-08,
+      "loss": 0.1733,
+      "step": 26508
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.485105359346762e-08,
+      "loss": 0.1673,
+      "step": 26509
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.476691888527838e-08,
+      "loss": 0.1723,
+      "step": 26510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4682926663613938e-08,
+      "loss": 0.1751,
+      "step": 26511
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4599076929672226e-08,
+      "loss": 0.1844,
+      "step": 26512
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.451536968465118e-08,
+      "loss": 0.1688,
+      "step": 26513
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4431804929746506e-08,
+      "loss": 0.1759,
+      "step": 26514
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4348382666152804e-08,
+      "loss": 0.1725,
+      "step": 26515
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4265102895060233e-08,
+      "loss": 0.1594,
+      "step": 26516
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4181965617658953e-08,
+      "loss": 0.1729,
+      "step": 26517
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4098970835135794e-08,
+      "loss": 0.1918,
+      "step": 26518
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4016118548677582e-08,
+      "loss": 0.1721,
+      "step": 26519
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3933408759467814e-08,
+      "loss": 0.1828,
+      "step": 26520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3850841468686658e-08,
+      "loss": 0.1915,
+      "step": 26521
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3768416677514283e-08,
+      "loss": 0.1738,
+      "step": 26522
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3686134387128634e-08,
+      "loss": 0.1868,
+      "step": 26523
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3603994598703218e-08,
+      "loss": 0.1618,
+      "step": 26524
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.352199731341265e-08,
+      "loss": 0.1846,
+      "step": 26525
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.344014253242821e-08,
+      "loss": 0.1696,
+      "step": 26526
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3358430256918974e-08,
+      "loss": 0.1604,
+      "step": 26527
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3276860488051777e-08,
+      "loss": 0.1764,
+      "step": 26528
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.319543322699236e-08,
+      "loss": 0.168,
+      "step": 26529
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.311414847490312e-08,
+      "loss": 0.1603,
+      "step": 26530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3033006232945354e-08,
+      "loss": 0.1719,
+      "step": 26531
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2952006502278135e-08,
+      "loss": 0.1764,
+      "step": 26532
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2871149284059425e-08,
+      "loss": 0.1903,
+      "step": 26533
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2790434579442742e-08,
+      "loss": 0.1807,
+      "step": 26534
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2709862389581616e-08,
+      "loss": 0.1675,
+      "step": 26535
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.262943271562623e-08,
+      "loss": 0.183,
+      "step": 26536
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.254914555872678e-08,
+      "loss": 0.1783,
+      "step": 26537
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2469000920029017e-08,
+      "loss": 0.1883,
+      "step": 26538
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2388998800678684e-08,
+      "loss": 0.1773,
+      "step": 26539
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.230913920181821e-08,
+      "loss": 0.1625,
+      "step": 26540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2229422124587786e-08,
+      "loss": 0.1763,
+      "step": 26541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2149847570126502e-08,
+      "loss": 0.1673,
+      "step": 26542
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2070415539570124e-08,
+      "loss": 0.1667,
+      "step": 26543
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.19911260340544e-08,
+      "loss": 0.189,
+      "step": 26544
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1911979054712873e-08,
+      "loss": 0.1901,
+      "step": 26545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.183297460267353e-08,
+      "loss": 0.1765,
+      "step": 26546
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1754112679065464e-08,
+      "loss": 0.1566,
+      "step": 26547
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1675393285016667e-08,
+      "loss": 0.1636,
+      "step": 26548
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.159681642165068e-08,
+      "loss": 0.203,
+      "step": 26549
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.151838209008994e-08,
+      "loss": 0.1716,
+      "step": 26550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.144009029145466e-08,
+      "loss": 0.1632,
+      "step": 26551
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1361941026863952e-08,
+      "loss": 0.1665,
+      "step": 26552
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1283934297432472e-08,
+      "loss": 0.1831,
+      "step": 26553
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1206070104276e-08,
+      "loss": 0.1864,
+      "step": 26554
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1128348448505865e-08,
+      "loss": 0.1521,
+      "step": 26555
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.105076933123229e-08,
+      "loss": 0.1841,
+      "step": 26556
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0973332753564392e-08,
+      "loss": 0.1601,
+      "step": 26557
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0896038716606837e-08,
+      "loss": 0.1825,
+      "step": 26558
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.08188872214643e-08,
+      "loss": 0.1529,
+      "step": 26559
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0741878269239235e-08,
+      "loss": 0.1677,
+      "step": 26560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.066501186103187e-08,
+      "loss": 0.1727,
+      "step": 26561
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0588287997939104e-08,
+      "loss": 0.175,
+      "step": 26562
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.051170668105784e-08,
+      "loss": 0.1763,
+      "step": 26563
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0435267911480537e-08,
+      "loss": 0.1861,
+      "step": 26564
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0358971690300768e-08,
+      "loss": 0.17,
+      "step": 26565
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.028281801860765e-08,
+      "loss": 0.165,
+      "step": 26566
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.020680689748922e-08,
+      "loss": 0.1612,
+      "step": 26567
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.013093832803015e-08,
+      "loss": 0.1975,
+      "step": 26568
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0055212311315132e-08,
+      "loss": 0.1709,
+      "step": 26569
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9979628848426636e-08,
+      "loss": 0.1829,
+      "step": 26570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9904187940441578e-08,
+      "loss": 0.1804,
+      "step": 26571
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9828889588440204e-08,
+      "loss": 0.1853,
+      "step": 26572
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.975373379349721e-08,
+      "loss": 0.1696,
+      "step": 26573
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.967872055668618e-08,
+      "loss": 0.1608,
+      "step": 26574
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9603849879078486e-08,
+      "loss": 0.1855,
+      "step": 26575
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9529121761743265e-08,
+      "loss": 0.154,
+      "step": 26576
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9454536205748553e-08,
+      "loss": 0.1661,
+      "step": 26577
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9380093212159057e-08,
+      "loss": 0.1718,
+      "step": 26578
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9305792782039478e-08,
+      "loss": 0.1756,
+      "step": 26579
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.923163491644897e-08,
+      "loss": 0.1717,
+      "step": 26580
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.915761961644891e-08,
+      "loss": 0.1717,
+      "step": 26581
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9083746883095113e-08,
+      "loss": 0.1808,
+      "step": 26582
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.901001671744229e-08,
+      "loss": 0.1674,
+      "step": 26583
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8936429120546274e-08,
+      "loss": 0.1653,
+      "step": 26584
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8862984093456215e-08,
+      "loss": 0.1713,
+      "step": 26585
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.878968163722128e-08,
+      "loss": 0.1709,
+      "step": 26586
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8716521752888406e-08,
+      "loss": 0.1688,
+      "step": 26587
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8643504441502314e-08,
+      "loss": 0.1713,
+      "step": 26588
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8570629704107723e-08,
+      "loss": 0.1981,
+      "step": 26589
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8497897541744914e-08,
+      "loss": 0.175,
+      "step": 26590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8425307955451943e-08,
+      "loss": 0.148,
+      "step": 26591
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.835286094626576e-08,
+      "loss": 0.1799,
+      "step": 26592
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8280556515221092e-08,
+      "loss": 0.1732,
+      "step": 26593
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8208394663352668e-08,
+      "loss": 0.1668,
+      "step": 26594
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8136375391688554e-08,
+      "loss": 0.1834,
+      "step": 26595
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8064498701259036e-08,
+      "loss": 0.1578,
+      "step": 26596
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7992764593089964e-08,
+      "loss": 0.1602,
+      "step": 26597
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7921173068207177e-08,
+      "loss": 0.1758,
+      "step": 26598
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7849724127632085e-08,
+      "loss": 0.1734,
+      "step": 26599
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.777841777238609e-08,
+      "loss": 0.1908,
+      "step": 26600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7707254003487274e-08,
+      "loss": 0.175,
+      "step": 26601
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7636232821952594e-08,
+      "loss": 0.1681,
+      "step": 26602
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7565354228795685e-08,
+      "loss": 0.1559,
+      "step": 26603
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.749461822503018e-08,
+      "loss": 0.1999,
+      "step": 26604
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7424024811665274e-08,
+      "loss": 0.1693,
+      "step": 26605
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.735357398971016e-08,
+      "loss": 0.1744,
+      "step": 26606
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7283265760170696e-08,
+      "loss": 0.1619,
+      "step": 26607
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.721310012405164e-08,
+      "loss": 0.1835,
+      "step": 26608
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7143077082354408e-08,
+      "loss": 0.1652,
+      "step": 26609
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7073196636079315e-08,
+      "loss": 0.1659,
+      "step": 26610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7003458786226666e-08,
+      "loss": 0.1651,
+      "step": 26611
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6933863533789007e-08,
+      "loss": 0.1664,
+      "step": 26612
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6864410879763316e-08,
+      "loss": 0.1658,
+      "step": 26613
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6795100825141018e-08,
+      "loss": 0.182,
+      "step": 26614
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6725933370910217e-08,
+      "loss": 0.1836,
+      "step": 26615
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6656908518062343e-08,
+      "loss": 0.1682,
+      "step": 26616
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6588026267579936e-08,
+      "loss": 0.1784,
+      "step": 26617
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6519286620448883e-08,
+      "loss": 0.1656,
+      "step": 26618
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6450689577651724e-08,
+      "loss": 0.176,
+      "step": 26619
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.638223514016546e-08,
+      "loss": 0.1739,
+      "step": 26620
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6313923308971525e-08,
+      "loss": 0.168,
+      "step": 26621
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6245754085042475e-08,
+      "loss": 0.1774,
+      "step": 26622
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.617772746935309e-08,
+      "loss": 0.168,
+      "step": 26623
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6109843462875917e-08,
+      "loss": 0.1839,
+      "step": 26624
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.604210206658019e-08,
+      "loss": 0.1561,
+      "step": 26625
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5974503281432907e-08,
+      "loss": 0.191,
+      "step": 26626
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5907047108401076e-08,
+      "loss": 0.176,
+      "step": 26627
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5839733548446146e-08,
+      "loss": 0.1673,
+      "step": 26628
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5772562602531793e-08,
+      "loss": 0.1539,
+      "step": 26629
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.570553427161503e-08,
+      "loss": 0.1729,
+      "step": 26630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5638648556656198e-08,
+      "loss": 0.1723,
+      "step": 26631
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5571905458608984e-08,
+      "loss": 0.1452,
+      "step": 26632
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.550530497842595e-08,
+      "loss": 0.165,
+      "step": 26633
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5438847117060785e-08,
+      "loss": 0.1668,
+      "step": 26634
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5372531875460505e-08,
+      "loss": 0.1754,
+      "step": 26635
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5306359254573244e-08,
+      "loss": 0.1829,
+      "step": 26636
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.524032925534491e-08,
+      "loss": 0.1613,
+      "step": 26637
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5174441878718083e-08,
+      "loss": 0.1778,
+      "step": 26638
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.510869712563312e-08,
+      "loss": 0.18,
+      "step": 26639
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5043094997030383e-08,
+      "loss": 0.1654,
+      "step": 26640
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4977635493845788e-08,
+      "loss": 0.1744,
+      "step": 26641
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4912318617015255e-08,
+      "loss": 0.1714,
+      "step": 26642
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4847144367471367e-08,
+      "loss": 0.1636,
+      "step": 26643
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4782112746145605e-08,
+      "loss": 0.1599,
+      "step": 26644
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4717223753967224e-08,
+      "loss": 0.1857,
+      "step": 26645
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4652477391861042e-08,
+      "loss": 0.1714,
+      "step": 26646
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4587873660754092e-08,
+      "loss": 0.1786,
+      "step": 26647
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4523412561567862e-08,
+      "loss": 0.1927,
+      "step": 26648
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4459094095222725e-08,
+      "loss": 0.1956,
+      "step": 26649
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4394918262640167e-08,
+      "loss": 0.1688,
+      "step": 26650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4330885064732791e-08,
+      "loss": 0.1658,
+      "step": 26651
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4266994502418752e-08,
+      "loss": 0.1745,
+      "step": 26652
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4203246576609542e-08,
+      "loss": 0.1634,
+      "step": 26653
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4139641288214434e-08,
+      "loss": 0.1673,
+      "step": 26654
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4076178638143811e-08,
+      "loss": 0.1821,
+      "step": 26655
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4012858627303616e-08,
+      "loss": 0.1782,
+      "step": 26656
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3949681256597569e-08,
+      "loss": 0.1724,
+      "step": 26657
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3886646526928283e-08,
+      "loss": 0.1864,
+      "step": 26658
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3823754439197257e-08,
+      "loss": 0.1728,
+      "step": 26659
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3761004994301553e-08,
+      "loss": 0.1544,
+      "step": 26660
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.369839819313823e-08,
+      "loss": 0.1665,
+      "step": 26661
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3635934036602128e-08,
+      "loss": 0.1682,
+      "step": 26662
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3573612525583646e-08,
+      "loss": 0.1654,
+      "step": 26663
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3511433660974294e-08,
+      "loss": 0.1701,
+      "step": 26664
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3449397443662248e-08,
+      "loss": 0.1756,
+      "step": 26665
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3387503874532359e-08,
+      "loss": 0.1873,
+      "step": 26666
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3325752954471693e-08,
+      "loss": 0.1625,
+      "step": 26667
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3264144684358437e-08,
+      "loss": 0.1777,
+      "step": 26668
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.320267906507522e-08,
+      "loss": 0.1849,
+      "step": 26669
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3141356097500225e-08,
+      "loss": 0.1746,
+      "step": 26670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3080175782507198e-08,
+      "loss": 0.1798,
+      "step": 26671
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3019138120972107e-08,
+      "loss": 0.1656,
+      "step": 26672
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2958243113765368e-08,
+      "loss": 0.1861,
+      "step": 26673
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2897490761758503e-08,
+      "loss": 0.1671,
+      "step": 26674
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2836881065817485e-08,
+      "loss": 0.1887,
+      "step": 26675
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2776414026808293e-08,
+      "loss": 0.1635,
+      "step": 26676
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2716089645595787e-08,
+      "loss": 0.183,
+      "step": 26677
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.26559079230415e-08,
+      "loss": 0.171,
+      "step": 26678
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2595868860004746e-08,
+      "loss": 0.1655,
+      "step": 26679
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2535972457342615e-08,
+      "loss": 0.1605,
+      "step": 26680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2476218715909983e-08,
+      "loss": 0.1706,
+      "step": 26681
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2416607636562828e-08,
+      "loss": 0.1669,
+      "step": 26682
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2357139220150471e-08,
+      "loss": 0.1772,
+      "step": 26683
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2297813467524456e-08,
+      "loss": 0.1714,
+      "step": 26684
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.223863037952966e-08,
+      "loss": 0.1686,
+      "step": 26685
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2179589957013183e-08,
+      "loss": 0.1515,
+      "step": 26686
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2120692200817686e-08,
+      "loss": 0.1698,
+      "step": 26687
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2061937111784715e-08,
+      "loss": 0.1722,
+      "step": 26688
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2003324690752494e-08,
+      "loss": 0.1581,
+      "step": 26689
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1944854938560347e-08,
+      "loss": 0.1819,
+      "step": 26690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1886527856042051e-08,
+      "loss": 0.1568,
+      "step": 26691
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1828343444031387e-08,
+      "loss": 0.1718,
+      "step": 26692
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1770301703357689e-08,
+      "loss": 0.1721,
+      "step": 26693
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1712402634852516e-08,
+      "loss": 0.1981,
+      "step": 26694
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1654646239341872e-08,
+      "loss": 0.1831,
+      "step": 26695
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1597032517650653e-08,
+      "loss": 0.168,
+      "step": 26696
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1539561470601535e-08,
+      "loss": 0.1685,
+      "step": 26697
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1482233099016082e-08,
+      "loss": 0.1786,
+      "step": 26698
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.142504740371253e-08,
+      "loss": 0.1619,
+      "step": 26699
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1368004385508002e-08,
+      "loss": 0.1915,
+      "step": 26700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1311104045217403e-08,
+      "loss": 0.173,
+      "step": 26701
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1254346383654525e-08,
+      "loss": 0.1618,
+      "step": 26702
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.119773140162761e-08,
+      "loss": 0.1687,
+      "step": 26703
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1141259099947122e-08,
+      "loss": 0.1593,
+      "step": 26704
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1084929479419082e-08,
+      "loss": 0.187,
+      "step": 26705
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1028742540848403e-08,
+      "loss": 0.1581,
+      "step": 26706
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0972698285037775e-08,
+      "loss": 0.1865,
+      "step": 26707
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0916796712787669e-08,
+      "loss": 0.1569,
+      "step": 26708
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0861037824896337e-08,
+      "loss": 0.1958,
+      "step": 26709
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0805421622160917e-08,
+      "loss": 0.1784,
+      "step": 26710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0749948105375218e-08,
+      "loss": 0.1774,
+      "step": 26711
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0694617275330832e-08,
+      "loss": 0.1796,
+      "step": 26712
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0639429132819345e-08,
+      "loss": 0.1852,
+      "step": 26713
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0584383678629019e-08,
+      "loss": 0.1835,
+      "step": 26714
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0529480913545887e-08,
+      "loss": 0.1777,
+      "step": 26715
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0474720838353769e-08,
+      "loss": 0.1914,
+      "step": 26716
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0420103453835373e-08,
+      "loss": 0.1794,
+      "step": 26717
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0365628760771185e-08,
+      "loss": 0.1829,
+      "step": 26718
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.031129675993836e-08,
+      "loss": 0.1841,
+      "step": 26719
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0257107452112946e-08,
+      "loss": 0.1737,
+      "step": 26720
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0203060838069878e-08,
+      "loss": 0.178,
+      "step": 26721
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.014915691858076e-08,
+      "loss": 0.1825,
+      "step": 26722
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0095395694416088e-08,
+      "loss": 0.17,
+      "step": 26723
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0041777166343025e-08,
+      "loss": 0.18,
+      "step": 26724
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.988301335126516e-09,
+      "loss": 0.1677,
+      "step": 26725
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.934968201532613e-09,
+      "loss": 0.1714,
+      "step": 26726
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.88177776632182e-09,
+      "loss": 0.175,
+      "step": 26727
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.82873003025464e-09,
+      "loss": 0.1725,
+      "step": 26728
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.775824994088246e-09,
+      "loss": 0.1896,
+      "step": 26729
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.723062658578696e-09,
+      "loss": 0.1403,
+      "step": 26730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.670443024478727e-09,
+      "loss": 0.191,
+      "step": 26731
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.617966092541064e-09,
+      "loss": 0.1978,
+      "step": 26732
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.565631863514002e-09,
+      "loss": 0.1718,
+      "step": 26733
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.51344033814694e-09,
+      "loss": 0.183,
+      "step": 26734
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.461391517183726e-09,
+      "loss": 0.1907,
+      "step": 26735
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.409485401369322e-09,
+      "loss": 0.1804,
+      "step": 26736
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.357721991444247e-09,
+      "loss": 0.1892,
+      "step": 26737
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.306101288146796e-09,
+      "loss": 0.1777,
+      "step": 26738
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.254623292216381e-09,
+      "loss": 0.1814,
+      "step": 26739
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.203288004387967e-09,
+      "loss": 0.1858,
+      "step": 26740
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.152095425393193e-09,
+      "loss": 0.1854,
+      "step": 26741
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.101045555964805e-09,
+      "loss": 0.188,
+      "step": 26742
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.05013839683111e-09,
+      "loss": 0.1755,
+      "step": 26743
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.999373948720414e-09,
+      "loss": 0.1668,
+      "step": 26744
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.948752212357691e-09,
+      "loss": 0.1632,
+      "step": 26745
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.89827318846459e-09,
+      "loss": 0.1848,
+      "step": 26746
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.847936877762752e-09,
+      "loss": 0.1889,
+      "step": 26747
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.797743280972715e-09,
+      "loss": 0.1674,
+      "step": 26748
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.74769239880835e-09,
+      "loss": 0.1741,
+      "step": 26749
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.697784231987971e-09,
+      "loss": 0.1876,
+      "step": 26750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.648018781223233e-09,
+      "loss": 0.1921,
+      "step": 26751
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.598396047223567e-09,
+      "loss": 0.1785,
+      "step": 26752
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.548916030699517e-09,
+      "loss": 0.1966,
+      "step": 26753
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.499578732357183e-09,
+      "loss": 0.1652,
+      "step": 26754
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.450384152901558e-09,
+      "loss": 0.201,
+      "step": 26755
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.401332293034304e-09,
+      "loss": 0.1696,
+      "step": 26756
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.352423153458188e-09,
+      "loss": 0.1855,
+      "step": 26757
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.303656734870435e-09,
+      "loss": 0.1816,
+      "step": 26758
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.255033037967152e-09,
+      "loss": 0.1759,
+      "step": 26759
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.20655206344334e-09,
+      "loss": 0.1637,
+      "step": 26760
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.158213811991778e-09,
+      "loss": 0.1855,
+      "step": 26761
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.110018284304132e-09,
+      "loss": 0.1404,
+      "step": 26762
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.061965481066524e-09,
+      "loss": 0.181,
+      "step": 26763
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.014055402966181e-09,
+      "loss": 0.2039,
+      "step": 26764
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.966288050686998e-09,
+      "loss": 0.1861,
+      "step": 26765
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.918663424912875e-09,
+      "loss": 0.1661,
+      "step": 26766
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.871181526323267e-09,
+      "loss": 0.1827,
+      "step": 26767
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.823842355595412e-09,
+      "loss": 0.1827,
+      "step": 26768
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.776645913406543e-09,
+      "loss": 0.1587,
+      "step": 26769
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.729592200430568e-09,
+      "loss": 0.1725,
+      "step": 26770
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.682681217340282e-09,
+      "loss": 0.1576,
+      "step": 26771
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.635912964805147e-09,
+      "loss": 0.1871,
+      "step": 26772
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.58928744349241e-09,
+      "loss": 0.192,
+      "step": 26773
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.542804654069313e-09,
+      "loss": 0.1653,
+      "step": 26774
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.496464597198661e-09,
+      "loss": 0.183,
+      "step": 26775
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.450267273543255e-09,
+      "loss": 0.1812,
+      "step": 26776
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.404212683763678e-09,
+      "loss": 0.1578,
+      "step": 26777
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.358300828516074e-09,
+      "loss": 0.1599,
+      "step": 26778
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.312531708457693e-09,
+      "loss": 0.1771,
+      "step": 26779
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.266905324241347e-09,
+      "loss": 0.155,
+      "step": 26780
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.2214216765187365e-09,
+      "loss": 0.1649,
+      "step": 26781
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.176080765940452e-09,
+      "loss": 0.1609,
+      "step": 26782
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.130882593152644e-09,
+      "loss": 0.1779,
+      "step": 26783
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.0858271588025716e-09,
+      "loss": 0.161,
+      "step": 26784
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.040914463533055e-09,
+      "loss": 0.1643,
+      "step": 26785
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.996144507985803e-09,
+      "loss": 0.177,
+      "step": 26786
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.951517292800303e-09,
+      "loss": 0.1741,
+      "step": 26787
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.9070328186138236e-09,
+      "loss": 0.1713,
+      "step": 26788
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.862691086061413e-09,
+      "loss": 0.1596,
+      "step": 26789
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.818492095777007e-09,
+      "loss": 0.1698,
+      "step": 26790
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.774435848392325e-09,
+      "loss": 0.1779,
+      "step": 26791
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.730522344535751e-09,
+      "loss": 0.1571,
+      "step": 26792
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.686751584835671e-09,
+      "loss": 0.1761,
+      "step": 26793
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6431235699160326e-09,
+      "loss": 0.1602,
+      "step": 26794
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.599638300399669e-09,
+      "loss": 0.184,
+      "step": 26795
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.556295776909416e-09,
+      "loss": 0.1597,
+      "step": 26796
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.5130960000636674e-09,
+      "loss": 0.1805,
+      "step": 26797
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.470038970479708e-09,
+      "loss": 0.177,
+      "step": 26798
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.427124688772601e-09,
+      "loss": 0.1763,
+      "step": 26799
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.38435315555408e-09,
+      "loss": 0.1779,
+      "step": 26800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.341724371436986e-09,
+      "loss": 0.178,
+      "step": 26801
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.2992383370286125e-09,
+      "loss": 0.1618,
+      "step": 26802
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.256895052936251e-09,
+      "loss": 0.1801,
+      "step": 26803
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.2146945197660845e-09,
+      "loss": 0.1774,
+      "step": 26804
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.1726367381187426e-09,
+      "loss": 0.1779,
+      "step": 26805
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.130721708595966e-09,
+      "loss": 0.1741,
+      "step": 26806
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.088949431797276e-09,
+      "loss": 0.1772,
+      "step": 26807
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.047319908318861e-09,
+      "loss": 0.1797,
+      "step": 26808
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.005833138754691e-09,
+      "loss": 0.1758,
+      "step": 26809
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.964489123696515e-09,
+      "loss": 0.1775,
+      "step": 26810
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.923287863738303e-09,
+      "loss": 0.1722,
+      "step": 26811
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.882229359465141e-09,
+      "loss": 0.2061,
+      "step": 26812
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.841313611465449e-09,
+      "loss": 0.1522,
+      "step": 26813
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.800540620322092e-09,
+      "loss": 0.1758,
+      "step": 26814
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.7599103866190495e-09,
+      "loss": 0.1721,
+      "step": 26815
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.719422910936967e-09,
+      "loss": 0.1667,
+      "step": 26816
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6790781938520505e-09,
+      "loss": 0.1817,
+      "step": 26817
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.638876235941615e-09,
+      "loss": 0.1867,
+      "step": 26818
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.598817037780757e-09,
+      "loss": 0.1916,
+      "step": 26819
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.558900599941241e-09,
+      "loss": 0.167,
+      "step": 26820
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.519126922992612e-09,
+      "loss": 0.186,
+      "step": 26821
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.479496007503304e-09,
+      "loss": 0.1844,
+      "step": 26822
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.440007854039531e-09,
+      "loss": 0.1956,
+      "step": 26823
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.400662463165285e-09,
+      "loss": 0.1815,
+      "step": 26824
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.361459835442339e-09,
+      "loss": 0.1867,
+      "step": 26825
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.322399971431358e-09,
+      "loss": 0.1883,
+      "step": 26826
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.283482871688561e-09,
+      "loss": 0.1842,
+      "step": 26827
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.244708536772392e-09,
+      "loss": 0.1639,
+      "step": 26828
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.20607696723574e-09,
+      "loss": 0.2009,
+      "step": 26829
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.167588163629278e-09,
+      "loss": 0.1824,
+      "step": 26830
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.129242126503675e-09,
+      "loss": 0.1706,
+      "step": 26831
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.091038856407382e-09,
+      "loss": 0.1794,
+      "step": 26832
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.052978353884408e-09,
+      "loss": 0.175,
+      "step": 26833
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.015060619479873e-09,
+      "loss": 0.1772,
+      "step": 26834
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.977285653734454e-09,
+      "loss": 0.1685,
+      "step": 26835
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.939653457187721e-09,
+      "loss": 0.1626,
+      "step": 26836
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.902164030378132e-09,
+      "loss": 0.1683,
+      "step": 26837
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.864817373840813e-09,
+      "loss": 0.1853,
+      "step": 26838
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.827613488109784e-09,
+      "loss": 0.1511,
+      "step": 26839
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.790552373714619e-09,
+      "loss": 0.163,
+      "step": 26840
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.753634031186005e-09,
+      "loss": 0.2021,
+      "step": 26841
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.716858461051299e-09,
+      "loss": 0.1744,
+      "step": 26842
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.680225663835636e-09,
+      "loss": 0.1861,
+      "step": 26843
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6437356400619305e-09,
+      "loss": 0.1627,
+      "step": 26844
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6073883902519876e-09,
+      "loss": 0.1721,
+      "step": 26845
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.571183914924282e-09,
+      "loss": 0.1618,
+      "step": 26846
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5351222145972875e-09,
+      "loss": 0.1906,
+      "step": 26847
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.499203289785037e-09,
+      "loss": 0.1939,
+      "step": 26848
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.463427141000454e-09,
+      "loss": 0.1781,
+      "step": 26849
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4277937687542405e-09,
+      "loss": 0.2009,
+      "step": 26850
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3923031735571e-09,
+      "loss": 0.1682,
+      "step": 26851
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.356955355915293e-09,
+      "loss": 0.1811,
+      "step": 26852
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.321750316332862e-09,
+      "loss": 0.1689,
+      "step": 26853
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2866880553127375e-09,
+      "loss": 0.1796,
+      "step": 26854
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2517685733567404e-09,
+      "loss": 0.1804,
+      "step": 26855
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.21699187096336e-09,
+      "loss": 0.1611,
+      "step": 26856
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.182357948627758e-09,
+      "loss": 0.1798,
+      "step": 26857
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.147866806847312e-09,
+      "loss": 0.1814,
+      "step": 26858
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.113518446112741e-09,
+      "loss": 0.1829,
+      "step": 26859
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.079312866915874e-09,
+      "loss": 0.1844,
+      "step": 26860
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0452500697441e-09,
+      "loss": 0.1767,
+      "step": 26861
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0113300550859155e-09,
+      "loss": 0.1736,
+      "step": 26862
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.977552823423158e-09,
+      "loss": 0.178,
+      "step": 26863
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.943918375239885e-09,
+      "loss": 0.1742,
+      "step": 26864
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9104267110168235e-09,
+      "loss": 0.1701,
+      "step": 26865
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.8770778312313684e-09,
+      "loss": 0.1746,
+      "step": 26866
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.843871736360916e-09,
+      "loss": 0.1433,
+      "step": 26867
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.810808426879531e-09,
+      "loss": 0.1763,
+      "step": 26868
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.777887903259058e-09,
+      "loss": 0.1757,
+      "step": 26869
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.745110165969124e-09,
+      "loss": 0.1789,
+      "step": 26870
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7124752154793498e-09,
+      "loss": 0.1778,
+      "step": 26871
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6799830522549205e-09,
+      "loss": 0.1677,
+      "step": 26872
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.64763367676102e-09,
+      "loss": 0.1875,
+      "step": 26873
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6154270894572796e-09,
+      "loss": 0.1791,
+      "step": 26874
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5833632908066627e-09,
+      "loss": 0.1921,
+      "step": 26875
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5514422812665816e-09,
+      "loss": 0.1861,
+      "step": 26876
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.519664061291117e-09,
+      "loss": 0.1646,
+      "step": 26877
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4880286313365707e-09,
+      "loss": 0.1779,
+      "step": 26878
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4565359918525833e-09,
+      "loss": 0.1599,
+      "step": 26879
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.425186143289905e-09,
+      "loss": 0.1769,
+      "step": 26880
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.393979086098176e-09,
+      "loss": 0.1722,
+      "step": 26881
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3629148207203753e-09,
+      "loss": 0.1828,
+      "step": 26882
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.331993347601703e-09,
+      "loss": 0.1954,
+      "step": 26883
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3012146671829172e-09,
+      "loss": 0.1671,
+      "step": 26884
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2705787799047762e-09,
+      "loss": 0.1669,
+      "step": 26885
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2400856862035977e-09,
+      "loss": 0.1702,
+      "step": 26886
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2097353865168103e-09,
+      "loss": 0.1795,
+      "step": 26887
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1795278812762898e-09,
+      "loss": 0.175,
+      "step": 26888
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1494631709139132e-09,
+      "loss": 0.1799,
+      "step": 26889
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.119541255860448e-09,
+      "loss": 0.1657,
+      "step": 26890
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0897621365411078e-09,
+      "loss": 0.1705,
+      "step": 26891
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.06012581338222e-09,
+      "loss": 0.1833,
+      "step": 26892
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0306322868067784e-09,
+      "loss": 0.1714,
+      "step": 26893
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0012815572377785e-09,
+      "loss": 0.1826,
+      "step": 26894
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.972073625092664e-09,
+      "loss": 0.1805,
+      "step": 26895
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.943008490788879e-09,
+      "loss": 0.1667,
+      "step": 26896
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9140861547416464e-09,
+      "loss": 0.1884,
+      "step": 26897
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.88530661736397e-09,
+      "loss": 0.1664,
+      "step": 26898
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8566698790677415e-09,
+      "loss": 0.1745,
+      "step": 26899
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.828175940261524e-09,
+      "loss": 0.1776,
+      "step": 26900
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7998248013527685e-09,
+      "loss": 0.171,
+      "step": 26901
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7716164627455964e-09,
+      "loss": 0.1855,
+      "step": 26902
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7435509248430193e-09,
+      "loss": 0.1842,
+      "step": 26903
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.715628188046937e-09,
+      "loss": 0.1658,
+      "step": 26904
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.68784825275481e-09,
+      "loss": 0.1735,
+      "step": 26905
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6602111193652082e-09,
+      "loss": 0.1833,
+      "step": 26906
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6327167882711503e-09,
+      "loss": 0.1914,
+      "step": 26907
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.605365259865655e-09,
+      "loss": 0.1731,
+      "step": 26908
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5781565345406324e-09,
+      "loss": 0.1612,
+      "step": 26909
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.551090612683549e-09,
+      "loss": 0.175,
+      "step": 26910
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5241674946807626e-09,
+      "loss": 0.1644,
+      "step": 26911
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4973871809186313e-09,
+      "loss": 0.1739,
+      "step": 26912
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.470749671777961e-09,
+      "loss": 0.1683,
+      "step": 26913
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4442549676395588e-09,
+      "loss": 0.1711,
+      "step": 26914
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4179030688820103e-09,
+      "loss": 0.1686,
+      "step": 26915
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3916939758827916e-09,
+      "loss": 0.1707,
+      "step": 26916
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3656276890138275e-09,
+      "loss": 0.1687,
+      "step": 26917
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.339704208650373e-09,
+      "loss": 0.1697,
+      "step": 26918
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3139235351610222e-09,
+      "loss": 0.1888,
+      "step": 26919
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2882856689132593e-09,
+      "loss": 0.1711,
+      "step": 26920
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.262790610275678e-09,
+      "loss": 0.1506,
+      "step": 26921
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2374383596102112e-09,
+      "loss": 0.1791,
+      "step": 26922
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.212228917279902e-09,
+      "loss": 0.1909,
+      "step": 26923
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.187162283645572e-09,
+      "loss": 0.1869,
+      "step": 26924
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1622384590636038e-09,
+      "loss": 0.1751,
+      "step": 26925
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1374574438914885e-09,
+      "loss": 0.1707,
+      "step": 26926
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1128192384822775e-09,
+      "loss": 0.178,
+      "step": 26927
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0883238431879117e-09,
+      "loss": 0.1804,
+      "step": 26928
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0639712583592208e-09,
+      "loss": 0.1738,
+      "step": 26929
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0397614843437053e-09,
+      "loss": 0.1719,
+      "step": 26930
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.015694521486644e-09,
+      "loss": 0.1818,
+      "step": 26931
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9917703701322065e-09,
+      "loss": 0.1639,
+      "step": 26932
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9679890306223413e-09,
+      "loss": 0.1787,
+      "step": 26933
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9443505032956666e-09,
+      "loss": 0.1851,
+      "step": 26934
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9208547884908004e-09,
+      "loss": 0.1717,
+      "step": 26935
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8975018865441395e-09,
+      "loss": 0.1741,
+      "step": 26936
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8742917977865316e-09,
+      "loss": 0.1691,
+      "step": 26937
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.851224522552153e-09,
+      "loss": 0.1689,
+      "step": 26938
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.82830006116963e-09,
+      "loss": 0.1641,
+      "step": 26939
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8055184139664782e-09,
+      "loss": 0.1852,
+      "step": 26940
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7828795812679934e-09,
+      "loss": 0.1869,
+      "step": 26941
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.76038356339836e-09,
+      "loss": 0.1915,
+      "step": 26942
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.738030360677323e-09,
+      "loss": 0.1916,
+      "step": 26943
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7158199734246262e-09,
+      "loss": 0.1807,
+      "step": 26944
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6937524019577934e-09,
+      "loss": 0.1691,
+      "step": 26945
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6718276465932381e-09,
+      "loss": 0.1771,
+      "step": 26946
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.650045707641823e-09,
+      "loss": 0.1583,
+      "step": 26947
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.628406585416631e-09,
+      "loss": 0.1623,
+      "step": 26948
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6069102802251935e-09,
+      "loss": 0.1807,
+      "step": 26949
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5855567923761528e-09,
+      "loss": 0.1785,
+      "step": 26950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.56434612217371e-09,
+      "loss": 0.1618,
+      "step": 26951
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5432782699209559e-09,
+      "loss": 0.1752,
+      "step": 26952
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5223532359176507e-09,
+      "loss": 0.2004,
+      "step": 26953
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.501571020464665e-09,
+      "loss": 0.1912,
+      "step": 26954
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4809316238584282e-09,
+      "loss": 0.1735,
+      "step": 26955
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4604350463920392e-09,
+      "loss": 0.1659,
+      "step": 26956
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4400812883608173e-09,
+      "loss": 0.1719,
+      "step": 26957
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4198703500534207e-09,
+      "loss": 0.1788,
+      "step": 26958
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3998022317596171e-09,
+      "loss": 0.1653,
+      "step": 26959
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3798769337658446e-09,
+      "loss": 0.1803,
+      "step": 26960
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.36009445635632e-09,
+      "loss": 0.1786,
+      "step": 26961
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3404547998141504e-09,
+      "loss": 0.1615,
+      "step": 26962
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3209579644191117e-09,
+      "loss": 0.1658,
+      "step": 26963
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3016039504509803e-09,
+      "loss": 0.1817,
+      "step": 26964
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2823927581850915e-09,
+      "loss": 0.1672,
+      "step": 26965
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2633243878967804e-09,
+      "loss": 0.1838,
+      "step": 26966
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2443988398569418e-09,
+      "loss": 0.1861,
+      "step": 26967
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.22561611433758e-09,
+      "loss": 0.1537,
+      "step": 26968
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2069762116062588e-09,
+      "loss": 0.1732,
+      "step": 26969
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1884791319283219e-09,
+      "loss": 0.1855,
+      "step": 26970
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1701248755702221e-09,
+      "loss": 0.1729,
+      "step": 26971
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1519134427917522e-09,
+      "loss": 0.1677,
+      "step": 26972
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1338448338549246e-09,
+      "loss": 0.176,
+      "step": 26973
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.115919049016201e-09,
+      "loss": 0.1572,
+      "step": 26974
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0981360885331526e-09,
+      "loss": 0.164,
+      "step": 26975
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0804959526578007e-09,
+      "loss": 0.1681,
+      "step": 26976
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0629986416443861e-09,
+      "loss": 0.1676,
+      "step": 26977
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0456441557415986e-09,
+      "loss": 0.1829,
+      "step": 26978
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0284324951981284e-09,
+      "loss": 0.1628,
+      "step": 26979
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0113636602593346e-09,
+      "loss": 0.1635,
+      "step": 26980
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.944376511683562e-10,
+      "loss": 0.1678,
+      "step": 26981
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.776544681672218e-10,
+      "loss": 0.1708,
+      "step": 26982
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.610141114968496e-10,
+      "loss": 0.1547,
+      "step": 26983
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.445165813937175e-10,
+      "loss": 0.1808,
+      "step": 26984
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.281618780931923e-10,
+      "loss": 0.1674,
+      "step": 26985
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.119500018306415e-10,
+      "loss": 0.1794,
+      "step": 26986
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.958809528358814e-10,
+      "loss": 0.1641,
+      "step": 26987
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.799547313387279e-10,
+      "loss": 0.181,
+      "step": 26988
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.641713375678873e-10,
+      "loss": 0.1888,
+      "step": 26989
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.485307717476244e-10,
+      "loss": 0.1594,
+      "step": 26990
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.330330341022042e-10,
+      "loss": 0.1769,
+      "step": 26991
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.176781248514509e-10,
+      "loss": 0.1766,
+      "step": 26992
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.024660442162991e-10,
+      "loss": 0.1664,
+      "step": 26993
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.873967924143522e-10,
+      "loss": 0.1773,
+      "step": 26994
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.724703696587732e-10,
+      "loss": 0.1883,
+      "step": 26995
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.576867761638352e-10,
+      "loss": 0.1892,
+      "step": 26996
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.430460121415906e-10,
+      "loss": 0.1633,
+      "step": 26997
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.285480778007614e-10,
+      "loss": 0.1906,
+      "step": 26998
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.141929733467389e-10,
+      "loss": 0.2037,
+      "step": 26999
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.999806989871349e-10,
+      "loss": 0.1778,
+      "step": 27000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.859112549228997e-10,
+      "loss": 0.1803,
+      "step": 27001
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.71984641357204e-10,
+      "loss": 0.1722,
+      "step": 27002
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.582008584865573e-10,
+      "loss": 0.1895,
+      "step": 27003
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.445599065085795e-10,
+      "loss": 0.186,
+      "step": 27004
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.310617856186696e-10,
+      "loss": 0.1764,
+      "step": 27005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.177064960100066e-10,
+      "loss": 0.1532,
+      "step": 27006
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.044940378724385e-10,
+      "loss": 0.1781,
+      "step": 27007
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.914244113958134e-10,
+      "loss": 0.1908,
+      "step": 27008
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.784976167644285e-10,
+      "loss": 0.184,
+      "step": 27009
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.657136541659114e-10,
+      "loss": 0.1928,
+      "step": 27010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.530725237812285e-10,
+      "loss": 0.1698,
+      "step": 27011
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.405742257902358e-10,
+      "loss": 0.1696,
+      "step": 27012
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.282187603738997e-10,
+      "loss": 0.1612,
+      "step": 27013
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.160061277065253e-10,
+      "loss": 0.1618,
+      "step": 27014
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.039363279624177e-10,
+      "loss": 0.1729,
+      "step": 27015
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.920093613158816e-10,
+      "loss": 0.1903,
+      "step": 27016
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.802252279367813e-10,
+      "loss": 0.1676,
+      "step": 27017
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.685839279916504e-10,
+      "loss": 0.1763,
+      "step": 27018
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.570854616481324e-10,
+      "loss": 0.1617,
+      "step": 27019
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.4572982907165053e-10,
+      "loss": 0.1848,
+      "step": 27020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3451703042207694e-10,
+      "loss": 0.181,
+      "step": 27021
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2344706586039396e-10,
+      "loss": 0.1751,
+      "step": 27022
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.125199355453635e-10,
+      "loss": 0.1775,
+      "step": 27023
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0173563963241677e-10,
+      "loss": 0.1786,
+      "step": 27024
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9109417827587483e-10,
+      "loss": 0.1766,
+      "step": 27025
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8059555162783813e-10,
+      "loss": 0.1655,
+      "step": 27026
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7023975983818685e-10,
+      "loss": 0.1781,
+      "step": 27027
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.600268030556908e-10,
+      "loss": 0.1854,
+      "step": 27028
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.499566814235689e-10,
+      "loss": 0.1656,
+      "step": 27029
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.400293950883704e-10,
+      "loss": 0.1871,
+      "step": 27030
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3024494419109377e-10,
+      "loss": 0.1585,
+      "step": 27031
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2060332887051683e-10,
+      "loss": 0.1594,
+      "step": 27032
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1110454926652767e-10,
+      "loss": 0.1689,
+      "step": 27033
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.017486055123531e-10,
+      "loss": 0.169,
+      "step": 27034
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9253549774344024e-10,
+      "loss": 0.1624,
+      "step": 27035
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.834652260907955e-10,
+      "loss": 0.1745,
+      "step": 27036
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7453779068320475e-10,
+      "loss": 0.1529,
+      "step": 27037
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.657531916494538e-10,
+      "loss": 0.1596,
+      "step": 27038
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5711142911388764e-10,
+      "loss": 0.1532,
+      "step": 27039
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.486125032008513e-10,
+      "loss": 0.1685,
+      "step": 27040
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.40256414031359e-10,
+      "loss": 0.1777,
+      "step": 27041
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3204316172531494e-10,
+      "loss": 0.1678,
+      "step": 27042
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2397274639818223e-10,
+      "loss": 0.1817,
+      "step": 27043
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1604516816764454e-10,
+      "loss": 0.1806,
+      "step": 27044
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0826042714583438e-10,
+      "loss": 0.1734,
+      "step": 27045
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0061852344377408e-10,
+      "loss": 0.1687,
+      "step": 27046
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.931194571713757e-10,
+      "loss": 0.1631,
+      "step": 27047
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8576322843411042e-10,
+      "loss": 0.1667,
+      "step": 27048
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.785498373385597e-10,
+      "loss": 0.1567,
+      "step": 27049
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.714792839879742e-10,
+      "loss": 0.165,
+      "step": 27050
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6455156848227405e-10,
+      "loss": 0.1746,
+      "step": 27051
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.577666909213793e-10,
+      "loss": 0.1809,
+      "step": 27052
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5112465140076916e-10,
+      "loss": 0.1835,
+      "step": 27053
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4462545001703299e-10,
+      "loss": 0.1591,
+      "step": 27054
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3826908686231933e-10,
+      "loss": 0.1616,
+      "step": 27055
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3205556202655624e-10,
+      "loss": 0.1787,
+      "step": 27056
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2598487559967176e-10,
+      "loss": 0.1839,
+      "step": 27057
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2005702766826334e-10,
+      "loss": 0.1761,
+      "step": 27058
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1427201831670786e-10,
+      "loss": 0.1747,
+      "step": 27059
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0862984762716189e-10,
+      "loss": 0.1711,
+      "step": 27060
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.031305156817819e-10,
+      "loss": 0.1597,
+      "step": 27061
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.777402255717327e-11,
+      "loss": 0.1732,
+      "step": 27062
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.256036833105165e-11,
+      "loss": 0.1678,
+      "step": 27063
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.748955307669171e-11,
+      "loss": 0.1828,
+      "step": 27064
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.256157686847843e-11,
+      "loss": 0.1786,
+      "step": 27065
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.777643977524563e-11,
+      "loss": 0.1849,
+      "step": 27066
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.313414186582713e-11,
+      "loss": 0.1656,
+      "step": 27067
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.86346832068363e-11,
+      "loss": 0.1775,
+      "step": 27068
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.427806386155588e-11,
+      "loss": 0.1912,
+      "step": 27069
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.006428389326858e-11,
+      "loss": 0.1696,
+      "step": 27070
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.599334336192641e-11,
+      "loss": 0.1807,
+      "step": 27071
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.2065242326371226e-11,
+      "loss": 0.1814,
+      "step": 27072
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.827998084100394e-11,
+      "loss": 0.1622,
+      "step": 27073
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.463755896022548e-11,
+      "loss": 0.1563,
+      "step": 27074
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1137976737326556e-11,
+      "loss": 0.17,
+      "step": 27075
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.778123422115698e-11,
+      "loss": 0.1761,
+      "step": 27076
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4567331460566566e-11,
+      "loss": 0.1875,
+      "step": 27077
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.149626850107446e-11,
+      "loss": 0.1686,
+      "step": 27078
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8568045385979347e-11,
+      "loss": 0.172,
+      "step": 27079
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5782662157469716e-11,
+      "loss": 0.1731,
+      "step": 27080
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.3140118855513594e-11,
+      "loss": 0.1705,
+      "step": 27081
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0640415518968783e-11,
+      "loss": 0.1657,
+      "step": 27082
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.828355218114197e-11,
+      "loss": 0.1838,
+      "step": 27083
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.606952887645008e-11,
+      "loss": 0.1646,
+      "step": 27084
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3998345638199795e-11,
+      "loss": 0.1814,
+      "step": 27085
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2070002494146694e-11,
+      "loss": 0.1884,
+      "step": 27086
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0284499473156572e-11,
+      "loss": 0.1838,
+      "step": 27087
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.641836599654341e-12,
+      "loss": 0.1737,
+      "step": 27088
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.1420138969546805e-12,
+      "loss": 0.1734,
+      "step": 27089
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.785031387262052e-12,
+      "loss": 0.1901,
+      "step": 27090
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.570889089450248e-12,
+      "loss": 0.1617,
+      "step": 27091
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4995870223930582e-12,
+      "loss": 0.1826,
+      "step": 27092
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5711251983029374e-12,
+      "loss": 0.1645,
+      "step": 27093
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.7855036338332298e-12,
+      "loss": 0.148,
+      "step": 27094
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1427223378657204e-12,
+      "loss": 0.1684,
+      "step": 27095
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.42781320392416e-13,
+      "loss": 0.1676,
+      "step": 27096
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8568058807465494e-13,
+      "loss": 0.1713,
+      "step": 27097
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.142014757377525e-14,
+      "loss": 0.1803,
+      "step": 27098
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1621,
+      "step": 27099
+    },
+    {
+      "epoch": 3.0,
+      "step": 27099,
+      "total_flos": 3.795760161023748e+19,
+      "train_loss": 0.24085783187505208,
+      "train_runtime": 34559.1787,
+      "train_samples_per_second": 100.359,
+      "train_steps_per_second": 0.784
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 27099,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1355,
+  "total_flos": 3.795760161023748e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}