diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,53526 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 8916,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.462686567164179e-08,
+      "loss": 1.2126,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4925373134328358e-07,
+      "loss": 1.2718,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2388059701492537e-07,
+      "loss": 1.377,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9850746268656716e-07,
+      "loss": 1.3273,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7313432835820895e-07,
+      "loss": 1.185,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4776119402985074e-07,
+      "loss": 1.4293,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.223880597014925e-07,
+      "loss": 1.5003,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.970149253731343e-07,
+      "loss": 1.3988,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.716417910447762e-07,
+      "loss": 1.2289,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.462686567164179e-07,
+      "loss": 1.1258,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.208955223880598e-07,
+      "loss": 1.0757,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.955223880597015e-07,
+      "loss": 1.1924,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.701492537313434e-07,
+      "loss": 1.2584,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.044776119402985e-06,
+      "loss": 1.4514,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.119402985074627e-06,
+      "loss": 0.9769,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1940298507462686e-06,
+      "loss": 1.1954,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2686567164179105e-06,
+      "loss": 1.0492,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3432835820895524e-06,
+      "loss": 0.971,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.417910447761194e-06,
+      "loss": 1.3094,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4925373134328358e-06,
+      "loss": 1.2807,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5671641791044779e-06,
+      "loss": 1.1918,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6417910447761196e-06,
+      "loss": 1.2005,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7164179104477613e-06,
+      "loss": 1.0645,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.791044776119403e-06,
+      "loss": 1.0096,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.865671641791045e-06,
+      "loss": 1.1313,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9402985074626867e-06,
+      "loss": 0.7637,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0149253731343284e-06,
+      "loss": 1.1214,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.08955223880597e-06,
+      "loss": 1.0284,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1641791044776118e-06,
+      "loss": 1.0451,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.238805970149254e-06,
+      "loss": 1.1014,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3134328358208956e-06,
+      "loss": 1.0779,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3880597014925373e-06,
+      "loss": 1.097,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4626865671641794e-06,
+      "loss": 0.8753,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.537313432835821e-06,
+      "loss": 0.9426,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6119402985074627e-06,
+      "loss": 1.0342,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.686567164179105e-06,
+      "loss": 1.0563,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7611940298507465e-06,
+      "loss": 1.0278,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.835820895522388e-06,
+      "loss": 1.0648,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9104477611940303e-06,
+      "loss": 0.9939,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9850746268656716e-06,
+      "loss": 0.9683,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0597014925373137e-06,
+      "loss": 1.0531,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1343283582089558e-06,
+      "loss": 0.873,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.208955223880597e-06,
+      "loss": 0.9544,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.283582089552239e-06,
+      "loss": 0.9348,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3582089552238813e-06,
+      "loss": 0.8899,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4328358208955225e-06,
+      "loss": 1.0018,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5074626865671646e-06,
+      "loss": 1.0771,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.582089552238806e-06,
+      "loss": 0.8388,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.656716417910448e-06,
+      "loss": 0.8724,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.73134328358209e-06,
+      "loss": 0.9213,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8059701492537314e-06,
+      "loss": 0.9177,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8805970149253735e-06,
+      "loss": 0.9278,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.955223880597015e-06,
+      "loss": 0.889,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.029850746268657e-06,
+      "loss": 0.8578,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.104477611940299e-06,
+      "loss": 0.9339,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.17910447761194e-06,
+      "loss": 0.9848,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.253731343283583e-06,
+      "loss": 0.9589,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3283582089552236e-06,
+      "loss": 0.9455,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.402985074626866e-06,
+      "loss": 0.9847,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.477611940298508e-06,
+      "loss": 0.9527,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5522388059701495e-06,
+      "loss": 0.843,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.626865671641791e-06,
+      "loss": 0.8556,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.701492537313434e-06,
+      "loss": 0.8897,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7761194029850745e-06,
+      "loss": 0.9458,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.850746268656717e-06,
+      "loss": 0.8806,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.925373134328359e-06,
+      "loss": 0.8658,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.7902,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.074626865671642e-06,
+      "loss": 0.8737,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.149253731343285e-06,
+      "loss": 0.9197,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2238805970149255e-06,
+      "loss": 0.7632,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.298507462686567e-06,
+      "loss": 0.8686,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.37313432835821e-06,
+      "loss": 0.8646,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.447761194029851e-06,
+      "loss": 0.8325,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.522388059701493e-06,
+      "loss": 0.867,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.597014925373134e-06,
+      "loss": 0.8843,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.671641791044776e-06,
+      "loss": 0.896,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.746268656716418e-06,
+      "loss": 0.8118,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.820895522388061e-06,
+      "loss": 0.981,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.895522388059702e-06,
+      "loss": 0.9047,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.970149253731343e-06,
+      "loss": 0.8302,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.044776119402986e-06,
+      "loss": 0.9722,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.119402985074627e-06,
+      "loss": 0.8719,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.194029850746269e-06,
+      "loss": 0.8026,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2686567164179116e-06,
+      "loss": 0.8514,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.343283582089553e-06,
+      "loss": 0.9115,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.417910447761194e-06,
+      "loss": 0.8595,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.492537313432837e-06,
+      "loss": 0.809,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.567164179104478e-06,
+      "loss": 0.8034,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.64179104477612e-06,
+      "loss": 0.8581,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7164179104477625e-06,
+      "loss": 0.8677,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.791044776119403e-06,
+      "loss": 0.9504,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.865671641791045e-06,
+      "loss": 0.8771,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9402985074626876e-06,
+      "loss": 0.853,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.014925373134329e-06,
+      "loss": 0.7964,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.089552238805971e-06,
+      "loss": 0.8515,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.164179104477612e-06,
+      "loss": 0.8542,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.238805970149254e-06,
+      "loss": 0.7784,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.313432835820896e-06,
+      "loss": 0.8756,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3880597014925385e-06,
+      "loss": 0.839,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.46268656716418e-06,
+      "loss": 0.8452,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.537313432835821e-06,
+      "loss": 0.7946,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.611940298507463e-06,
+      "loss": 0.784,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.686567164179105e-06,
+      "loss": 0.8746,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.761194029850747e-06,
+      "loss": 0.9198,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.835820895522389e-06,
+      "loss": 0.8102,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.91044776119403e-06,
+      "loss": 0.7727,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.985074626865672e-06,
+      "loss": 0.9227,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.059701492537314e-06,
+      "loss": 0.8723,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.134328358208955e-06,
+      "loss": 0.8054,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.208955223880599e-06,
+      "loss": 0.885,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.283582089552239e-06,
+      "loss": 0.7112,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.35820895522388e-06,
+      "loss": 0.9772,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.432835820895524e-06,
+      "loss": 0.7661,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.507462686567165e-06,
+      "loss": 0.8951,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.582089552238807e-06,
+      "loss": 0.8073,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.656716417910447e-06,
+      "loss": 0.7365,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.73134328358209e-06,
+      "loss": 0.7713,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.805970149253732e-06,
+      "loss": 0.864,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.880597014925374e-06,
+      "loss": 0.9211,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.955223880597016e-06,
+      "loss": 0.9363,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.029850746268657e-06,
+      "loss": 0.7003,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.104477611940299e-06,
+      "loss": 0.8365,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.17910447761194e-06,
+      "loss": 0.8195,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.253731343283582e-06,
+      "loss": 0.7832,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.328358208955226e-06,
+      "loss": 0.827,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.402985074626867e-06,
+      "loss": 0.8999,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.477611940298507e-06,
+      "loss": 0.7769,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.552238805970149e-06,
+      "loss": 0.7863,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.626865671641792e-06,
+      "loss": 0.8628,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.701492537313434e-06,
+      "loss": 0.8485,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.776119402985076e-06,
+      "loss": 0.8041,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.850746268656717e-06,
+      "loss": 0.7714,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.925373134328359e-06,
+      "loss": 0.8235,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.7325,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0074626865671643e-05,
+      "loss": 0.869,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0149253731343284e-05,
+      "loss": 0.8708,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0223880597014926e-05,
+      "loss": 0.9009,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.029850746268657e-05,
+      "loss": 0.8373,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.037313432835821e-05,
+      "loss": 0.8306,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0447761194029851e-05,
+      "loss": 0.865,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0522388059701494e-05,
+      "loss": 0.8249,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0597014925373134e-05,
+      "loss": 0.7834,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0671641791044778e-05,
+      "loss": 0.8357,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.074626865671642e-05,
+      "loss": 0.8432,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.082089552238806e-05,
+      "loss": 0.8343,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0895522388059703e-05,
+      "loss": 0.8556,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0970149253731344e-05,
+      "loss": 0.7508,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1044776119402986e-05,
+      "loss": 0.6653,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1119402985074628e-05,
+      "loss": 0.9305,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1194029850746268e-05,
+      "loss": 0.7513,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1268656716417911e-05,
+      "loss": 0.7814,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1343283582089553e-05,
+      "loss": 0.8466,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1417910447761196e-05,
+      "loss": 0.8821,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1492537313432836e-05,
+      "loss": 0.6644,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1567164179104478e-05,
+      "loss": 0.8716,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1641791044776121e-05,
+      "loss": 0.7955,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1716417910447761e-05,
+      "loss": 0.7127,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1791044776119405e-05,
+      "loss": 0.7241,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1865671641791046e-05,
+      "loss": 0.751,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1940298507462686e-05,
+      "loss": 0.8576,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.201492537313433e-05,
+      "loss": 0.7812,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2089552238805971e-05,
+      "loss": 0.8639,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2164179104477613e-05,
+      "loss": 0.6958,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2238805970149255e-05,
+      "loss": 0.8212,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2313432835820895e-05,
+      "loss": 0.9486,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2388059701492538e-05,
+      "loss": 0.6809,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.246268656716418e-05,
+      "loss": 0.8581,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2537313432835823e-05,
+      "loss": 0.7163,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2611940298507463e-05,
+      "loss": 0.7303,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2686567164179107e-05,
+      "loss": 0.9329,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2761194029850748e-05,
+      "loss": 0.6341,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2835820895522388e-05,
+      "loss": 0.781,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2910447761194032e-05,
+      "loss": 0.8507,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2985074626865673e-05,
+      "loss": 0.8054,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3059701492537315e-05,
+      "loss": 0.8326,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3134328358208957e-05,
+      "loss": 0.823,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3208955223880597e-05,
+      "loss": 0.7286,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.328358208955224e-05,
+      "loss": 0.7865,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3358208955223882e-05,
+      "loss": 0.7937,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3432835820895525e-05,
+      "loss": 0.7629,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3507462686567165e-05,
+      "loss": 0.7971,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3582089552238807e-05,
+      "loss": 0.7394,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.365671641791045e-05,
+      "loss": 0.8251,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.373134328358209e-05,
+      "loss": 0.9122,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3805970149253733e-05,
+      "loss": 0.882,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3880597014925375e-05,
+      "loss": 0.8002,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3955223880597015e-05,
+      "loss": 0.8264,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4029850746268658e-05,
+      "loss": 0.6971,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4104477611940298e-05,
+      "loss": 0.8755,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4179104477611942e-05,
+      "loss": 0.82,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4253731343283584e-05,
+      "loss": 0.8497,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4328358208955224e-05,
+      "loss": 0.7729,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4402985074626867e-05,
+      "loss": 0.7858,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4477611940298509e-05,
+      "loss": 0.8187,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4552238805970152e-05,
+      "loss": 0.7559,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4626865671641792e-05,
+      "loss": 0.8091,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4701492537313434e-05,
+      "loss": 0.852,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4776119402985077e-05,
+      "loss": 0.7406,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4850746268656717e-05,
+      "loss": 0.8001,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.492537313432836e-05,
+      "loss": 0.7795,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6977,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5074626865671642e-05,
+      "loss": 0.8587,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5149253731343285e-05,
+      "loss": 0.7235,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5223880597014925e-05,
+      "loss": 0.7838,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.529850746268657e-05,
+      "loss": 0.812,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.537313432835821e-05,
+      "loss": 0.8456,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5447761194029852e-05,
+      "loss": 0.6912,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5522388059701494e-05,
+      "loss": 0.8711,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5597014925373136e-05,
+      "loss": 0.7898,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5671641791044777e-05,
+      "loss": 0.6656,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.574626865671642e-05,
+      "loss": 0.8439,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.582089552238806e-05,
+      "loss": 0.6193,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5895522388059702e-05,
+      "loss": 0.8065,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5970149253731344e-05,
+      "loss": 0.9185,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6044776119402986e-05,
+      "loss": 0.8675,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6119402985074627e-05,
+      "loss": 0.7621,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.619402985074627e-05,
+      "loss": 0.8069,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.626865671641791e-05,
+      "loss": 0.8828,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6343283582089552e-05,
+      "loss": 0.8075,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6417910447761197e-05,
+      "loss": 0.861,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6492537313432836e-05,
+      "loss": 0.8203,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6567164179104477e-05,
+      "loss": 0.8674,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6641791044776122e-05,
+      "loss": 0.7539,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671641791044776e-05,
+      "loss": 0.7862,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6791044776119406e-05,
+      "loss": 0.7373,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6865671641791048e-05,
+      "loss": 0.6183,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6940298507462686e-05,
+      "loss": 0.7177,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.701492537313433e-05,
+      "loss": 0.8176,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7089552238805973e-05,
+      "loss": 0.8109,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7164179104477614e-05,
+      "loss": 0.7592,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7238805970149256e-05,
+      "loss": 0.789,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7313432835820894e-05,
+      "loss": 0.7778,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.738805970149254e-05,
+      "loss": 0.8477,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.746268656716418e-05,
+      "loss": 0.8337,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7537313432835823e-05,
+      "loss": 0.8164,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7611940298507464e-05,
+      "loss": 0.7119,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7686567164179106e-05,
+      "loss": 0.8832,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7761194029850748e-05,
+      "loss": 0.7386,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783582089552239e-05,
+      "loss": 0.9632,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.791044776119403e-05,
+      "loss": 0.7898,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7985074626865673e-05,
+      "loss": 0.7649,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8059701492537314e-05,
+      "loss": 0.832,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8134328358208956e-05,
+      "loss": 0.7294,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8208955223880598e-05,
+      "loss": 0.8697,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.828358208955224e-05,
+      "loss": 0.6806,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.835820895522388e-05,
+      "loss": 0.8931,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8432835820895526e-05,
+      "loss": 0.8203,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8507462686567165e-05,
+      "loss": 0.7199,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8582089552238806e-05,
+      "loss": 0.8152,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.865671641791045e-05,
+      "loss": 0.7693,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.873134328358209e-05,
+      "loss": 0.7244,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8805970149253735e-05,
+      "loss": 0.8404,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8880597014925376e-05,
+      "loss": 0.739,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8955223880597015e-05,
+      "loss": 0.876,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.902985074626866e-05,
+      "loss": 0.7207,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9104477611940298e-05,
+      "loss": 0.8365,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9179104477611943e-05,
+      "loss": 0.7949,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9253731343283585e-05,
+      "loss": 0.7115,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9328358208955223e-05,
+      "loss": 0.8616,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9402985074626868e-05,
+      "loss": 0.8562,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.947761194029851e-05,
+      "loss": 0.6716,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.955223880597015e-05,
+      "loss": 0.7295,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9626865671641793e-05,
+      "loss": 0.7483,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9701492537313435e-05,
+      "loss": 0.8638,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9776119402985077e-05,
+      "loss": 0.8409,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9850746268656718e-05,
+      "loss": 0.7864,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.992537313432836e-05,
+      "loss": 0.911,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.7258,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999934016039e-05,
+      "loss": 0.7211,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997360641645e-05,
+      "loss": 0.718,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999406144403e-05,
+      "loss": 0.7243,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998944256797e-05,
+      "loss": 0.7489,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983504014086e-05,
+      "loss": 0.7801,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997624578316e-05,
+      "loss": 0.7094,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999967667876143e-05,
+      "loss": 0.7647,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999957770294172e-05,
+      "loss": 0.8032,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999946553038556e-05,
+      "loss": 0.7315,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999934016110768e-05,
+      "loss": 0.8757,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999920159512468e-05,
+      "loss": 0.7353,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990498324548e-05,
+      "loss": 0.6807,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999888487311812e-05,
+      "loss": 0.8292,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987067171364e-05,
+      "loss": 0.8751,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999851536453315e-05,
+      "loss": 0.7018,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999831081533358e-05,
+      "loss": 0.6095,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980930695647e-05,
+      "loss": 0.8085,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999786212725528e-05,
+      "loss": 0.7106,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999761798843577e-05,
+      "loss": 0.7547,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999736065313843e-05,
+      "loss": 0.8208,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999709012139712e-05,
+      "loss": 0.8239,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999680639324764e-05,
+      "loss": 0.7603,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999965094687274e-05,
+      "loss": 0.8186,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961993478756e-05,
+      "loss": 0.7875,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999587603073308e-05,
+      "loss": 0.856,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999553951734265e-05,
+      "loss": 0.8485,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999518980774862e-05,
+      "loss": 0.6969,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948269019972e-05,
+      "loss": 0.8222,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999944508001362e-05,
+      "loss": 0.8359,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999940615022153e-05,
+      "loss": 0.7362,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999936590082859e-05,
+      "loss": 0.8132,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999932433184011e-05,
+      "loss": 0.7598,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999281443261575e-05,
+      "loss": 0.7316,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999237235098644e-05,
+      "loss": 0.8867,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999191707357152e-05,
+      "loss": 0.8995,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999144860043106e-05,
+      "loss": 0.832,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999096693162692e-05,
+      "loss": 0.7837,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999047206722265e-05,
+      "loss": 0.9137,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999899640072835e-05,
+      "loss": 0.8244,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998944275187658e-05,
+      "loss": 0.7782,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999889083010707e-05,
+      "loss": 0.7928,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999883606549363e-05,
+      "loss": 0.8385,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998779981354573e-05,
+      "loss": 0.7364,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99987225776973e-05,
+      "loss": 0.7953,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999866385452938e-05,
+      "loss": 0.7906,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998603811858573e-05,
+      "loss": 0.8135,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998542449692794e-05,
+      "loss": 0.8224,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999847976804014e-05,
+      "loss": 0.7113,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998415766908888e-05,
+      "loss": 0.7611,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998350446307486e-05,
+      "loss": 0.7618,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998283806244548e-05,
+      "loss": 0.7766,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999821584672887e-05,
+      "loss": 0.8035,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999814656776942e-05,
+      "loss": 0.7019,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999807596937535e-05,
+      "loss": 0.6833,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999800405155596e-05,
+      "loss": 0.7996,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997930814320754e-05,
+      "loss": 0.7189,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999785625767939e-05,
+      "loss": 0.7638,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999778038164171e-05,
+      "loss": 0.7913,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997703186217732e-05,
+      "loss": 0.7025,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997624671417636e-05,
+      "loss": 0.7684,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997544837251783e-05,
+      "loss": 0.7938,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997463683730712e-05,
+      "loss": 0.8522,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997381210865133e-05,
+      "loss": 0.8114,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999729741866593e-05,
+      "loss": 0.6877,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997212307144157e-05,
+      "loss": 0.7644,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999712587631105e-05,
+      "loss": 0.772,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997038126178015e-05,
+      "loss": 0.7863,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999694905675663e-05,
+      "loss": 0.8002,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996858668058646e-05,
+      "loss": 0.7865,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996766960096e-05,
+      "loss": 0.6895,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999667393288079e-05,
+      "loss": 0.7406,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999657958642529e-05,
+      "loss": 0.839,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996483920741955e-05,
+      "loss": 0.7049,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999638693584341e-05,
+      "loss": 0.7384,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996288631742452e-05,
+      "loss": 0.8314,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999618900845205e-05,
+      "loss": 0.7754,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999608806598536e-05,
+      "loss": 0.808,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995985804355696e-05,
+      "loss": 0.7962,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995882223576556e-05,
+      "loss": 0.7988,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995777323661606e-05,
+      "loss": 0.7489,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995671104624696e-05,
+      "loss": 0.68,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995563566479837e-05,
+      "loss": 0.719,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999545470924123e-05,
+      "loss": 0.7772,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995344532923225e-05,
+      "loss": 0.8091,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995233037540376e-05,
+      "loss": 0.8976,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995120223107393e-05,
+      "loss": 0.774,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995006089639163e-05,
+      "loss": 0.7897,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994890637150743e-05,
+      "loss": 0.7932,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994773865657375e-05,
+      "loss": 0.7914,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994655775174467e-05,
+      "loss": 0.8467,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994536365717607e-05,
+      "loss": 0.8295,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994415637302545e-05,
+      "loss": 0.7801,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994293589945224e-05,
+      "loss": 0.8764,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999417022366174e-05,
+      "loss": 0.7519,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999404553846838e-05,
+      "loss": 0.8235,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993919534381598e-05,
+      "loss": 0.7945,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999379221141802e-05,
+      "loss": 0.6548,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993663569594448e-05,
+      "loss": 0.7323,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993533608927864e-05,
+      "loss": 0.7141,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999340232943541e-05,
+      "loss": 0.8704,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993269731134414e-05,
+      "loss": 0.6581,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993135814042382e-05,
+      "loss": 0.7233,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993000578176978e-05,
+      "loss": 0.7785,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992864023556052e-05,
+      "loss": 0.754,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992726150197624e-05,
+      "loss": 0.7004,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999258695811989e-05,
+      "loss": 0.6826,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992446447341218e-05,
+      "loss": 0.6787,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999230461788015e-05,
+      "loss": 0.7965,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992161469755404e-05,
+      "loss": 0.7643,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992017002985877e-05,
+      "loss": 0.8042,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991871217590623e-05,
+      "loss": 0.7357,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991724113588883e-05,
+      "loss": 0.7764,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991575691000075e-05,
+      "loss": 0.7931,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999142594984379e-05,
+      "loss": 0.7278,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991274890139777e-05,
+      "loss": 0.8296,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999112251190798e-05,
+      "loss": 0.731,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990968815168504e-05,
+      "loss": 0.8221,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990813799941633e-05,
+      "loss": 0.8587,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990657466247826e-05,
+      "loss": 0.8859,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990499814107708e-05,
+      "loss": 0.8007,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999034084354209e-05,
+      "loss": 0.6855,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990180554571955e-05,
+      "loss": 0.7288,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990018947218445e-05,
+      "loss": 0.8399,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989856021502893e-05,
+      "loss": 0.8395,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989691777446797e-05,
+      "loss": 0.8611,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989526215071837e-05,
+      "loss": 0.7487,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998935933439986e-05,
+      "loss": 0.7938,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989191135452888e-05,
+      "loss": 0.8456,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989021618253118e-05,
+      "loss": 0.7984,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998885078282292e-05,
+      "loss": 0.8909,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988678629184835e-05,
+      "loss": 0.7596,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998850515736159e-05,
+      "loss": 0.7324,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998833036737608e-05,
+      "loss": 0.7864,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988154259251358e-05,
+      "loss": 0.8787,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987976833010676e-05,
+      "loss": 0.8004,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987798088677444e-05,
+      "loss": 0.7187,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998761802627525e-05,
+      "loss": 0.7385,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998743664582786e-05,
+      "loss": 0.7311,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987253947359208e-05,
+      "loss": 0.6694,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987069930893403e-05,
+      "loss": 0.8069,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998688459645473e-05,
+      "loss": 0.8475,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998669794406765e-05,
+      "loss": 0.792,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986509973756795e-05,
+      "loss": 0.7813,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986320685546966e-05,
+      "loss": 0.6981,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986130079463146e-05,
+      "loss": 0.8178,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985938155530494e-05,
+      "loss": 0.8165,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985744913774327e-05,
+      "loss": 0.7263,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985550354220156e-05,
+      "loss": 0.7691,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998535447689365e-05,
+      "loss": 0.7403,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985157281820667e-05,
+      "loss": 0.7881,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984958769027224e-05,
+      "loss": 0.7555,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984758938539516e-05,
+      "loss": 0.763,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998455779038392e-05,
+      "loss": 0.7485,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998435532458698e-05,
+      "loss": 0.8272,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998415154117541e-05,
+      "loss": 0.8611,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983946440176114e-05,
+      "loss": 0.8022,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998374002161614e-05,
+      "loss": 0.806,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983532285522752e-05,
+      "loss": 0.7826,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983323231923345e-05,
+      "loss": 0.8521,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983112860845518e-05,
+      "loss": 0.8132,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998290117231703e-05,
+      "loss": 0.8031,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982688166365817e-05,
+      "loss": 0.6815,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998247384301999e-05,
+      "loss": 0.8686,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998225820230783e-05,
+      "loss": 0.821,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982041244257798e-05,
+      "loss": 0.8214,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981822968898524e-05,
+      "loss": 0.7283,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981603376258815e-05,
+      "loss": 0.7037,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981382466367646e-05,
+      "loss": 0.75,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981160239254175e-05,
+      "loss": 0.7382,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980936694947726e-05,
+      "loss": 0.7855,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99807118334778e-05,
+      "loss": 0.8104,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980485654874072e-05,
+      "loss": 0.7939,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998025815916639e-05,
+      "loss": 0.7051,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980029346384774e-05,
+      "loss": 0.7528,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979799216559428e-05,
+      "loss": 0.758,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997956776972071e-05,
+      "loss": 0.7165,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979335005899174e-05,
+      "loss": 0.8441,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997910092512553e-05,
+      "loss": 0.7887,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978865527430672e-05,
+      "loss": 0.7532,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997862881284567e-05,
+      "loss": 0.7942,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978390781401752e-05,
+      "loss": 0.788,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997815143313034e-05,
+      "loss": 0.7847,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997791076806301e-05,
+      "loss": 0.7102,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977668786231536e-05,
+      "loss": 0.663,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997742548766784e-05,
+      "loss": 0.8626,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977180872404035e-05,
+      "loss": 0.8059,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976934940472402e-05,
+      "loss": 0.6801,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976687691905394e-05,
+      "loss": 0.8398,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976439126735642e-05,
+      "loss": 0.8389,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997618924499595e-05,
+      "loss": 0.8412,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975938046719288e-05,
+      "loss": 0.7424,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975685531938817e-05,
+      "loss": 0.8294,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997543170068785e-05,
+      "loss": 0.8764,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997517655299989e-05,
+      "loss": 0.7136,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974920088908605e-05,
+      "loss": 0.7395,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974662308447844e-05,
+      "loss": 0.6643,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974403211651627e-05,
+      "loss": 0.8145,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974142798554142e-05,
+      "loss": 0.8705,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973881069189754e-05,
+      "loss": 0.7327,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973618023593006e-05,
+      "loss": 0.774,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973353661798614e-05,
+      "loss": 0.8554,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973087983841458e-05,
+      "loss": 0.8114,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997282098975661e-05,
+      "loss": 0.7102,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972552679579295e-05,
+      "loss": 0.7742,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997228305334492e-05,
+      "loss": 0.701,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972012111089076e-05,
+      "loss": 0.7583,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971739852847514e-05,
+      "loss": 0.8408,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997146627865616e-05,
+      "loss": 0.7932,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971191388551125e-05,
+      "loss": 0.7606,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997091518256868e-05,
+      "loss": 0.6812,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970637660745272e-05,
+      "loss": 0.6969,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970358823117534e-05,
+      "loss": 0.8427,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997007866972226e-05,
+      "loss": 0.8755,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969797200596418e-05,
+      "loss": 0.698,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969514415777155e-05,
+      "loss": 0.7634,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996923031530179e-05,
+      "loss": 0.7945,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968944899207817e-05,
+      "loss": 0.7962,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968658167532898e-05,
+      "loss": 0.8198,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968370120314872e-05,
+      "loss": 0.7864,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968080757591758e-05,
+      "loss": 0.7704,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967790079401737e-05,
+      "loss": 0.7877,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996749808578317e-05,
+      "loss": 0.7953,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996720477677459e-05,
+      "loss": 0.7641,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966910152414706e-05,
+      "loss": 0.6294,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966614212742402e-05,
+      "loss": 0.803,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996631695779673e-05,
+      "loss": 0.8164,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966018387616918e-05,
+      "loss": 0.8033,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965718502242365e-05,
+      "loss": 0.8221,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965417301712645e-05,
+      "loss": 0.8527,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965114786067515e-05,
+      "loss": 0.7548,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964810955346895e-05,
+      "loss": 0.7685,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964505809590874e-05,
+      "loss": 0.7864,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996419934883973e-05,
+      "loss": 0.9177,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99638915731339e-05,
+      "loss": 0.8093,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963582482514003e-05,
+      "loss": 0.8045,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963272077020827e-05,
+      "loss": 0.6352,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962960356695338e-05,
+      "loss": 0.7622,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962647321578674e-05,
+      "loss": 0.7351,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962332971712143e-05,
+      "loss": 0.7777,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962017307137226e-05,
+      "loss": 0.7371,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996170032789559e-05,
+      "loss": 0.8164,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996138203402906e-05,
+      "loss": 0.725,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996106242557964e-05,
+      "loss": 0.8156,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996074150258951e-05,
+      "loss": 0.7095,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960419265101022e-05,
+      "loss": 0.7155,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960095713156695e-05,
+      "loss": 0.8429,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959770846799237e-05,
+      "loss": 0.7549,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959444666071515e-05,
+      "loss": 0.7815,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959117171016572e-05,
+      "loss": 0.826,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995878836167763e-05,
+      "loss": 0.7302,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958458238098082e-05,
+      "loss": 0.7795,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958126800321495e-05,
+      "loss": 0.6786,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957794048391603e-05,
+      "loss": 0.779,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995745998235232e-05,
+      "loss": 0.7629,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957124602247735e-05,
+      "loss": 0.8204,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956787908122105e-05,
+      "loss": 0.8473,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956449900019864e-05,
+      "loss": 0.7958,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956110577985618e-05,
+      "loss": 0.7742,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955769942064143e-05,
+      "loss": 0.8131,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99554279923004e-05,
+      "loss": 0.7458,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995508472873951e-05,
+      "loss": 0.835,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954740151426774e-05,
+      "loss": 0.7304,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954394260407662e-05,
+      "loss": 0.8885,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954047055727827e-05,
+      "loss": 0.6453,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995369853743308e-05,
+      "loss": 0.7683,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953348705569425e-05,
+      "loss": 0.7948,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995299756018302e-05,
+      "loss": 0.7183,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952645101320207e-05,
+      "loss": 0.8912,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99522913290275e-05,
+      "loss": 0.6611,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995193624335159e-05,
+      "loss": 0.7304,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951579844339325e-05,
+      "loss": 0.7014,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951222132037752e-05,
+      "loss": 0.6323,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995086310649407e-05,
+      "loss": 0.6571,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995050276775566e-05,
+      "loss": 0.6589,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950141115870075e-05,
+      "loss": 0.6843,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994977815088504e-05,
+      "loss": 0.7333,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994941387284846e-05,
+      "loss": 0.7957,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949048281808407e-05,
+      "loss": 0.8141,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994868137781312e-05,
+      "loss": 0.7007,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948313160911024e-05,
+      "loss": 0.7192,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947943631150714e-05,
+      "loss": 0.7286,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994757278858095e-05,
+      "loss": 0.8592,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947200633250675e-05,
+      "loss": 0.7491,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946827165209003e-05,
+      "loss": 0.8064,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994645238450522e-05,
+      "loss": 0.7068,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946076291188778e-05,
+      "loss": 0.7608,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945698885309318e-05,
+      "loss": 0.7179,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945320166916643e-05,
+      "loss": 0.7617,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944940136060727e-05,
+      "loss": 0.755,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944558792791726e-05,
+      "loss": 0.7689,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944176137159966e-05,
+      "loss": 0.7631,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994379216921594e-05,
+      "loss": 0.8134,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943406889010327e-05,
+      "loss": 0.7499,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943020296593967e-05,
+      "loss": 0.736,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994263239201788e-05,
+      "loss": 0.7124,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942243175333253e-05,
+      "loss": 0.6728,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941852646591452e-05,
+      "loss": 0.82,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941460805844018e-05,
+      "loss": 0.7748,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941067653142658e-05,
+      "loss": 0.7942,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940673188539254e-05,
+      "loss": 0.8524,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940277412085866e-05,
+      "loss": 0.6276,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939880323834723e-05,
+      "loss": 0.7845,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939481923838225e-05,
+      "loss": 0.7382,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939082212148954e-05,
+      "loss": 0.7171,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993868118881965e-05,
+      "loss": 0.7235,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938278853903243e-05,
+      "loss": 0.793,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993787520745283e-05,
+      "loss": 0.7656,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937470249521667e-05,
+      "loss": 0.7951,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993706398016321e-05,
+      "loss": 0.7307,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936656399431066e-05,
+      "loss": 0.7918,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936247507379023e-05,
+      "loss": 0.7403,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935837304061044e-05,
+      "loss": 0.7616,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935425789531262e-05,
+      "loss": 0.6806,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993501296384398e-05,
+      "loss": 0.8528,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993459882705368e-05,
+      "loss": 0.7794,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993418337921502e-05,
+      "loss": 0.7494,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933766620382817e-05,
+      "loss": 0.7963,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933348550612076e-05,
+      "loss": 0.8179,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932929169957967e-05,
+      "loss": 0.7431,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932508478475835e-05,
+      "loss": 0.6602,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932086476221198e-05,
+      "loss": 0.8034,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931663163249744e-05,
+      "loss": 0.7632,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931238539617335e-05,
+      "loss": 0.8072,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930812605380023e-05,
+      "loss": 0.8479,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930385360593995e-05,
+      "loss": 0.811,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992995680531565e-05,
+      "loss": 0.7373,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929526939601536e-05,
+      "loss": 0.8184,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992909576350839e-05,
+      "loss": 0.7686,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99286632770931e-05,
+      "loss": 0.8166,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992822948041275e-05,
+      "loss": 0.8088,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927794373524585e-05,
+      "loss": 0.7311,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927357956486027e-05,
+      "loss": 0.7357,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926920229354664e-05,
+      "loss": 0.6954,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926481192188267e-05,
+      "loss": 0.7659,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926040845044773e-05,
+      "loss": 0.7447,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992559918798229e-05,
+      "loss": 0.7972,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925156221059112e-05,
+      "loss": 0.8423,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924711944333688e-05,
+      "loss": 0.7498,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992426635786465e-05,
+      "loss": 0.681,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99238194617108e-05,
+      "loss": 0.7844,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992337125593112e-05,
+      "loss": 0.6925,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922921740584754e-05,
+      "loss": 0.7421,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992247091573102e-05,
+      "loss": 0.7269,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992201878142942e-05,
+      "loss": 0.7735,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921565337739622e-05,
+      "loss": 0.6472,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921110584721455e-05,
+      "loss": 0.7698,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992065452243494e-05,
+      "loss": 0.7686,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920197150940263e-05,
+      "loss": 0.8352,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919738470297784e-05,
+      "loss": 0.8433,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919278480568026e-05,
+      "loss": 0.7063,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918817181811698e-05,
+      "loss": 0.7392,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918354574089678e-05,
+      "loss": 0.7123,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917890657463013e-05,
+      "loss": 0.8131,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917425431992927e-05,
+      "loss": 0.6499,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916958897740813e-05,
+      "loss": 0.7791,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916491054768236e-05,
+      "loss": 0.7688,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916021903136945e-05,
+      "loss": 0.7161,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991555144290884e-05,
+      "loss": 0.7783,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915079674146022e-05,
+      "loss": 0.7784,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991460659691074e-05,
+      "loss": 0.7096,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914132211265423e-05,
+      "loss": 0.7321,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913656517272682e-05,
+      "loss": 0.765,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913179514995287e-05,
+      "loss": 0.753,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991270120449619e-05,
+      "loss": 0.7355,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912221585838513e-05,
+      "loss": 0.7287,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991174065908555e-05,
+      "loss": 0.7061,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911258424300767e-05,
+      "loss": 0.6674,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910774881547803e-05,
+      "loss": 0.7148,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910290030890472e-05,
+      "loss": 0.8166,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909803872392753e-05,
+      "loss": 0.7559,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909316406118816e-05,
+      "loss": 0.7909,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990882763213298e-05,
+      "loss": 0.8599,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908337550499748e-05,
+      "loss": 0.7119,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99078461612838e-05,
+      "loss": 0.7236,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907353464549982e-05,
+      "loss": 0.7366,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906859460363307e-05,
+      "loss": 0.753,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990636414878898e-05,
+      "loss": 0.7809,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905867529892358e-05,
+      "loss": 0.7367,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905369603738985e-05,
+      "loss": 0.6949,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904870370394563e-05,
+      "loss": 0.7987,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990436982992498e-05,
+      "loss": 0.7024,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990386798239629e-05,
+      "loss": 0.7776,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903364827874724e-05,
+      "loss": 0.7783,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902860366426676e-05,
+      "loss": 0.8535,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902354598118724e-05,
+      "loss": 0.7254,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990184752301761e-05,
+      "loss": 0.7104,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901339141190254e-05,
+      "loss": 0.7982,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900829452703745e-05,
+      "loss": 0.7019,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900318457625344e-05,
+      "loss": 0.7896,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899806156022488e-05,
+      "loss": 0.7642,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899292547962787e-05,
+      "loss": 0.7068,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898777633514017e-05,
+      "loss": 0.7612,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898261412744128e-05,
+      "loss": 0.7928,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989774388572125e-05,
+      "loss": 0.6918,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897225052513676e-05,
+      "loss": 0.7145,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896704913189876e-05,
+      "loss": 0.6036,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896183467818493e-05,
+      "loss": 0.7519,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895660716468345e-05,
+      "loss": 0.7585,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989513665920841e-05,
+      "loss": 0.7787,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989461129610785e-05,
+      "loss": 0.7711,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894084627235998e-05,
+      "loss": 0.7396,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893556652662357e-05,
+      "loss": 0.7233,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98930273724566e-05,
+      "loss": 0.7706,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892496786688577e-05,
+      "loss": 0.8574,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989196489542831e-05,
+      "loss": 0.6984,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891431698745988e-05,
+      "loss": 0.7136,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989089719671198e-05,
+      "loss": 0.6994,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989036138939682e-05,
+      "loss": 0.77,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889824276871218e-05,
+      "loss": 0.8159,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889285859206056e-05,
+      "loss": 0.8142,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888746136472384e-05,
+      "loss": 0.698,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888205108741437e-05,
+      "loss": 0.6217,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887662776084606e-05,
+      "loss": 0.7708,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887119138573462e-05,
+      "loss": 0.7266,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886574196279752e-05,
+      "loss": 0.7971,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886027949275387e-05,
+      "loss": 0.6917,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885480397632454e-05,
+      "loss": 0.7008,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884931541423214e-05,
+      "loss": 0.8389,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98843813807201e-05,
+      "loss": 0.7115,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988382991559571e-05,
+      "loss": 0.7834,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988327714612283e-05,
+      "loss": 0.7017,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882723072374396e-05,
+      "loss": 0.7331,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988216769442353e-05,
+      "loss": 0.7118,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988161101234353e-05,
+      "loss": 0.8006,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988105302620786e-05,
+      "loss": 0.8034,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988049373609015e-05,
+      "loss": 0.7114,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879933142064213e-05,
+      "loss": 0.8105,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879371244204026e-05,
+      "loss": 0.7469,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878808042583747e-05,
+      "loss": 0.7935,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878243537277698e-05,
+      "loss": 0.8306,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987767772836037e-05,
+      "loss": 0.8366,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877110615906442e-05,
+      "loss": 0.68,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876542199990747e-05,
+      "loss": 0.8576,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98759724806883e-05,
+      "loss": 0.736,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875401458074285e-05,
+      "loss": 0.6751,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987482913222406e-05,
+      "loss": 0.6856,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874255503213154e-05,
+      "loss": 0.8109,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873680571117265e-05,
+      "loss": 0.8501,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873104336012267e-05,
+      "loss": 0.7554,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872526797974206e-05,
+      "loss": 0.7637,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871947957079295e-05,
+      "loss": 0.7137,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871367813403927e-05,
+      "loss": 0.8223,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870786367024656e-05,
+      "loss": 0.7182,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987020361801822e-05,
+      "loss": 0.7013,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869619566461522e-05,
+      "loss": 0.7917,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986903421243164e-05,
+      "loss": 0.6967,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868447556005813e-05,
+      "loss": 0.7292,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867859597261476e-05,
+      "loss": 0.663,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867270336276207e-05,
+      "loss": 0.7729,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866679773127774e-05,
+      "loss": 0.822,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866087907894108e-05,
+      "loss": 0.7511,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986549474065333e-05,
+      "loss": 0.7861,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864900271483706e-05,
+      "loss": 0.9032,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986430450046369e-05,
+      "loss": 0.8044,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986370742767191e-05,
+      "loss": 0.8073,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986310905318715e-05,
+      "loss": 0.6761,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986250937708839e-05,
+      "loss": 0.6802,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861908399454753e-05,
+      "loss": 0.7487,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986130612036556e-05,
+      "loss": 0.7166,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860702539900288e-05,
+      "loss": 0.818,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860097658138588e-05,
+      "loss": 0.8119,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985949147516029e-05,
+      "loss": 0.7424,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985888399104539e-05,
+      "loss": 0.8059,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858275205874056e-05,
+      "loss": 0.7197,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857665119726625e-05,
+      "loss": 0.6914,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857053732683613e-05,
+      "loss": 0.8203,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98564410448257e-05,
+      "loss": 0.7115,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985582705623374e-05,
+      "loss": 0.761,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855211766988763e-05,
+      "loss": 0.7327,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854595177171968e-05,
+      "loss": 0.7741,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853977286864727e-05,
+      "loss": 0.6895,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853358096148577e-05,
+      "loss": 0.7692,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985273760510523e-05,
+      "loss": 0.8106,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852115813816578e-05,
+      "loss": 0.7729,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985149272236467e-05,
+      "loss": 0.7635,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850868330831736e-05,
+      "loss": 0.7111,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985024263930018e-05,
+      "loss": 0.7695,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984961564785257e-05,
+      "loss": 0.7871,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984898735657165e-05,
+      "loss": 0.7139,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984835776554033e-05,
+      "loss": 0.7371,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98477268748417e-05,
+      "loss": 0.7732,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984709468455902e-05,
+      "loss": 0.7466,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984646119477571e-05,
+      "loss": 0.7579,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984582640557538e-05,
+      "loss": 0.81,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845190317041797e-05,
+      "loss": 0.5983,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844552929258903e-05,
+      "loss": 0.6689,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843914242310817e-05,
+      "loss": 0.7427,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843274256281818e-05,
+      "loss": 0.7402,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984263297125637e-05,
+      "loss": 0.6896,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98419903873191e-05,
+      "loss": 0.7416,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984134650455481e-05,
+      "loss": 0.5821,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840701323048473e-05,
+      "loss": 0.8064,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840054842885228e-05,
+      "loss": 0.7836,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983940706415039e-05,
+      "loss": 0.7657,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983875798692945e-05,
+      "loss": 0.8694,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838107611308057e-05,
+      "loss": 0.7522,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983745593737205e-05,
+      "loss": 0.649,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983680296520742e-05,
+      "loss": 0.7399,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836148694900347e-05,
+      "loss": 0.7856,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835493126537162e-05,
+      "loss": 0.7408,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834836260204388e-05,
+      "loss": 0.7565,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834178095988708e-05,
+      "loss": 0.6779,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833518633976983e-05,
+      "loss": 0.6959,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832857874256234e-05,
+      "loss": 0.7383,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983219581691366e-05,
+      "loss": 0.7625,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831532462036634e-05,
+      "loss": 0.8146,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830867809712703e-05,
+      "loss": 0.7685,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983020186002957e-05,
+      "loss": 0.655,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829534613075122e-05,
+      "loss": 0.7224,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982886606893742e-05,
+      "loss": 0.7572,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828196227704682e-05,
+      "loss": 0.6962,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982752508946531e-05,
+      "loss": 0.7335,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826852654307873e-05,
+      "loss": 0.772,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826178922321108e-05,
+      "loss": 0.7591,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825503893593928e-05,
+      "loss": 0.7534,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824827568215418e-05,
+      "loss": 0.8155,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824149946274827e-05,
+      "loss": 0.8151,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982347102786158e-05,
+      "loss": 0.7037,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822790813065275e-05,
+      "loss": 0.6644,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822109301975674e-05,
+      "loss": 0.6322,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982142649468272e-05,
+      "loss": 0.6787,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982074239127652e-05,
+      "loss": 0.7802,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820056991847348e-05,
+      "loss": 0.8091,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819370296485662e-05,
+      "loss": 0.764,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818682305282083e-05,
+      "loss": 0.748,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98179930183274e-05,
+      "loss": 0.7719,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981730243571258e-05,
+      "loss": 0.7698,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981661055752875e-05,
+      "loss": 0.6787,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981591738386723e-05,
+      "loss": 0.8356,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815222914819486e-05,
+      "loss": 0.7831,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981452715047717e-05,
+      "loss": 0.7145,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98138300909321e-05,
+      "loss": 0.7026,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981313173627626e-05,
+      "loss": 0.6849,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981243208660182e-05,
+      "loss": 0.7695,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811731142001106e-05,
+      "loss": 0.8278,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811028902566617e-05,
+      "loss": 0.7701,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981032536839103e-05,
+      "loss": 0.77,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809620539567193e-05,
+      "loss": 0.7434,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808914416188108e-05,
+      "loss": 0.6694,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808206998346975e-05,
+      "loss": 0.7436,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807498286137146e-05,
+      "loss": 0.7878,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980678827965214e-05,
+      "loss": 0.6766,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980607697898567e-05,
+      "loss": 0.7895,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980536438423159e-05,
+      "loss": 0.7041,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980465049548395e-05,
+      "loss": 0.7789,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803935312836956e-05,
+      "loss": 0.7022,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980321883638499e-05,
+      "loss": 0.7096,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802501066222603e-05,
+      "loss": 0.7308,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801782002444518e-05,
+      "loss": 0.7842,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801061645145632e-05,
+      "loss": 0.728,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800339994421004e-05,
+      "loss": 0.7334,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979961705036587e-05,
+      "loss": 0.8154,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798892813075636e-05,
+      "loss": 0.794,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798167282645878e-05,
+      "loss": 0.712,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797440459172343e-05,
+      "loss": 0.7836,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796712342750946e-05,
+      "loss": 0.7278,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795982933477785e-05,
+      "loss": 0.8332,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795252231449105e-05,
+      "loss": 0.8693,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979452023676134e-05,
+      "loss": 0.7319,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793786949511095e-05,
+      "loss": 0.7076,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979305236979513e-05,
+      "loss": 0.7853,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792316497710397e-05,
+      "loss": 0.8042,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791579333354005e-05,
+      "loss": 0.6065,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979084087682323e-05,
+      "loss": 0.6016,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790101128215528e-05,
+      "loss": 0.7422,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789360087628524e-05,
+      "loss": 0.713,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788617755160012e-05,
+      "loss": 0.7908,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787874130907952e-05,
+      "loss": 0.8667,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978712921497048e-05,
+      "loss": 0.6454,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786383007445905e-05,
+      "loss": 0.7265,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785635508432697e-05,
+      "loss": 0.6856,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784886718029503e-05,
+      "loss": 0.7631,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978413663633514e-05,
+      "loss": 0.6488,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97833852634486e-05,
+      "loss": 0.7746,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978263259946903e-05,
+      "loss": 0.7586,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781878644495763e-05,
+      "loss": 0.8537,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97811233986283e-05,
+      "loss": 0.7502,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780366861966302e-05,
+      "loss": 0.7273,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977960903460961e-05,
+      "loss": 0.8154,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977884991665824e-05,
+      "loss": 0.7632,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778089508212365e-05,
+      "loss": 0.8709,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777327809372333e-05,
+      "loss": 0.695,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776564820238664e-05,
+      "loss": 0.7019,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977580054091205e-05,
+      "loss": 0.7731,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775034971493355e-05,
+      "loss": 0.7631,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774268112083605e-05,
+      "loss": 0.7578,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773499962784e-05,
+      "loss": 0.7123,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772730523695914e-05,
+      "loss": 0.8458,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771959794920886e-05,
+      "loss": 0.6862,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771187776560632e-05,
+      "loss": 0.763,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770414468717026e-05,
+      "loss": 0.7459,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769639871492126e-05,
+      "loss": 0.7545,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976886398498815e-05,
+      "loss": 0.7631,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768086809307495e-05,
+      "loss": 0.7334,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976730834455272e-05,
+      "loss": 0.7962,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766528590826556e-05,
+      "loss": 0.6562,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976574754823191e-05,
+      "loss": 0.6792,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764965216871848e-05,
+      "loss": 0.7639,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764181596849617e-05,
+      "loss": 0.7807,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763396688268632e-05,
+      "loss": 0.7495,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762610491232475e-05,
+      "loss": 0.7898,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976182300584489e-05,
+      "loss": 0.745,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761034232209812e-05,
+      "loss": 0.7497,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760244170431328e-05,
+      "loss": 0.7884,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97594528206137e-05,
+      "loss": 0.7178,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975866018286136e-05,
+      "loss": 0.6428,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757866257278912e-05,
+      "loss": 0.835,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757071043971135e-05,
+      "loss": 0.7903,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975627454304296e-05,
+      "loss": 0.7999,
+      "step": 877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975547675459951e-05,
+      "loss": 0.7958,
+      "step": 878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754677678746064e-05,
+      "loss": 0.7077,
+      "step": 879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753877315588072e-05,
+      "loss": 0.8088,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753075665231156e-05,
+      "loss": 0.7561,
+      "step": 881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975227272778111e-05,
+      "loss": 0.6564,
+      "step": 882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97514685033439e-05,
+      "loss": 0.8114,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750662992025645e-05,
+      "loss": 0.7117,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749856193932664e-05,
+      "loss": 0.8016,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749048109171415e-05,
+      "loss": 0.792,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748238737848546e-05,
+      "loss": 0.827,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747428080070868e-05,
+      "loss": 0.7045,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746616135945358e-05,
+      "loss": 0.7722,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974580290557917e-05,
+      "loss": 0.757,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974498838907962e-05,
+      "loss": 0.6911,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744172586554202e-05,
+      "loss": 0.7307,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743355498110577e-05,
+      "loss": 0.7634,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974253712385657e-05,
+      "loss": 0.8007,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741717463900183e-05,
+      "loss": 0.7038,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740896518349583e-05,
+      "loss": 0.718,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740074287313115e-05,
+      "loss": 0.739,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739250770899274e-05,
+      "loss": 0.7127,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738425969216748e-05,
+      "loss": 0.7467,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737599882374387e-05,
+      "loss": 0.7975,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736772510481198e-05,
+      "loss": 0.729,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973594385364637e-05,
+      "loss": 0.6646,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735113911979265e-05,
+      "loss": 0.805,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734282685589405e-05,
+      "loss": 0.7146,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733450174586487e-05,
+      "loss": 0.722,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973261637908037e-05,
+      "loss": 0.7325,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731781299181098e-05,
+      "loss": 0.7396,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730944934998862e-05,
+      "loss": 0.7487,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973010728664405e-05,
+      "loss": 0.7338,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972926835422719e-05,
+      "loss": 0.7339,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728428137859006e-05,
+      "loss": 0.7297,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727586637650373e-05,
+      "loss": 0.8699,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726743853712344e-05,
+      "loss": 0.7914,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972589978615614e-05,
+      "loss": 0.7163,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972505443509315e-05,
+      "loss": 0.722,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724207800634935e-05,
+      "loss": 0.6903,
+      "step": 916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723359882893218e-05,
+      "loss": 0.7822,
+      "step": 917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722510681979904e-05,
+      "loss": 0.6573,
+      "step": 918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972166019800706e-05,
+      "loss": 0.7089,
+      "step": 919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720808431086917e-05,
+      "loss": 0.7119,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719955381331883e-05,
+      "loss": 0.8038,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971910104885453e-05,
+      "loss": 0.7669,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718245433767612e-05,
+      "loss": 0.7593,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971738853618404e-05,
+      "loss": 0.8201,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971653035621689e-05,
+      "loss": 0.7028,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715670893979416e-05,
+      "loss": 0.8305,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714810149585046e-05,
+      "loss": 0.6706,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713948123147365e-05,
+      "loss": 0.6569,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713084814780134e-05,
+      "loss": 0.6636,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712220224597283e-05,
+      "loss": 0.7329,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711354352712908e-05,
+      "loss": 0.76,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971048719924128e-05,
+      "loss": 0.7894,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709618764296833e-05,
+      "loss": 0.7369,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970874904799417e-05,
+      "loss": 0.7728,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707878050448074e-05,
+      "loss": 0.796,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707005771773477e-05,
+      "loss": 0.7529,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97061322120855e-05,
+      "loss": 0.7548,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705257371499426e-05,
+      "loss": 0.8145,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704381250130698e-05,
+      "loss": 0.8304,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703503848094944e-05,
+      "loss": 0.7873,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702625165507953e-05,
+      "loss": 0.6422,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701745202485673e-05,
+      "loss": 0.7716,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970086395914424e-05,
+      "loss": 0.692,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969998143559995e-05,
+      "loss": 0.6954,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969909763196926e-05,
+      "loss": 0.7682,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969821254836881e-05,
+      "loss": 0.7276,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697326184915403e-05,
+      "loss": 0.8337,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969643854172601e-05,
+      "loss": 0.7649,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969554961891777e-05,
+      "loss": 0.7265,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694659416607992e-05,
+      "loss": 0.6669,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693767934914155e-05,
+      "loss": 0.7102,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692875173953904e-05,
+      "loss": 0.8152,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691981133845058e-05,
+      "loss": 0.7165,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96910858147056e-05,
+      "loss": 0.7602,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690189216653687e-05,
+      "loss": 0.8273,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689291339807633e-05,
+      "loss": 0.7227,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688392184285937e-05,
+      "loss": 0.7717,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687491750207255e-05,
+      "loss": 0.6346,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686590037690413e-05,
+      "loss": 0.7661,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685687046854415e-05,
+      "loss": 0.7745,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968478277781842e-05,
+      "loss": 0.7628,
+      "step": 961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683877230701766e-05,
+      "loss": 0.6956,
+      "step": 962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968297040562396e-05,
+      "loss": 0.7117,
+      "step": 963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682062302704664e-05,
+      "loss": 0.7371,
+      "step": 964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681152922063725e-05,
+      "loss": 0.7165,
+      "step": 965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968024226382115e-05,
+      "loss": 0.7503,
+      "step": 966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967933032809712e-05,
+      "loss": 0.6893,
+      "step": 967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678417115011978e-05,
+      "loss": 0.7123,
+      "step": 968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967750262468624e-05,
+      "loss": 0.7465,
+      "step": 969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676586857240588e-05,
+      "loss": 0.6553,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675669812795876e-05,
+      "loss": 0.7714,
+      "step": 971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674751491473123e-05,
+      "loss": 0.7515,
+      "step": 972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967383189339352e-05,
+      "loss": 0.7487,
+      "step": 973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672911018678418e-05,
+      "loss": 0.7072,
+      "step": 974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671988867449353e-05,
+      "loss": 0.7671,
+      "step": 975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967106543982801e-05,
+      "loss": 0.7616,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670140735936258e-05,
+      "loss": 0.7622,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669214755896125e-05,
+      "loss": 0.712,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668287499829813e-05,
+      "loss": 0.7204,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667358967859693e-05,
+      "loss": 0.7062,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966642916010829e-05,
+      "loss": 0.7529,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966549807669832e-05,
+      "loss": 0.781,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664565717752652e-05,
+      "loss": 0.7563,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663632083394327e-05,
+      "loss": 0.7054,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662697173746555e-05,
+      "loss": 0.7973,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661760988932717e-05,
+      "loss": 0.8691,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660823529076355e-05,
+      "loss": 0.6269,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659884794301186e-05,
+      "loss": 0.6977,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965894478473109e-05,
+      "loss": 0.739,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658003500490127e-05,
+      "loss": 0.7641,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965706094170251e-05,
+      "loss": 0.8416,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656117108492618e-05,
+      "loss": 0.6546,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655172000985023e-05,
+      "loss": 0.8075,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965422561930444e-05,
+      "loss": 0.6939,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653277963575757e-05,
+      "loss": 0.7202,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652329033924043e-05,
+      "loss": 0.7988,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965137883047452e-05,
+      "loss": 0.6282,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965042735335259e-05,
+      "loss": 0.7166,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649474602683814e-05,
+      "loss": 0.762,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648520578593927e-05,
+      "loss": 0.7203,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647565281208822e-05,
+      "loss": 0.6685,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646608710654576e-05,
+      "loss": 0.7907,
+      "step": 1002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645650867057423e-05,
+      "loss": 0.6949,
+      "step": 1003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964469175054377e-05,
+      "loss": 0.6592,
+      "step": 1004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964373136124018e-05,
+      "loss": 0.7091,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642769699273406e-05,
+      "loss": 0.764,
+      "step": 1006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964180676477035e-05,
+      "loss": 0.713,
+      "step": 1007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964084255785809e-05,
+      "loss": 0.801,
+      "step": 1008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639877078663867e-05,
+      "loss": 0.7001,
+      "step": 1009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638910327315096e-05,
+      "loss": 0.7317,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963794230393936e-05,
+      "loss": 0.8251,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636973008664406e-05,
+      "loss": 0.8246,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636002441618146e-05,
+      "loss": 0.7267,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635030602928663e-05,
+      "loss": 0.6651,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634057492724213e-05,
+      "loss": 0.8078,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963308311113322e-05,
+      "loss": 0.7275,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963210745828426e-05,
+      "loss": 0.8378,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631130534306094e-05,
+      "loss": 0.7026,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963015233932764e-05,
+      "loss": 0.8603,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629172873477995e-05,
+      "loss": 0.7538,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628192136886414e-05,
+      "loss": 0.771,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962721012968232e-05,
+      "loss": 0.7613,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962622685199531e-05,
+      "loss": 0.7856,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962524230395515e-05,
+      "loss": 0.659,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624256485691756e-05,
+      "loss": 0.7259,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962326939733523e-05,
+      "loss": 0.8065,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622281039015843e-05,
+      "loss": 0.8034,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962129141086402e-05,
+      "loss": 0.8524,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962030051301036e-05,
+      "loss": 0.7851,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961930834558563e-05,
+      "loss": 0.7263,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618314908720768e-05,
+      "loss": 0.7817,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961732020254687e-05,
+      "loss": 0.7956,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616324227195212e-05,
+      "loss": 0.6761,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615326982797224e-05,
+      "loss": 0.6778,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614328469484514e-05,
+      "loss": 0.7498,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961332868738885e-05,
+      "loss": 0.7929,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961232763664218e-05,
+      "loss": 0.8099,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96113253173766e-05,
+      "loss": 0.6605,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961032172972439e-05,
+      "loss": 0.6871,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609316873817992e-05,
+      "loss": 0.6576,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960831074979001e-05,
+      "loss": 0.6642,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607303357773228e-05,
+      "loss": 0.7671,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606294697900582e-05,
+      "loss": 0.6644,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605284770305186e-05,
+      "loss": 0.734,
+      "step": 1044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604273575120315e-05,
+      "loss": 0.7588,
+      "step": 1045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603261112479415e-05,
+      "loss": 0.7463,
+      "step": 1046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602247382516103e-05,
+      "loss": 0.7062,
+      "step": 1047
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601232385364154e-05,
+      "loss": 0.7547,
+      "step": 1048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960021612115752e-05,
+      "loss": 0.7076,
+      "step": 1049
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599198590030308e-05,
+      "loss": 0.6958,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959817979211681e-05,
+      "loss": 0.7752,
+      "step": 1051
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597159727551466e-05,
+      "loss": 0.7766,
+      "step": 1052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959613839646889e-05,
+      "loss": 0.8088,
+      "step": 1053
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595115799003874e-05,
+      "loss": 0.7324,
+      "step": 1054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959409193529136e-05,
+      "loss": 0.7572,
+      "step": 1055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959306680546647e-05,
+      "loss": 0.7126,
+      "step": 1056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592040409664487e-05,
+      "loss": 0.6736,
+      "step": 1057
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591012748020863e-05,
+      "loss": 0.71,
+      "step": 1058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589983820671216e-05,
+      "loss": 0.776,
+      "step": 1059
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958895362775133e-05,
+      "loss": 0.6884,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587922169397153e-05,
+      "loss": 0.7332,
+      "step": 1061
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586889445744814e-05,
+      "loss": 0.744,
+      "step": 1062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585855456930597e-05,
+      "loss": 0.7722,
+      "step": 1063
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958482020309095e-05,
+      "loss": 0.7391,
+      "step": 1064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95837836843625e-05,
+      "loss": 0.7011,
+      "step": 1065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582745900882028e-05,
+      "loss": 0.7844,
+      "step": 1066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581706852786492e-05,
+      "loss": 0.7425,
+      "step": 1067
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958066654021301e-05,
+      "loss": 0.8174,
+      "step": 1068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579624963298872e-05,
+      "loss": 0.8529,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578582122181535e-05,
+      "loss": 0.8852,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577538016998617e-05,
+      "loss": 0.7327,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576492647887908e-05,
+      "loss": 0.8806,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575446014987362e-05,
+      "loss": 0.7352,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574398118435103e-05,
+      "loss": 0.6881,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957334895836942e-05,
+      "loss": 0.7377,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572298534928764e-05,
+      "loss": 0.7459,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571246848251762e-05,
+      "loss": 0.6338,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95701938984772e-05,
+      "loss": 0.6946,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569139685744036e-05,
+      "loss": 0.6963,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568084210191388e-05,
+      "loss": 0.7366,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956702747195855e-05,
+      "loss": 0.7003,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565969471184978e-05,
+      "loss": 0.7923,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956491020801029e-05,
+      "loss": 0.8025,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956384968257427e-05,
+      "loss": 0.6889,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956278789501689e-05,
+      "loss": 0.6066,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561724845478254e-05,
+      "loss": 0.8314,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560660534098662e-05,
+      "loss": 0.6881,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559594961018562e-05,
+      "loss": 0.7507,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955852812637858e-05,
+      "loss": 0.7867,
+      "step": 1089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557460030319505e-05,
+      "loss": 0.7659,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556390672982283e-05,
+      "loss": 0.7339,
+      "step": 1091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555320054508044e-05,
+      "loss": 0.8127,
+      "step": 1092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554248175038072e-05,
+      "loss": 0.7568,
+      "step": 1093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553175034713822e-05,
+      "loss": 0.6275,
+      "step": 1094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955210063367691e-05,
+      "loss": 0.683,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551024972069127e-05,
+      "loss": 0.6754,
+      "step": 1096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549948050032422e-05,
+      "loss": 0.6546,
+      "step": 1097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954886986770892e-05,
+      "loss": 0.7549,
+      "step": 1098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95477904252409e-05,
+      "loss": 0.8057,
+      "step": 1099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546709722770817e-05,
+      "loss": 0.6952,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954562776044129e-05,
+      "loss": 0.6903,
+      "step": 1101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95445445383951e-05,
+      "loss": 0.7607,
+      "step": 1102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543460056775196e-05,
+      "loss": 0.7235,
+      "step": 1103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542374315724704e-05,
+      "loss": 0.5996,
+      "step": 1104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95412873153869e-05,
+      "loss": 0.7276,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540199055905232e-05,
+      "loss": 0.717,
+      "step": 1106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539109537423318e-05,
+      "loss": 0.732,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953801876008494e-05,
+      "loss": 0.695,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536926724034047e-05,
+      "loss": 0.6824,
+      "step": 1109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953583342941475e-05,
+      "loss": 0.7467,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534738876371328e-05,
+      "loss": 0.699,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533643065048227e-05,
+      "loss": 0.8403,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532545995590064e-05,
+      "loss": 0.7037,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953144766814161e-05,
+      "loss": 0.7637,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953034808284781e-05,
+      "loss": 0.8408,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529247239853782e-05,
+      "loss": 0.7321,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528145139304794e-05,
+      "loss": 0.6401,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952704178134629e-05,
+      "loss": 0.7154,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525937166123876e-05,
+      "loss": 0.8025,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952483129378333e-05,
+      "loss": 0.6798,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523724164470586e-05,
+      "loss": 0.72,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522615778331758e-05,
+      "loss": 0.7315,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952150613551311e-05,
+      "loss": 0.8579,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952039523616108e-05,
+      "loss": 0.8416,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519283080422276e-05,
+      "loss": 0.6483,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951816966844346e-05,
+      "loss": 0.7305,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517055000371573e-05,
+      "loss": 0.7606,
+      "step": 1127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515939076353712e-05,
+      "loss": 0.78,
+      "step": 1128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514821896537143e-05,
+      "loss": 0.7353,
+      "step": 1129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513703461069297e-05,
+      "loss": 0.6398,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951258377009778e-05,
+      "loss": 0.7379,
+      "step": 1131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511462823770343e-05,
+      "loss": 0.7293,
+      "step": 1132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951034062223492e-05,
+      "loss": 0.6936,
+      "step": 1133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950921716563961e-05,
+      "loss": 0.7277,
+      "step": 1134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508092454132664e-05,
+      "loss": 0.6102,
+      "step": 1135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950696648786252e-05,
+      "loss": 0.7432,
+      "step": 1136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505839266977757e-05,
+      "loss": 0.7711,
+      "step": 1137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950471079162714e-05,
+      "loss": 0.8024,
+      "step": 1138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503581061959588e-05,
+      "loss": 0.6619,
+      "step": 1139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950245007812419e-05,
+      "loss": 0.7359,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95013178402702e-05,
+      "loss": 0.6904,
+      "step": 1141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950018434854704e-05,
+      "loss": 0.7651,
+      "step": 1142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499049603104292e-05,
+      "loss": 0.7567,
+      "step": 1143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94979136040917e-05,
+      "loss": 0.7307,
+      "step": 1144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949677635165919e-05,
+      "loss": 0.6959,
+      "step": 1145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495637845956837e-05,
+      "loss": 0.6773,
+      "step": 1146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494498087134888e-05,
+      "loss": 0.7513,
+      "step": 1147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949335707534375e-05,
+      "loss": 0.6576,
+      "step": 1148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949221481073401e-05,
+      "loss": 0.7287,
+      "step": 1149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491071293456402e-05,
+      "loss": 0.8441,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948992652366184e-05,
+      "loss": 0.6556,
+      "step": 1151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948878050150139e-05,
+      "loss": 0.6844,
+      "step": 1152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487633227126294e-05,
+      "loss": 0.8306,
+      "step": 1153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486484700687955e-05,
+      "loss": 0.7668,
+      "step": 1154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948533492233794e-05,
+      "loss": 0.655,
+      "step": 1155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948418389222799e-05,
+      "loss": 0.662,
+      "step": 1156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483031610509996e-05,
+      "loss": 0.6914,
+      "step": 1157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948187807733602e-05,
+      "loss": 0.6555,
+      "step": 1158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480723292858302e-05,
+      "loss": 0.7506,
+      "step": 1159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479567257229227e-05,
+      "loss": 0.7283,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947840997060136e-05,
+      "loss": 0.8466,
+      "step": 1161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947725143312742e-05,
+      "loss": 0.7787,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476091644960305e-05,
+      "loss": 0.7722,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474930606253065e-05,
+      "loss": 0.5848,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947376831715892e-05,
+      "loss": 0.8099,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472604777831258e-05,
+      "loss": 0.6137,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471439988423625e-05,
+      "loss": 0.7392,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470273949089733e-05,
+      "loss": 0.7918,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469106659983467e-05,
+      "loss": 0.6327,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467938121258873e-05,
+      "loss": 0.8433,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466768333070157e-05,
+      "loss": 0.647,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465597295571694e-05,
+      "loss": 0.7484,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464425008918025e-05,
+      "loss": 0.7301,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463251473263853e-05,
+      "loss": 0.6387,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462076688764047e-05,
+      "loss": 0.7802,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946090065557364e-05,
+      "loss": 0.7406,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459723373847833e-05,
+      "loss": 0.6786,
+      "step": 1177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458544843741988e-05,
+      "loss": 0.7357,
+      "step": 1178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457365065411637e-05,
+      "loss": 0.6665,
+      "step": 1179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456184039012465e-05,
+      "loss": 0.7095,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455001764700334e-05,
+      "loss": 0.775,
+      "step": 1181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945381824263127e-05,
+      "loss": 0.7742,
+      "step": 1182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945263347296145e-05,
+      "loss": 0.7191,
+      "step": 1183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945144745584724e-05,
+      "loss": 0.8053,
+      "step": 1184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945026019144514e-05,
+      "loss": 0.712,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449071679911843e-05,
+      "loss": 0.6326,
+      "step": 1186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447881921404187e-05,
+      "loss": 0.7609,
+      "step": 1187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944669091607919e-05,
+      "loss": 0.7471,
+      "step": 1188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944549866409402e-05,
+      "loss": 0.6507,
+      "step": 1189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444305165606017e-05,
+      "loss": 0.767,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443110420772684e-05,
+      "loss": 0.759,
+      "step": 1191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441914429751692e-05,
+      "loss": 0.7226,
+      "step": 1192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440717192700868e-05,
+      "loss": 0.7213,
+      "step": 1193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943951870977822e-05,
+      "loss": 0.7277,
+      "step": 1194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438318981141894e-05,
+      "loss": 0.6387,
+      "step": 1195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437118006950227e-05,
+      "loss": 0.7097,
+      "step": 1196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435915787361705e-05,
+      "loss": 0.7842,
+      "step": 1197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434712322534983e-05,
+      "loss": 0.7442,
+      "step": 1198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943350761262888e-05,
+      "loss": 0.7964,
+      "step": 1199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432301657802378e-05,
+      "loss": 0.7161,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431094458214627e-05,
+      "loss": 0.7339,
+      "step": 1201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429886014024935e-05,
+      "loss": 0.7045,
+      "step": 1202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428676325392782e-05,
+      "loss": 0.7695,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427465392477803e-05,
+      "loss": 0.8016,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426253215439807e-05,
+      "loss": 0.8313,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942503979443876e-05,
+      "loss": 0.776,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942382512963479e-05,
+      "loss": 0.7044,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422609221188208e-05,
+      "loss": 0.8057,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942139206925946e-05,
+      "loss": 0.695,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942017367400918e-05,
+      "loss": 0.595,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941895403559815e-05,
+      "loss": 0.6524,
+      "step": 1211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941773315418733e-05,
+      "loss": 0.7317,
+      "step": 1212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416511029937837e-05,
+      "loss": 0.7713,
+      "step": 1213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415287663010945e-05,
+      "loss": 0.7723,
+      "step": 1214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414063053568105e-05,
+      "loss": 0.5608,
+      "step": 1215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412837201770925e-05,
+      "loss": 0.6705,
+      "step": 1216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411610107781178e-05,
+      "loss": 0.701,
+      "step": 1217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94103817717608e-05,
+      "loss": 0.7369,
+      "step": 1218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409152193871895e-05,
+      "loss": 0.6859,
+      "step": 1219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407921374276726e-05,
+      "loss": 0.7354,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940668931313772e-05,
+      "loss": 0.8076,
+      "step": 1221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940545601061747e-05,
+      "loss": 0.735,
+      "step": 1222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404221466878737e-05,
+      "loss": 0.7537,
+      "step": 1223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402985682084434e-05,
+      "loss": 0.7623,
+      "step": 1224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940174865639765e-05,
+      "loss": 0.6966,
+      "step": 1225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400510389981632e-05,
+      "loss": 0.6711,
+      "step": 1226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939927088299979e-05,
+      "loss": 0.8318,
+      "step": 1227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398030135615703e-05,
+      "loss": 0.6505,
+      "step": 1228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396788147993103e-05,
+      "loss": 0.7488,
+      "step": 1229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395544920295894e-05,
+      "loss": 0.7631,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394300452688148e-05,
+      "loss": 0.6455,
+      "step": 1231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393054745334092e-05,
+      "loss": 0.6482,
+      "step": 1232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391807798398117e-05,
+      "loss": 0.6954,
+      "step": 1233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390559612044783e-05,
+      "loss": 0.6595,
+      "step": 1234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389310186438804e-05,
+      "loss": 0.7315,
+      "step": 1235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388059521745078e-05,
+      "loss": 0.7206,
+      "step": 1236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386807618128637e-05,
+      "loss": 0.8691,
+      "step": 1237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385554475754703e-05,
+      "loss": 0.8329,
+      "step": 1238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384300094788645e-05,
+      "loss": 0.7322,
+      "step": 1239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383044475396003e-05,
+      "loss": 0.7268,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381787617742476e-05,
+      "loss": 0.7635,
+      "step": 1241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380529521993935e-05,
+      "loss": 0.8744,
+      "step": 1242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93792701883164e-05,
+      "loss": 0.7214,
+      "step": 1243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378009616876072e-05,
+      "loss": 0.8168,
+      "step": 1244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376747807839295e-05,
+      "loss": 0.8171,
+      "step": 1245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375484761372597e-05,
+      "loss": 0.8476,
+      "step": 1246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374220477642658e-05,
+      "loss": 0.7155,
+      "step": 1247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372954956816318e-05,
+      "loss": 0.7994,
+      "step": 1248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371688199060586e-05,
+      "loss": 0.7659,
+      "step": 1249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370420204542642e-05,
+      "loss": 0.6753,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936915097342981e-05,
+      "loss": 0.7266,
+      "step": 1251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367880505889587e-05,
+      "loss": 0.7059,
+      "step": 1252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366608802089647e-05,
+      "loss": 0.7633,
+      "step": 1253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93653358621978e-05,
+      "loss": 0.6862,
+      "step": 1254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364061686382042e-05,
+      "loss": 0.749,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362786274810518e-05,
+      "loss": 0.7158,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361509627651548e-05,
+      "loss": 0.7528,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93602317450736e-05,
+      "loss": 0.6971,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935895262724532e-05,
+      "loss": 0.7992,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357672274335506e-05,
+      "loss": 0.8407,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935639068651313e-05,
+      "loss": 0.7877,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355107863947314e-05,
+      "loss": 0.8269,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935382380680735e-05,
+      "loss": 0.8013,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352538515262697e-05,
+      "loss": 0.8651,
+      "step": 1264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935125198948297e-05,
+      "loss": 0.6967,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349964229637945e-05,
+      "loss": 0.7659,
+      "step": 1266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934867523589757e-05,
+      "loss": 0.6992,
+      "step": 1267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347385008431948e-05,
+      "loss": 0.7143,
+      "step": 1268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934609354741135e-05,
+      "loss": 0.6586,
+      "step": 1269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344800853006208e-05,
+      "loss": 0.7601,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343506925387112e-05,
+      "loss": 0.733,
+      "step": 1271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342211764724824e-05,
+      "loss": 0.7418,
+      "step": 1272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934091537119026e-05,
+      "loss": 0.6201,
+      "step": 1273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339617744954503e-05,
+      "loss": 0.7564,
+      "step": 1274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338318886188797e-05,
+      "loss": 0.662,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337018795064556e-05,
+      "loss": 0.7837,
+      "step": 1276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933571747175334e-05,
+      "loss": 0.8497,
+      "step": 1277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933441491642689e-05,
+      "loss": 0.7165,
+      "step": 1278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333111129257102e-05,
+      "loss": 0.7471,
+      "step": 1279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331806110416027e-05,
+      "loss": 0.7368,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933049986007589e-05,
+      "loss": 0.7284,
+      "step": 1281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329192378409075e-05,
+      "loss": 0.6385,
+      "step": 1282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327883665588128e-05,
+      "loss": 0.8087,
+      "step": 1283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326573721785755e-05,
+      "loss": 0.9634,
+      "step": 1284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325262547174826e-05,
+      "loss": 0.7195,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323950141928373e-05,
+      "loss": 0.7084,
+      "step": 1286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322636506219597e-05,
+      "loss": 0.7593,
+      "step": 1287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321321640221852e-05,
+      "loss": 0.9706,
+      "step": 1288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320005544108658e-05,
+      "loss": 0.8215,
+      "step": 1289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93186882180537e-05,
+      "loss": 0.7988,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317369662230817e-05,
+      "loss": 0.6974,
+      "step": 1291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316049876814023e-05,
+      "loss": 0.7071,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931472886197748e-05,
+      "loss": 0.8116,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313406617895526e-05,
+      "loss": 0.6239,
+      "step": 1294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312083144742655e-05,
+      "loss": 0.7474,
+      "step": 1295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310758442693516e-05,
+      "loss": 0.6956,
+      "step": 1296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309432511922935e-05,
+      "loss": 0.7564,
+      "step": 1297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930810535260589e-05,
+      "loss": 0.8151,
+      "step": 1298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930677696491752e-05,
+      "loss": 0.8104,
+      "step": 1299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930544734903313e-05,
+      "loss": 0.737,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304116505128194e-05,
+      "loss": 0.7605,
+      "step": 1301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302784433378333e-05,
+      "loss": 0.8155,
+      "step": 1302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301451133959342e-05,
+      "loss": 0.7306,
+      "step": 1303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300116607047167e-05,
+      "loss": 0.7569,
+      "step": 1304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298780852817932e-05,
+      "loss": 0.7249,
+      "step": 1305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297443871447906e-05,
+      "loss": 0.7436,
+      "step": 1306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296105663113533e-05,
+      "loss": 0.8548,
+      "step": 1307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294766227991414e-05,
+      "loss": 0.7365,
+      "step": 1308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293425566258306e-05,
+      "loss": 0.7814,
+      "step": 1309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292083678091136e-05,
+      "loss": 0.8101,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290740563666993e-05,
+      "loss": 0.7662,
+      "step": 1311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289396223163123e-05,
+      "loss": 0.7451,
+      "step": 1312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288050656756932e-05,
+      "loss": 0.9105,
+      "step": 1313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286703864625995e-05,
+      "loss": 0.7797,
+      "step": 1314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285355846948048e-05,
+      "loss": 0.7949,
+      "step": 1315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284006603900983e-05,
+      "loss": 0.6851,
+      "step": 1316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282656135662857e-05,
+      "loss": 0.756,
+      "step": 1317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281304442411887e-05,
+      "loss": 0.8276,
+      "step": 1318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279951524326457e-05,
+      "loss": 0.7858,
+      "step": 1319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278597381585108e-05,
+      "loss": 0.6911,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277242014366537e-05,
+      "loss": 0.9263,
+      "step": 1321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927588542284962e-05,
+      "loss": 0.8107,
+      "step": 1322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274527607213373e-05,
+      "loss": 0.6579,
+      "step": 1323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927316856763699e-05,
+      "loss": 0.7964,
+      "step": 1324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927180830429982e-05,
+      "loss": 0.7318,
+      "step": 1325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270446817381377e-05,
+      "loss": 0.8133,
+      "step": 1326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269084107061327e-05,
+      "loss": 0.7052,
+      "step": 1327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267720173519507e-05,
+      "loss": 0.6884,
+      "step": 1328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266355016935915e-05,
+      "loss": 0.6093,
+      "step": 1329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92649886374907e-05,
+      "loss": 0.7948,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263621035364192e-05,
+      "loss": 0.731,
+      "step": 1331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926225221073686e-05,
+      "loss": 0.7075,
+      "step": 1332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260882163789353e-05,
+      "loss": 0.7254,
+      "step": 1333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259510894702466e-05,
+      "loss": 0.6875,
+      "step": 1334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258138403657168e-05,
+      "loss": 0.7649,
+      "step": 1335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256764690834583e-05,
+      "loss": 0.782,
+      "step": 1336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255389756415992e-05,
+      "loss": 0.6991,
+      "step": 1337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925401360058285e-05,
+      "loss": 0.6793,
+      "step": 1338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925263622351676e-05,
+      "loss": 0.7336,
+      "step": 1339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251257625399494e-05,
+      "loss": 0.7484,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924987780641298e-05,
+      "loss": 0.7191,
+      "step": 1341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248496766739316e-05,
+      "loss": 0.8104,
+      "step": 1342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924711450656075e-05,
+      "loss": 0.7686,
+      "step": 1343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245731026059693e-05,
+      "loss": 0.7908,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244346325418726e-05,
+      "loss": 0.76,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242960404820582e-05,
+      "loss": 0.7802,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924157326444816e-05,
+      "loss": 0.7144,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240184904484514e-05,
+      "loss": 0.7598,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.7494,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237404526516602e-05,
+      "loss": 0.7821,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923601250887925e-05,
+      "loss": 0.7793,
+      "step": 1351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234619272384524e-05,
+      "loss": 0.7083,
+      "step": 1352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233224817216276e-05,
+      "loss": 0.7239,
+      "step": 1353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231829143558536e-05,
+      "loss": 0.8042,
+      "step": 1354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230432251595486e-05,
+      "loss": 0.7724,
+      "step": 1355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922903414151147e-05,
+      "loss": 0.8018,
+      "step": 1356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227634813491e-05,
+      "loss": 0.7903,
+      "step": 1357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226234267718734e-05,
+      "loss": 0.6903,
+      "step": 1358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92248325043795e-05,
+      "loss": 0.7328,
+      "step": 1359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922342952365829e-05,
+      "loss": 0.716,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922202532574025e-05,
+      "loss": 0.6891,
+      "step": 1361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922061991081069e-05,
+      "loss": 0.7344,
+      "step": 1362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219213279055082e-05,
+      "loss": 0.7773,
+      "step": 1363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217805430659053e-05,
+      "loss": 0.7238,
+      "step": 1364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921639636580839e-05,
+      "loss": 0.7235,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214986084689054e-05,
+      "loss": 0.7295,
+      "step": 1366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213574587487152e-05,
+      "loss": 0.7252,
+      "step": 1367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212161874388956e-05,
+      "loss": 0.7092,
+      "step": 1368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210747945580898e-05,
+      "loss": 0.7149,
+      "step": 1369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920933280124957e-05,
+      "loss": 0.7195,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920791644158173e-05,
+      "loss": 0.7397,
+      "step": 1371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920649886676429e-05,
+      "loss": 0.5974,
+      "step": 1372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205080076984323e-05,
+      "loss": 0.7378,
+      "step": 1373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203660072429063e-05,
+      "loss": 0.7803,
+      "step": 1374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920223885328591e-05,
+      "loss": 0.7434,
+      "step": 1375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920081641974242e-05,
+      "loss": 0.7417,
+      "step": 1376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199392771986302e-05,
+      "loss": 0.7784,
+      "step": 1377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919796791020543e-05,
+      "loss": 0.801,
+      "step": 1378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196541834587852e-05,
+      "loss": 0.6717,
+      "step": 1379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195114545321756e-05,
+      "loss": 0.7505,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193686042595495e-05,
+      "loss": 0.7394,
+      "step": 1381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192256326597594e-05,
+      "loss": 0.7517,
+      "step": 1382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190825397516727e-05,
+      "loss": 0.771,
+      "step": 1383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189393255541726e-05,
+      "loss": 0.794,
+      "step": 1384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187959900861594e-05,
+      "loss": 0.7197,
+      "step": 1385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186525333665484e-05,
+      "loss": 0.7384,
+      "step": 1386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918508955414271e-05,
+      "loss": 0.8457,
+      "step": 1387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183652562482758e-05,
+      "loss": 0.7647,
+      "step": 1388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918221435887526e-05,
+      "loss": 0.6595,
+      "step": 1389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918077494351001e-05,
+      "loss": 0.7294,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179334316576966e-05,
+      "loss": 0.8004,
+      "step": 1391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177892478266247e-05,
+      "loss": 0.7367,
+      "step": 1392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176449428768128e-05,
+      "loss": 0.6883,
+      "step": 1393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175005168273045e-05,
+      "loss": 0.7826,
+      "step": 1394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173559696971594e-05,
+      "loss": 0.7676,
+      "step": 1395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917211301505453e-05,
+      "loss": 0.8441,
+      "step": 1396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170665122712774e-05,
+      "loss": 0.7428,
+      "step": 1397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169216020137395e-05,
+      "loss": 0.6944,
+      "step": 1398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916776570751963e-05,
+      "loss": 0.7753,
+      "step": 1399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916631418505087e-05,
+      "loss": 0.7413,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164861452922676e-05,
+      "loss": 0.7432,
+      "step": 1401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163407511326762e-05,
+      "loss": 0.709,
+      "step": 1402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161952360454995e-05,
+      "loss": 0.7781,
+      "step": 1403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160496000499417e-05,
+      "loss": 0.7037,
+      "step": 1404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915903843165221e-05,
+      "loss": 0.772,
+      "step": 1405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157579654105737e-05,
+      "loss": 0.7172,
+      "step": 1406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156119668052503e-05,
+      "loss": 0.7252,
+      "step": 1407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915465847368518e-05,
+      "loss": 0.6974,
+      "step": 1408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153196071196603e-05,
+      "loss": 0.6429,
+      "step": 1409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915173246077976e-05,
+      "loss": 0.7305,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91502676426278e-05,
+      "loss": 0.7835,
+      "step": 1411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914880161693403e-05,
+      "loss": 0.7498,
+      "step": 1412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914733438389192e-05,
+      "loss": 0.795,
+      "step": 1413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91458659436951e-05,
+      "loss": 0.8031,
+      "step": 1414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144396296537356e-05,
+      "loss": 0.826,
+      "step": 1415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914292544261263e-05,
+      "loss": 0.7025,
+      "step": 1416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141453382115032e-05,
+      "loss": 0.7374,
+      "step": 1417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139980115238827e-05,
+      "loss": 0.7364,
+      "step": 1418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913850564217844e-05,
+      "loss": 0.6541,
+      "step": 1419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913702996312845e-05,
+      "loss": 0.7333,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91355530782836e-05,
+      "loss": 0.7799,
+      "step": 1421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134074987838794e-05,
+      "loss": 0.8248,
+      "step": 1422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132595691989092e-05,
+      "loss": 1.0505,
+      "step": 1423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913111519092971e-05,
+      "loss": 0.7923,
+      "step": 1424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912963348485603e-05,
+      "loss": 0.6917,
+      "step": 1425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912815057396359e-05,
+      "loss": 0.6676,
+      "step": 1426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126666458448083e-05,
+      "loss": 0.7387,
+      "step": 1427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912518113850537e-05,
+      "loss": 0.6663,
+      "step": 1428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123694614331467e-05,
+      "loss": 0.8356,
+      "step": 1429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122206886122536e-05,
+      "loss": 0.7789,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912071795407492e-05,
+      "loss": 0.8458,
+      "step": 1431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119227818385106e-05,
+      "loss": 0.7871,
+      "step": 1432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117736479249747e-05,
+      "loss": 0.7548,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911624393686565e-05,
+      "loss": 0.7693,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114750191429783e-05,
+      "loss": 0.7408,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911325524313927e-05,
+      "loss": 0.8189,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91117590921914e-05,
+      "loss": 0.8989,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110261738783616e-05,
+      "loss": 0.6974,
+      "step": 1438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108763183113522e-05,
+      "loss": 0.7385,
+      "step": 1439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107263425378873e-05,
+      "loss": 0.7357,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91057624657776e-05,
+      "loss": 0.803,
+      "step": 1441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910426030450777e-05,
+      "loss": 0.8044,
+      "step": 1442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102756941767625e-05,
+      "loss": 0.6436,
+      "step": 1443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101252377755558e-05,
+      "loss": 0.7447,
+      "step": 1444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909974661267013e-05,
+      "loss": 0.7313,
+      "step": 1445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098239646710048e-05,
+      "loss": 0.6908,
+      "step": 1446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096731480074186e-05,
+      "loss": 0.677,
+      "step": 1447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909522211296157e-05,
+      "loss": 0.7958,
+      "step": 1448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909371154557139e-05,
+      "loss": 0.7789,
+      "step": 1449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092199778102996e-05,
+      "loss": 0.7857,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090686810755887e-05,
+      "loss": 0.7055,
+      "step": 1451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908917264372973e-05,
+      "loss": 0.694,
+      "step": 1452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087657277224345e-05,
+      "loss": 0.7078,
+      "step": 1453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086140711439713e-05,
+      "loss": 0.7349,
+      "step": 1454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084622946575968e-05,
+      "loss": 0.7395,
+      "step": 1455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083103982833408e-05,
+      "loss": 0.7154,
+      "step": 1456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908158382041249e-05,
+      "loss": 0.8052,
+      "step": 1457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080062459513832e-05,
+      "loss": 0.721,
+      "step": 1458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078539900338195e-05,
+      "loss": 0.7164,
+      "step": 1459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077016143086505e-05,
+      "loss": 0.788,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907549118795986e-05,
+      "loss": 0.9512,
+      "step": 1461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073965035159502e-05,
+      "loss": 0.7627,
+      "step": 1462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072437684886828e-05,
+      "loss": 0.8368,
+      "step": 1463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907090913734341e-05,
+      "loss": 0.7474,
+      "step": 1464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069379392730956e-05,
+      "loss": 0.7705,
+      "step": 1465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067848451251353e-05,
+      "loss": 0.619,
+      "step": 1466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066316313106628e-05,
+      "loss": 0.7617,
+      "step": 1467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064782978498978e-05,
+      "loss": 0.7085,
+      "step": 1468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063248447630755e-05,
+      "loss": 0.6964,
+      "step": 1469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061712720704465e-05,
+      "loss": 0.7481,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906017579792278e-05,
+      "loss": 0.8375,
+      "step": 1471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058637679488514e-05,
+      "loss": 0.8357,
+      "step": 1472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057098365604657e-05,
+      "loss": 0.7344,
+      "step": 1473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055557856474353e-05,
+      "loss": 0.8515,
+      "step": 1474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054016152300893e-05,
+      "loss": 0.665,
+      "step": 1475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052473253287734e-05,
+      "loss": 0.7243,
+      "step": 1476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050929159638487e-05,
+      "loss": 0.7562,
+      "step": 1477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049383871556927e-05,
+      "loss": 0.782,
+      "step": 1478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047837389246983e-05,
+      "loss": 0.7289,
+      "step": 1479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046289712912735e-05,
+      "loss": 0.7036,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044740842758433e-05,
+      "loss": 0.7704,
+      "step": 1481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043190778988477e-05,
+      "loss": 0.7444,
+      "step": 1482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904163952180742e-05,
+      "loss": 0.7699,
+      "step": 1483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040087071419987e-05,
+      "loss": 0.7736,
+      "step": 1484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038533428031045e-05,
+      "loss": 0.8707,
+      "step": 1485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036978591845624e-05,
+      "loss": 0.7732,
+      "step": 1486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903542256306892e-05,
+      "loss": 0.812,
+      "step": 1487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033865341906274e-05,
+      "loss": 0.7648,
+      "step": 1488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032306928563193e-05,
+      "loss": 0.7111,
+      "step": 1489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903074732324533e-05,
+      "loss": 0.7245,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902918652615851e-05,
+      "loss": 0.8285,
+      "step": 1491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90276245375087e-05,
+      "loss": 0.6968,
+      "step": 1492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902606135750204e-05,
+      "loss": 0.8336,
+      "step": 1493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902449698634482e-05,
+      "loss": 0.7528,
+      "step": 1494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022931424243484e-05,
+      "loss": 0.7552,
+      "step": 1495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902136467140464e-05,
+      "loss": 0.6849,
+      "step": 1496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901979672803504e-05,
+      "loss": 0.7865,
+      "step": 1497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018227594341607e-05,
+      "loss": 0.7046,
+      "step": 1498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016657270531416e-05,
+      "loss": 0.752,
+      "step": 1499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015085756811706e-05,
+      "loss": 0.7918,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013513053389858e-05,
+      "loss": 0.7545,
+      "step": 1501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901193916047342e-05,
+      "loss": 0.7791,
+      "step": 1502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010364078270095e-05,
+      "loss": 0.7305,
+      "step": 1503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008787806987748e-05,
+      "loss": 0.7759,
+      "step": 1504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007210346834392e-05,
+      "loss": 0.7273,
+      "step": 1505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005631698018203e-05,
+      "loss": 0.6747,
+      "step": 1506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900405186074751e-05,
+      "loss": 0.7463,
+      "step": 1507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002470835230805e-05,
+      "loss": 0.6301,
+      "step": 1508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000888621676728e-05,
+      "loss": 0.792,
+      "step": 1509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899930522029408e-05,
+      "loss": 0.6898,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997720631291823e-05,
+      "loss": 0.7703,
+      "step": 1511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899613485487907e-05,
+      "loss": 0.7515,
+      "step": 1512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899454789126509e-05,
+      "loss": 0.7516,
+      "step": 1513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899295974065932e-05,
+      "loss": 0.7685,
+      "step": 1514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991370403271333e-05,
+      "loss": 0.6661,
+      "step": 1515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898977987931088e-05,
+      "loss": 0.8042,
+      "step": 1516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988188168987856e-05,
+      "loss": 0.7318,
+      "step": 1517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986595272512317e-05,
+      "loss": 0.7419,
+      "step": 1518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898500119009447e-05,
+      "loss": 0.8313,
+      "step": 1519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983405921944686e-05,
+      "loss": 0.7717,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981809468273488e-05,
+      "loss": 0.6974,
+      "step": 1521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898021182929156e-05,
+      "loss": 0.8006,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978613005209734e-05,
+      "loss": 0.7421,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977012996239005e-05,
+      "loss": 0.7096,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975411802590526e-05,
+      "loss": 0.7633,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973809424475602e-05,
+      "loss": 0.7297,
+      "step": 1526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897220586210569e-05,
+      "loss": 0.7009,
+      "step": 1527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897060111569242e-05,
+      "loss": 0.8012,
+      "step": 1528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968995185447554e-05,
+      "loss": 0.7232,
+      "step": 1529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967388071583033e-05,
+      "loss": 0.8147,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896577977431094e-05,
+      "loss": 0.7187,
+      "step": 1531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896417029384352e-05,
+      "loss": 0.7099,
+      "step": 1532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962559630393173e-05,
+      "loss": 0.7247,
+      "step": 1533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960947784172456e-05,
+      "loss": 0.6888,
+      "step": 1534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959334755394075e-05,
+      "loss": 0.7029,
+      "step": 1535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957720544270905e-05,
+      "loss": 0.7655,
+      "step": 1536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956105151015966e-05,
+      "loss": 0.7286,
+      "step": 1537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954488575842444e-05,
+      "loss": 0.7365,
+      "step": 1538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895287081896367e-05,
+      "loss": 0.7601,
+      "step": 1539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951251880593133e-05,
+      "loss": 0.6832,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949631760944486e-05,
+      "loss": 0.6054,
+      "step": 1541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948010460231533e-05,
+      "loss": 0.9173,
+      "step": 1542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894638797866823e-05,
+      "loss": 0.7806,
+      "step": 1543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944764316468697e-05,
+      "loss": 0.6784,
+      "step": 1544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89431394738472e-05,
+      "loss": 0.6975,
+      "step": 1545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941513451018174e-05,
+      "loss": 0.7567,
+      "step": 1546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939886248196195e-05,
+      "loss": 0.6779,
+      "step": 1547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938257865596e-05,
+      "loss": 0.6921,
+      "step": 1548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893662830343249e-05,
+      "loss": 0.6374,
+      "step": 1549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893499756192071e-05,
+      "loss": 0.7075,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933365641275873e-05,
+      "loss": 0.6455,
+      "step": 1551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893173254171333e-05,
+      "loss": 0.7261,
+      "step": 1552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930098263448603e-05,
+      "loss": 0.7012,
+      "step": 1553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928462806697363e-05,
+      "loss": 0.7806,
+      "step": 1554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892682617167544e-05,
+      "loss": 0.7456,
+      "step": 1555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925188358598815e-05,
+      "loss": 0.8269,
+      "step": 1556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923549367683626e-05,
+      "loss": 0.7317,
+      "step": 1557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921909199146167e-05,
+      "loss": 0.7421,
+      "step": 1558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920267853202897e-05,
+      "loss": 0.7353,
+      "step": 1559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8918625330070407e-05,
+      "loss": 0.7101,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916981629965464e-05,
+      "loss": 0.7127,
+      "step": 1561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915336753104986e-05,
+      "loss": 0.7738,
+      "step": 1562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891369069970604e-05,
+      "loss": 0.7141,
+      "step": 1563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912043469985855e-05,
+      "loss": 0.7695,
+      "step": 1564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891039506416181e-05,
+      "loss": 0.7416,
+      "step": 1565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908745482451444e-05,
+      "loss": 0.8566,
+      "step": 1566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890709472507245e-05,
+      "loss": 0.77,
+      "step": 1567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905442792242667e-05,
+      "loss": 0.7133,
+      "step": 1568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903789684180106e-05,
+      "loss": 0.7934,
+      "step": 1569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902135401102924e-05,
+      "loss": 0.7492,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900479943229424e-05,
+      "loss": 0.7719,
+      "step": 1571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898823310778085e-05,
+      "loss": 0.8319,
+      "step": 1572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897165503967524e-05,
+      "loss": 0.7531,
+      "step": 1573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895506523016514e-05,
+      "loss": 1.1186,
+      "step": 1574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893846368143996e-05,
+      "loss": 0.7805,
+      "step": 1575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889218503956905e-05,
+      "loss": 0.6494,
+      "step": 1576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889052253751092e-05,
+      "loss": 0.7529,
+      "step": 1577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888885886218901e-05,
+      "loss": 0.982,
+      "step": 1578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887194013822864e-05,
+      "loss": 0.6696,
+      "step": 1579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885527992632186e-05,
+      "loss": 0.751,
+      "step": 1580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883860798836845e-05,
+      "loss": 0.7385,
+      "step": 1581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882192432656852e-05,
+      "loss": 0.7733,
+      "step": 1582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880522894312378e-05,
+      "loss": 0.7632,
+      "step": 1583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878852184023754e-05,
+      "loss": 0.7065,
+      "step": 1584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877180302011452e-05,
+      "loss": 0.7154,
+      "step": 1585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875507248496117e-05,
+      "loss": 0.7309,
+      "step": 1586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873833023698527e-05,
+      "loss": 0.811,
+      "step": 1587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872157627839636e-05,
+      "loss": 0.7737,
+      "step": 1588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870481061140532e-05,
+      "loss": 0.7404,
+      "step": 1589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868803323822477e-05,
+      "loss": 0.7294,
+      "step": 1590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886712441610688e-05,
+      "loss": 0.7063,
+      "step": 1591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865444338215293e-05,
+      "loss": 0.6614,
+      "step": 1592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886376309036944e-05,
+      "loss": 0.7882,
+      "step": 1593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862080672791184e-05,
+      "loss": 0.7328,
+      "step": 1594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886039708570256e-05,
+      "loss": 0.664,
+      "step": 1595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858712329325745e-05,
+      "loss": 0.6907,
+      "step": 1596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885702640388307e-05,
+      "loss": 0.6994,
+      "step": 1597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855339309597026e-05,
+      "loss": 0.6383,
+      "step": 1598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885365104669025e-05,
+      "loss": 0.6803,
+      "step": 1599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851961615385542e-05,
+      "loss": 0.6769,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885027101590585e-05,
+      "loss": 0.7452,
+      "step": 1601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848579248474286e-05,
+      "loss": 0.6943,
+      "step": 1602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846886313314105e-05,
+      "loss": 0.6953,
+      "step": 1603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884519221064872e-05,
+      "loss": 0.7836,
+      "step": 1604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843496940701695e-05,
+      "loss": 0.7601,
+      "step": 1605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841800503696754e-05,
+      "loss": 0.675,
+      "step": 1606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840102899857776e-05,
+      "loss": 0.6969,
+      "step": 1607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838404129408786e-05,
+      "loss": 0.8012,
+      "step": 1608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836704192573963e-05,
+      "loss": 0.6703,
+      "step": 1609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835003089577654e-05,
+      "loss": 0.7807,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833300820644345e-05,
+      "loss": 0.6851,
+      "step": 1611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883159738599868e-05,
+      "loss": 0.7301,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882989278586546e-05,
+      "loss": 0.7079,
+      "step": 1613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828187020469634e-05,
+      "loss": 0.7414,
+      "step": 1614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826480090036314e-05,
+      "loss": 0.6812,
+      "step": 1615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824771994790754e-05,
+      "loss": 0.8048,
+      "step": 1616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882306273495837e-05,
+      "loss": 0.7364,
+      "step": 1617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882135231076473e-05,
+      "loss": 0.7515,
+      "step": 1618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819640722435557e-05,
+      "loss": 0.7177,
+      "step": 1619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881792797019672e-05,
+      "loss": 0.8106,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816214054274256e-05,
+      "loss": 0.7597,
+      "step": 1621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814498974894338e-05,
+      "loss": 0.7123,
+      "step": 1622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812782732283308e-05,
+      "loss": 0.7044,
+      "step": 1623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811065326667653e-05,
+      "loss": 0.6869,
+      "step": 1624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809346758274014e-05,
+      "loss": 0.6939,
+      "step": 1625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807627027329186e-05,
+      "loss": 0.7519,
+      "step": 1626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880590613406012e-05,
+      "loss": 0.6942,
+      "step": 1627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804184078693923e-05,
+      "loss": 0.7218,
+      "step": 1628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802460861457846e-05,
+      "loss": 0.667,
+      "step": 1629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800736482579295e-05,
+      "loss": 0.6791,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879901094228584e-05,
+      "loss": 0.7777,
+      "step": 1631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797284240805192e-05,
+      "loss": 0.7484,
+      "step": 1632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795556378365226e-05,
+      "loss": 0.7263,
+      "step": 1633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879382735519396e-05,
+      "loss": 0.6316,
+      "step": 1634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879209717151957e-05,
+      "loss": 0.6896,
+      "step": 1635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790365827570383e-05,
+      "loss": 0.5804,
+      "step": 1636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788633323574882e-05,
+      "loss": 0.7153,
+      "step": 1637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786899659761706e-05,
+      "loss": 0.7819,
+      "step": 1638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785164836359637e-05,
+      "loss": 0.7312,
+      "step": 1639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878342885359762e-05,
+      "loss": 0.6862,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781691711704752e-05,
+      "loss": 0.5927,
+      "step": 1641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877995341091027e-05,
+      "loss": 0.7095,
+      "step": 1642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778213951443583e-05,
+      "loss": 0.673,
+      "step": 1643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776473333534238e-05,
+      "loss": 0.7135,
+      "step": 1644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774731557411945e-05,
+      "loss": 0.7394,
+      "step": 1645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877298862330656e-05,
+      "loss": 0.6455,
+      "step": 1646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771244531448096e-05,
+      "loss": 0.7901,
+      "step": 1647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769499282066716e-05,
+      "loss": 0.5674,
+      "step": 1648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767752875392737e-05,
+      "loss": 0.6953,
+      "step": 1649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876600531165663e-05,
+      "loss": 0.6823,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764256591089022e-05,
+      "loss": 0.7237,
+      "step": 1651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762506713920678e-05,
+      "loss": 0.7206,
+      "step": 1652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760755680382527e-05,
+      "loss": 0.7378,
+      "step": 1653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759003490705654e-05,
+      "loss": 0.7153,
+      "step": 1654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875725014512129e-05,
+      "loss": 0.7976,
+      "step": 1655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755495643860822e-05,
+      "loss": 0.7296,
+      "step": 1656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753739987155788e-05,
+      "loss": 0.7633,
+      "step": 1657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751983175237875e-05,
+      "loss": 0.6998,
+      "step": 1658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750225208338928e-05,
+      "loss": 0.6662,
+      "step": 1659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874846608669094e-05,
+      "loss": 0.6596,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746705810526063e-05,
+      "loss": 0.7009,
+      "step": 1661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744944380076597e-05,
+      "loss": 0.7867,
+      "step": 1662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743181795574986e-05,
+      "loss": 0.7323,
+      "step": 1663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741418057253843e-05,
+      "loss": 0.6776,
+      "step": 1664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739653165345925e-05,
+      "loss": 0.7425,
+      "step": 1665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737887120084135e-05,
+      "loss": 0.6991,
+      "step": 1666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736119921701538e-05,
+      "loss": 0.817,
+      "step": 1667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734351570431347e-05,
+      "loss": 1.115,
+      "step": 1668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873258206650693e-05,
+      "loss": 0.7499,
+      "step": 1669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730811410161804e-05,
+      "loss": 0.7026,
+      "step": 1670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729039601629637e-05,
+      "loss": 0.7065,
+      "step": 1671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872726664114425e-05,
+      "loss": 0.6493,
+      "step": 1672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725492528939618e-05,
+      "loss": 0.7068,
+      "step": 1673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723717265249866e-05,
+      "loss": 0.6881,
+      "step": 1674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721940850309278e-05,
+      "loss": 0.7103,
+      "step": 1675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720163284352274e-05,
+      "loss": 0.8314,
+      "step": 1676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871838456761344e-05,
+      "loss": 0.8053,
+      "step": 1677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716604700327516e-05,
+      "loss": 0.7902,
+      "step": 1678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714823682729374e-05,
+      "loss": 0.7136,
+      "step": 1679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713041515054065e-05,
+      "loss": 0.7658,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711258197536767e-05,
+      "loss": 0.7786,
+      "step": 1681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870947373041283e-05,
+      "loss": 0.816,
+      "step": 1682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870768811391774e-05,
+      "loss": 0.7482,
+      "step": 1683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870590134828714e-05,
+      "loss": 0.6702,
+      "step": 1684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704113433756833e-05,
+      "loss": 0.812,
+      "step": 1685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870232437056276e-05,
+      "loss": 0.6549,
+      "step": 1686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870053415894102e-05,
+      "loss": 0.6758,
+      "step": 1687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869874279912787e-05,
+      "loss": 0.7836,
+      "step": 1688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696950291359704e-05,
+      "loss": 0.7637,
+      "step": 1689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869515663587308e-05,
+      "loss": 0.7436,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86933618329047e-05,
+      "loss": 0.7235,
+      "step": 1691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869156588269143e-05,
+      "loss": 0.7414,
+      "step": 1692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689768785470265e-05,
+      "loss": 0.7997,
+      "step": 1693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687970541478367e-05,
+      "loss": 0.7408,
+      "step": 1694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868617115095305e-05,
+      "loss": 0.6922,
+      "step": 1695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684370614131777e-05,
+      "loss": 0.6946,
+      "step": 1696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682568931252158e-05,
+      "loss": 0.5995,
+      "step": 1697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680766102551955e-05,
+      "loss": 0.7456,
+      "step": 1698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678962128269094e-05,
+      "loss": 0.7495,
+      "step": 1699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867715700864163e-05,
+      "loss": 0.699,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675350743907788e-05,
+      "loss": 0.7174,
+      "step": 1701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867354333430593e-05,
+      "loss": 0.8328,
+      "step": 1702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671734780074585e-05,
+      "loss": 0.7145,
+      "step": 1703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669925081452417e-05,
+      "loss": 0.7459,
+      "step": 1704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866811423867825e-05,
+      "loss": 0.7905,
+      "step": 1705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866630225199106e-05,
+      "loss": 0.7824,
+      "step": 1706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866448912162997e-05,
+      "loss": 0.7641,
+      "step": 1707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662674847834253e-05,
+      "loss": 0.6819,
+      "step": 1708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660859430843333e-05,
+      "loss": 0.7032,
+      "step": 1709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659042870896793e-05,
+      "loss": 0.748,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865722516823436e-05,
+      "loss": 0.7404,
+      "step": 1711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655406323095904e-05,
+      "loss": 0.6428,
+      "step": 1712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653586335721467e-05,
+      "loss": 0.7446,
+      "step": 1713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651765206351216e-05,
+      "loss": 0.7261,
+      "step": 1714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864994293522549e-05,
+      "loss": 0.6731,
+      "step": 1715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864811952258477e-05,
+      "loss": 0.7199,
+      "step": 1716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646294968669685e-05,
+      "loss": 0.7382,
+      "step": 1717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864446927372102e-05,
+      "loss": 0.676,
+      "step": 1718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642642437979702e-05,
+      "loss": 0.6796,
+      "step": 1719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640814461686825e-05,
+      "loss": 0.7691,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638985345083616e-05,
+      "loss": 0.7763,
+      "step": 1721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863715508841146e-05,
+      "loss": 0.6338,
+      "step": 1722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635323691911894e-05,
+      "loss": 0.7742,
+      "step": 1723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633491155826607e-05,
+      "loss": 0.7283,
+      "step": 1724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863165748039743e-05,
+      "loss": 0.6794,
+      "step": 1725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629822665866348e-05,
+      "loss": 0.7555,
+      "step": 1726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627986712475503e-05,
+      "loss": 0.7064,
+      "step": 1727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626149620467177e-05,
+      "loss": 0.7889,
+      "step": 1728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862431139008381e-05,
+      "loss": 0.6791,
+      "step": 1729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862247202156799e-05,
+      "loss": 0.7407,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620631515162454e-05,
+      "loss": 0.7816,
+      "step": 1731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618789871110084e-05,
+      "loss": 0.6965,
+      "step": 1732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861694708965393e-05,
+      "loss": 0.7113,
+      "step": 1733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861510317103717e-05,
+      "loss": 0.6268,
+      "step": 1734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861325811550315e-05,
+      "loss": 0.663,
+      "step": 1735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611411923295345e-05,
+      "loss": 0.6927,
+      "step": 1736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860956459465741e-05,
+      "loss": 0.6257,
+      "step": 1737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860771612983312e-05,
+      "loss": 0.7526,
+      "step": 1738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860586652906642e-05,
+      "loss": 0.6539,
+      "step": 1739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604015792601395e-05,
+      "loss": 0.6979,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602163920682284e-05,
+      "loss": 0.6814,
+      "step": 1741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600310913553476e-05,
+      "loss": 0.7544,
+      "step": 1742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598456771459507e-05,
+      "loss": 0.731,
+      "step": 1743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859660149464506e-05,
+      "loss": 0.7372,
+      "step": 1744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859474508335498e-05,
+      "loss": 0.7055,
+      "step": 1745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592887537834246e-05,
+      "loss": 0.7198,
+      "step": 1746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591028858328e-05,
+      "loss": 0.6637,
+      "step": 1747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589169045081533e-05,
+      "loss": 0.6726,
+      "step": 1748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587308098340267e-05,
+      "loss": 0.7532,
+      "step": 1749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585446018349796e-05,
+      "loss": 0.6426,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583582805355853e-05,
+      "loss": 0.7267,
+      "step": 1751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581718459604324e-05,
+      "loss": 0.6087,
+      "step": 1752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857985298134124e-05,
+      "loss": 0.767,
+      "step": 1753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577986370812785e-05,
+      "loss": 0.5931,
+      "step": 1754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576118628265295e-05,
+      "loss": 0.727,
+      "step": 1755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857424975394525e-05,
+      "loss": 0.7089,
+      "step": 1756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572379748099277e-05,
+      "loss": 0.7477,
+      "step": 1757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857050861097416e-05,
+      "loss": 0.7118,
+      "step": 1758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568636342816838e-05,
+      "loss": 0.7023,
+      "step": 1759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566762943874376e-05,
+      "loss": 0.6631,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856488841439401e-05,
+      "loss": 0.7297,
+      "step": 1761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563012754623114e-05,
+      "loss": 0.7498,
+      "step": 1762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561135964809223e-05,
+      "loss": 0.6844,
+      "step": 1763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559258045200004e-05,
+      "loss": 0.767,
+      "step": 1764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557378996043284e-05,
+      "loss": 0.8077,
+      "step": 1765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555498817587043e-05,
+      "loss": 0.6693,
+      "step": 1766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553617510079397e-05,
+      "loss": 0.7884,
+      "step": 1767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855173507376862e-05,
+      "loss": 0.676,
+      "step": 1768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854985150890314e-05,
+      "loss": 0.7472,
+      "step": 1769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854796681573152e-05,
+      "loss": 0.7784,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854608099450248e-05,
+      "loss": 0.7512,
+      "step": 1771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544194045464888e-05,
+      "loss": 0.7236,
+      "step": 1772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542305968867758e-05,
+      "loss": 0.6949,
+      "step": 1773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540416764960266e-05,
+      "loss": 0.7062,
+      "step": 1774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538526433991716e-05,
+      "loss": 0.6028,
+      "step": 1775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536634976211575e-05,
+      "loss": 0.7507,
+      "step": 1776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853474239186945e-05,
+      "loss": 0.7282,
+      "step": 1777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853284868121511e-05,
+      "loss": 0.6882,
+      "step": 1778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853095384449846e-05,
+      "loss": 0.7639,
+      "step": 1779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529057881969554e-05,
+      "loss": 0.7551,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527160793878603e-05,
+      "loss": 0.7164,
+      "step": 1781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852526258047596e-05,
+      "loss": 0.6414,
+      "step": 1782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523363242012128e-05,
+      "loss": 0.7989,
+      "step": 1783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852146277873776e-05,
+      "loss": 0.734,
+      "step": 1784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519561190903657e-05,
+      "loss": 0.7153,
+      "step": 1785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851765847876076e-05,
+      "loss": 0.6533,
+      "step": 1786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851575464256018e-05,
+      "loss": 0.8028,
+      "step": 1787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851384968255315e-05,
+      "loss": 0.777,
+      "step": 1788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851194359899107e-05,
+      "loss": 0.6385,
+      "step": 1789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510036392125476e-05,
+      "loss": 0.659,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508128062208065e-05,
+      "loss": 0.7914,
+      "step": 1791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850621860949067e-05,
+      "loss": 0.6913,
+      "step": 1792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504308034225285e-05,
+      "loss": 0.6894,
+      "step": 1793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502396336664036e-05,
+      "loss": 0.7266,
+      "step": 1794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500483517059208e-05,
+      "loss": 0.8086,
+      "step": 1795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498569575663233e-05,
+      "loss": 0.7216,
+      "step": 1796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496654512728694e-05,
+      "loss": 0.7904,
+      "step": 1797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849473832850831e-05,
+      "loss": 0.7798,
+      "step": 1798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849282102325496e-05,
+      "loss": 0.7675,
+      "step": 1799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490902597221665e-05,
+      "loss": 0.7282,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84889830506616e-05,
+      "loss": 0.7238,
+      "step": 1801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487062383828078e-05,
+      "loss": 0.6727,
+      "step": 1802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848514059697457e-05,
+      "loss": 0.6053,
+      "step": 1803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483217690354687e-05,
+      "loss": 0.686,
+      "step": 1804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481293664222197e-05,
+      "loss": 0.8487,
+      "step": 1805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479368518831e-05,
+      "loss": 0.723,
+      "step": 1806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477442254435157e-05,
+      "loss": 0.7882,
+      "step": 1807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847551487128888e-05,
+      "loss": 0.7715,
+      "step": 1808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473586369646514e-05,
+      "loss": 0.7667,
+      "step": 1809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471656749762558e-05,
+      "loss": 0.791,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846972601189167e-05,
+      "loss": 0.6479,
+      "step": 1811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467794156288634e-05,
+      "loss": 0.7136,
+      "step": 1812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84658611832084e-05,
+      "loss": 0.6034,
+      "step": 1813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846392709290606e-05,
+      "loss": 0.7179,
+      "step": 1814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461991885636842e-05,
+      "loss": 0.7502,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846005556165614e-05,
+      "loss": 0.7936,
+      "step": 1816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458118121219483e-05,
+      "loss": 0.7845,
+      "step": 1817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456179564582557e-05,
+      "loss": 0.6505,
+      "step": 1818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845423989200118e-05,
+      "loss": 0.6769,
+      "step": 1819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845229910373133e-05,
+      "loss": 0.8207,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450357200029136e-05,
+      "loss": 0.6885,
+      "step": 1821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448414181150856e-05,
+      "loss": 0.7767,
+      "step": 1822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446470047352914e-05,
+      "loss": 0.7684,
+      "step": 1823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844452479889187e-05,
+      "loss": 0.6677,
+      "step": 1824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844257843602443e-05,
+      "loss": 0.7812,
+      "step": 1825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844063095900746e-05,
+      "loss": 0.735,
+      "step": 1826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843868236809796e-05,
+      "loss": 0.7872,
+      "step": 1827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436732663553085e-05,
+      "loss": 0.7525,
+      "step": 1828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843478184563013e-05,
+      "loss": 0.706,
+      "step": 1829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843282991458654e-05,
+      "loss": 0.7524,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843087687067991e-05,
+      "loss": 0.7791,
+      "step": 1831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842892271416797e-05,
+      "loss": 0.6898,
+      "step": 1832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842696744530862e-05,
+      "loss": 0.7408,
+      "step": 1833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425011064359885e-05,
+      "loss": 0.7492,
+      "step": 1834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423053571579948e-05,
+      "loss": 0.7733,
+      "step": 1835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842109496722713e-05,
+      "loss": 0.6635,
+      "step": 1836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841913525155991e-05,
+      "loss": 0.7477,
+      "step": 1837
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417174424836902e-05,
+      "loss": 0.7169,
+      "step": 1838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415212487316875e-05,
+      "loss": 0.7778,
+      "step": 1839
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413249439258743e-05,
+      "loss": 0.7152,
+      "step": 1840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411285280921564e-05,
+      "loss": 0.6755,
+      "step": 1841
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409320012564543e-05,
+      "loss": 0.707,
+      "step": 1842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840735363444703e-05,
+      "loss": 0.7374,
+      "step": 1843
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405386146828533e-05,
+      "loss": 0.6957,
+      "step": 1844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403417549968688e-05,
+      "loss": 0.8179,
+      "step": 1845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840144784412729e-05,
+      "loss": 0.7326,
+      "step": 1846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399477029564278e-05,
+      "loss": 0.7312,
+      "step": 1847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397505106539735e-05,
+      "loss": 0.701,
+      "step": 1848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839553207531389e-05,
+      "loss": 0.6046,
+      "step": 1849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393557936147127e-05,
+      "loss": 0.6595,
+      "step": 1850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391582689299962e-05,
+      "loss": 0.7039,
+      "step": 1851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838960633503306e-05,
+      "loss": 0.763,
+      "step": 1852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387628873607247e-05,
+      "loss": 0.6476,
+      "step": 1853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838565030528348e-05,
+      "loss": 0.6617,
+      "step": 1854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383670630322864e-05,
+      "loss": 0.8055,
+      "step": 1855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838168984898666e-05,
+      "loss": 0.8004,
+      "step": 1856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379707961536256e-05,
+      "loss": 0.6154,
+      "step": 1857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377724968233212e-05,
+      "loss": 0.7806,
+      "step": 1858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375740869339205e-05,
+      "loss": 0.7163,
+      "step": 1859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837375566511608e-05,
+      "loss": 0.694,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837176935582582e-05,
+      "loss": 0.6731,
+      "step": 1861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369781941730556e-05,
+      "loss": 0.6817,
+      "step": 1862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367793423092555e-05,
+      "loss": 0.7636,
+      "step": 1863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836580380017425e-05,
+      "loss": 0.7127,
+      "step": 1864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363813073238196e-05,
+      "loss": 0.698,
+      "step": 1865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836182124254711e-05,
+      "loss": 0.7414,
+      "step": 1866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835982830836385e-05,
+      "loss": 0.7579,
+      "step": 1867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835783427095142e-05,
+      "loss": 0.8173,
+      "step": 1868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355839130572963e-05,
+      "loss": 0.7173,
+      "step": 1869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353842887491783e-05,
+      "loss": 0.8364,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351845541971313e-05,
+      "loss": 0.7079,
+      "step": 1871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834984709427514e-05,
+      "loss": 0.7948,
+      "step": 1872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347847544666996e-05,
+      "loss": 0.7368,
+      "step": 1873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345846893410758e-05,
+      "loss": 0.7104,
+      "step": 1874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343845140770447e-05,
+      "loss": 0.6393,
+      "step": 1875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834184228701023e-05,
+      "loss": 0.7326,
+      "step": 1876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339838332394418e-05,
+      "loss": 0.7457,
+      "step": 1877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337833277187472e-05,
+      "loss": 0.7604,
+      "step": 1878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833582712165399e-05,
+      "loss": 0.7383,
+      "step": 1879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833381986605873e-05,
+      "loss": 0.6165,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331811510666572e-05,
+      "loss": 0.741,
+      "step": 1881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329802055742565e-05,
+      "loss": 0.6581,
+      "step": 1882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327791501551887e-05,
+      "loss": 0.838,
+      "step": 1883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325779848359868e-05,
+      "loss": 0.717,
+      "step": 1884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323767096431986e-05,
+      "loss": 0.6839,
+      "step": 1885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832175324603385e-05,
+      "loss": 0.6885,
+      "step": 1886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319738297431236e-05,
+      "loss": 0.6557,
+      "step": 1887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317722250890044e-05,
+      "loss": 0.6967,
+      "step": 1888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315705106676324e-05,
+      "loss": 0.6716,
+      "step": 1889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831368686505629e-05,
+      "loss": 0.7037,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311667526296267e-05,
+      "loss": 0.7816,
+      "step": 1891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830964709066275e-05,
+      "loss": 0.7536,
+      "step": 1892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307625558422377e-05,
+      "loss": 0.7616,
+      "step": 1893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305602929841922e-05,
+      "loss": 0.7192,
+      "step": 1894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303579205188305e-05,
+      "loss": 0.7103,
+      "step": 1895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301554384728594e-05,
+      "loss": 0.8155,
+      "step": 1896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299528468729998e-05,
+      "loss": 0.8039,
+      "step": 1897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297501457459877e-05,
+      "loss": 0.6822,
+      "step": 1898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829547335118573e-05,
+      "loss": 0.7735,
+      "step": 1899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293444150175198e-05,
+      "loss": 0.7209,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829141385469608e-05,
+      "loss": 0.7498,
+      "step": 1901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82893824650163e-05,
+      "loss": 0.7086,
+      "step": 1902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828734998140394e-05,
+      "loss": 0.8101,
+      "step": 1903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285316404127225e-05,
+      "loss": 0.7249,
+      "step": 1904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828328173345452e-05,
+      "loss": 0.7417,
+      "step": 1905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281245969654333e-05,
+      "loss": 0.7433,
+      "step": 1906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279209112995324e-05,
+      "loss": 0.7968,
+      "step": 1907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827717116374629e-05,
+      "loss": 0.7112,
+      "step": 1908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827513212217618e-05,
+      "loss": 0.7201,
+      "step": 1909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273091988554075e-05,
+      "loss": 0.7656,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271050763149213e-05,
+      "loss": 0.7683,
+      "step": 1911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269008446230966e-05,
+      "loss": 0.7339,
+      "step": 1912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266965038068856e-05,
+      "loss": 0.6891,
+      "step": 1913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264920538932553e-05,
+      "loss": 0.7205,
+      "step": 1914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262874949091852e-05,
+      "loss": 0.7417,
+      "step": 1915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260828268816717e-05,
+      "loss": 0.6956,
+      "step": 1916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258780498377243e-05,
+      "loss": 0.7127,
+      "step": 1917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825673163804366e-05,
+      "loss": 0.685,
+      "step": 1918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254681688086366e-05,
+      "loss": 0.8246,
+      "step": 1919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252630648775874e-05,
+      "loss": 0.751,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250578520382868e-05,
+      "loss": 0.7796,
+      "step": 1921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824852530317816e-05,
+      "loss": 0.6135,
+      "step": 1922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246470997432704e-05,
+      "loss": 0.7139,
+      "step": 1923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244415603417603e-05,
+      "loss": 0.6475,
+      "step": 1924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242359121404112e-05,
+      "loss": 0.7516,
+      "step": 1925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824030155166361e-05,
+      "loss": 0.7524,
+      "step": 1926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238242894467632e-05,
+      "loss": 0.8518,
+      "step": 1927
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236183150087862e-05,
+      "loss": 0.6186,
+      "step": 1928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234122318796112e-05,
+      "loss": 0.7537,
+      "step": 1929
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823206040086435e-05,
+      "loss": 0.644,
+      "step": 1930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229997396564683e-05,
+      "loss": 0.6075,
+      "step": 1931
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822793330616936e-05,
+      "loss": 0.705,
+      "step": 1932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225868129950775e-05,
+      "loss": 0.7055,
+      "step": 1933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223801868181462e-05,
+      "loss": 0.6741,
+      "step": 1934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221734521134105e-05,
+      "loss": 0.7388,
+      "step": 1935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821966608908153e-05,
+      "loss": 0.6818,
+      "step": 1936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217596572296702e-05,
+      "loss": 0.7466,
+      "step": 1937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215525971052727e-05,
+      "loss": 0.6965,
+      "step": 1938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821345428562286e-05,
+      "loss": 0.6829,
+      "step": 1939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211381516280494e-05,
+      "loss": 0.664,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209307663299176e-05,
+      "loss": 0.6895,
+      "step": 1941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820723272695258e-05,
+      "loss": 0.7039,
+      "step": 1942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205156707514535e-05,
+      "loss": 0.785,
+      "step": 1943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820307960525901e-05,
+      "loss": 0.6485,
+      "step": 1944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201001420460115e-05,
+      "loss": 0.7331,
+      "step": 1945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81989221533921e-05,
+      "loss": 0.7912,
+      "step": 1946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819684180432937e-05,
+      "loss": 0.7246,
+      "step": 1947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819476037354645e-05,
+      "loss": 0.7158,
+      "step": 1948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819267786131804e-05,
+      "loss": 0.7048,
+      "step": 1949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819059426791895e-05,
+      "loss": 0.7194,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188509593624157e-05,
+      "loss": 0.6383,
+      "step": 1951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186423838708767e-05,
+      "loss": 0.8147,
+      "step": 1952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184337003448033e-05,
+      "loss": 0.675,
+      "step": 1953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182249088117347e-05,
+      "loss": 0.7226,
+      "step": 1954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180160092992254e-05,
+      "loss": 0.7308,
+      "step": 1955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178070018348427e-05,
+      "loss": 0.5928,
+      "step": 1956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175978864461697e-05,
+      "loss": 0.772,
+      "step": 1957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173886631608024e-05,
+      "loss": 0.7644,
+      "step": 1958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171793320063517e-05,
+      "loss": 0.6954,
+      "step": 1959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816969893010442e-05,
+      "loss": 0.733,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167603462007135e-05,
+      "loss": 0.7292,
+      "step": 1961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165506916048193e-05,
+      "loss": 0.7026,
+      "step": 1962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816340929250427e-05,
+      "loss": 0.753,
+      "step": 1963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816131059165218e-05,
+      "loss": 0.8326,
+      "step": 1964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159210813768894e-05,
+      "loss": 0.7136,
+      "step": 1965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815710995913151e-05,
+      "loss": 0.6783,
+      "step": 1966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155008028017275e-05,
+      "loss": 0.7085,
+      "step": 1967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152905020703573e-05,
+      "loss": 0.7956,
+      "step": 1968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150800937467937e-05,
+      "loss": 0.7142,
+      "step": 1969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148695778588034e-05,
+      "loss": 0.7073,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146589544341685e-05,
+      "loss": 0.661,
+      "step": 1971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814448223500684e-05,
+      "loss": 0.7589,
+      "step": 1972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142373850861598e-05,
+      "loss": 0.7683,
+      "step": 1973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140264392184198e-05,
+      "loss": 0.6981,
+      "step": 1974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813815385925302e-05,
+      "loss": 0.7744,
+      "step": 1975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136042252346587e-05,
+      "loss": 0.7605,
+      "step": 1976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133929571743565e-05,
+      "loss": 0.5398,
+      "step": 1977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131815817722757e-05,
+      "loss": 0.6932,
+      "step": 1978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129700990563114e-05,
+      "loss": 0.7177,
+      "step": 1979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127585090543717e-05,
+      "loss": 0.7356,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812546811794381e-05,
+      "loss": 0.7254,
+      "step": 1981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123350073042758e-05,
+      "loss": 0.6457,
+      "step": 1982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121230956120075e-05,
+      "loss": 0.6836,
+      "step": 1983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811911076745542e-05,
+      "loss": 0.7301,
+      "step": 1984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116989507328584e-05,
+      "loss": 0.6951,
+      "step": 1985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811486717601951e-05,
+      "loss": 0.7686,
+      "step": 1986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112743773808275e-05,
+      "loss": 0.714,
+      "step": 1987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110619300975103e-05,
+      "loss": 0.748,
+      "step": 1988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108493757800353e-05,
+      "loss": 0.6761,
+      "step": 1989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106367144564532e-05,
+      "loss": 0.6482,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104239461548282e-05,
+      "loss": 0.6969,
+      "step": 1991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810211070903239e-05,
+      "loss": 0.66,
+      "step": 1992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809998088729778e-05,
+      "loss": 0.6571,
+      "step": 1993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097849996625524e-05,
+      "loss": 0.7274,
+      "step": 1994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095718037296835e-05,
+      "loss": 0.6951,
+      "step": 1995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093585009593053e-05,
+      "loss": 0.775,
+      "step": 1996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091450913795675e-05,
+      "loss": 0.6598,
+      "step": 1997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089315750186335e-05,
+      "loss": 0.6471,
+      "step": 1998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087179519046802e-05,
+      "loss": 0.8263,
+      "step": 1999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085042220658993e-05,
+      "loss": 0.6727,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808290385530496e-05,
+      "loss": 0.731,
+      "step": 2001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80807644232669e-05,
+      "loss": 0.736,
+      "step": 2002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078623924827153e-05,
+      "loss": 0.7957,
+      "step": 2003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807648236026819e-05,
+      "loss": 0.7004,
+      "step": 2004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074339729872635e-05,
+      "loss": 0.8048,
+      "step": 2005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072196033923238e-05,
+      "loss": 0.7088,
+      "step": 2006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070051272702905e-05,
+      "loss": 0.7215,
+      "step": 2007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806790544649468e-05,
+      "loss": 0.5993,
+      "step": 2008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065758555581728e-05,
+      "loss": 0.6923,
+      "step": 2009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063610600247386e-05,
+      "loss": 0.7532,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061461580775103e-05,
+      "loss": 0.6769,
+      "step": 2011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805931149744849e-05,
+      "loss": 0.6928,
+      "step": 2012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805716035055128e-05,
+      "loss": 0.6822,
+      "step": 2013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055008140367364e-05,
+      "loss": 0.7321,
+      "step": 2014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805285486718076e-05,
+      "loss": 0.7001,
+      "step": 2015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050700531275632e-05,
+      "loss": 0.6985,
+      "step": 2016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048545132936283e-05,
+      "loss": 0.6781,
+      "step": 2017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046388672447156e-05,
+      "loss": 0.7643,
+      "step": 2018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044231150092836e-05,
+      "loss": 0.6507,
+      "step": 2019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042072566158043e-05,
+      "loss": 0.7317,
+      "step": 2020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039912920927644e-05,
+      "loss": 0.7629,
+      "step": 2021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037752214686647e-05,
+      "loss": 0.7984,
+      "step": 2022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035590447720186e-05,
+      "loss": 0.7452,
+      "step": 2023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033427620313554e-05,
+      "loss": 0.7459,
+      "step": 2024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803126373275217e-05,
+      "loss": 0.7332,
+      "step": 2025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80290987853216e-05,
+      "loss": 0.7373,
+      "step": 2026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026932778307546e-05,
+      "loss": 0.744,
+      "step": 2027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802476571199585e-05,
+      "loss": 0.7246,
+      "step": 2028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80225975866725e-05,
+      "loss": 0.6983,
+      "step": 2029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020428402623618e-05,
+      "loss": 0.718,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018258160135463e-05,
+      "loss": 0.6489,
+      "step": 2031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801608685949444e-05,
+      "loss": 0.8595,
+      "step": 2032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013914500987085e-05,
+      "loss": 0.662,
+      "step": 2033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011741084900092e-05,
+      "loss": 0.6968,
+      "step": 2034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009566611520273e-05,
+      "loss": 0.7415,
+      "step": 2035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007391081134587e-05,
+      "loss": 0.8271,
+      "step": 2036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005214494030144e-05,
+      "loss": 0.7231,
+      "step": 2037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003036850494177e-05,
+      "loss": 0.7346,
+      "step": 2038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800085815081406e-05,
+      "loss": 0.6477,
+      "step": 2039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998678395277328e-05,
+      "loss": 0.7386,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996497584171618e-05,
+      "loss": 0.8208,
+      "step": 2041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799431571778474e-05,
+      "loss": 0.6956,
+      "step": 2042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992132796404634e-05,
+      "loss": 0.6685,
+      "step": 2043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989948820319364e-05,
+      "loss": 0.7335,
+      "step": 2044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987763789817152e-05,
+      "loss": 0.7032,
+      "step": 2045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985577705186347e-05,
+      "loss": 0.6123,
+      "step": 2046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983390566715448e-05,
+      "loss": 0.7239,
+      "step": 2047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981202374693083e-05,
+      "loss": 0.6241,
+      "step": 2048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979013129408024e-05,
+      "loss": 0.6922,
+      "step": 2049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976822831149183e-05,
+      "loss": 0.7471,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974631480205608e-05,
+      "loss": 0.777,
+      "step": 2051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972439076866486e-05,
+      "loss": 0.6617,
+      "step": 2052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970245621421144e-05,
+      "loss": 0.7308,
+      "step": 2053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968051114159046e-05,
+      "loss": 0.8706,
+      "step": 2054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965855555369804e-05,
+      "loss": 0.6934,
+      "step": 2055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963658945343153e-05,
+      "loss": 0.7395,
+      "step": 2056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961461284368978e-05,
+      "loss": 0.6865,
+      "step": 2057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79592625727373e-05,
+      "loss": 0.6988,
+      "step": 2058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795706281073828e-05,
+      "loss": 0.7503,
+      "step": 2059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954861998662216e-05,
+      "loss": 0.7371,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795266013679954e-05,
+      "loss": 0.826,
+      "step": 2061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795045722544083e-05,
+      "loss": 0.7729,
+      "step": 2062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79482532648768e-05,
+      "loss": 0.6923,
+      "step": 2063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946048255398305e-05,
+      "loss": 0.6623,
+      "step": 2064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943842197296332e-05,
+      "loss": 0.6783,
+      "step": 2065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941635090862007e-05,
+      "loss": 0.8185,
+      "step": 2066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939426936386603e-05,
+      "loss": 0.7237,
+      "step": 2067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937217734161523e-05,
+      "loss": 0.7032,
+      "step": 2068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935007484478312e-05,
+      "loss": 0.7253,
+      "step": 2069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932796187628648e-05,
+      "loss": 0.7224,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930583843904357e-05,
+      "loss": 0.7336,
+      "step": 2071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928370453597392e-05,
+      "loss": 0.6482,
+      "step": 2072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926156016999857e-05,
+      "loss": 0.8077,
+      "step": 2073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923940534403978e-05,
+      "loss": 0.6967,
+      "step": 2074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792172400610213e-05,
+      "loss": 0.6944,
+      "step": 2075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791950643238683e-05,
+      "loss": 0.736,
+      "step": 2076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917287813550717e-05,
+      "loss": 0.7018,
+      "step": 2077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915068149886586e-05,
+      "loss": 0.8043,
+      "step": 2078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912847441687354e-05,
+      "loss": 0.764,
+      "step": 2079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791062568924609e-05,
+      "loss": 0.6362,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908402892855988e-05,
+      "loss": 0.7053,
+      "step": 2081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906179052810394e-05,
+      "loss": 0.7828,
+      "step": 2082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903954169402772e-05,
+      "loss": 0.7326,
+      "step": 2083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901728242926747e-05,
+      "loss": 0.8058,
+      "step": 2084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789950127367606e-05,
+      "loss": 0.6764,
+      "step": 2085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897273261944605e-05,
+      "loss": 0.6631,
+      "step": 2086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895044208026407e-05,
+      "loss": 0.6092,
+      "step": 2087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789281411221563e-05,
+      "loss": 0.602,
+      "step": 2088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890582974806575e-05,
+      "loss": 0.68,
+      "step": 2089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888350796093683e-05,
+      "loss": 0.6852,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886117576371522e-05,
+      "loss": 0.6068,
+      "step": 2091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883883315934816e-05,
+      "loss": 0.7437,
+      "step": 2092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881648015078406e-05,
+      "loss": 0.8553,
+      "step": 2093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879411674097284e-05,
+      "loss": 0.688,
+      "step": 2094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787717429328658e-05,
+      "loss": 0.6667,
+      "step": 2095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874935872941547e-05,
+      "loss": 0.7732,
+      "step": 2096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872696413357596e-05,
+      "loss": 0.652,
+      "step": 2097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870455914830252e-05,
+      "loss": 0.7221,
+      "step": 2098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868214377655197e-05,
+      "loss": 0.7748,
+      "step": 2099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786597180212824e-05,
+      "loss": 0.7614,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863728188545326e-05,
+      "loss": 0.6908,
+      "step": 2101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861483537202544e-05,
+      "loss": 0.7616,
+      "step": 2102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859237848396116e-05,
+      "loss": 0.7886,
+      "step": 2103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856991122422398e-05,
+      "loss": 0.7694,
+      "step": 2104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854743359577888e-05,
+      "loss": 0.7251,
+      "step": 2105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852494560159218e-05,
+      "loss": 0.6777,
+      "step": 2106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850244724463157e-05,
+      "loss": 0.6874,
+      "step": 2107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847993852786612e-05,
+      "loss": 0.7548,
+      "step": 2108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845741945426622e-05,
+      "loss": 0.8206,
+      "step": 2109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843489002680374e-05,
+      "loss": 0.7127,
+      "step": 2110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841235024845177e-05,
+      "loss": 0.7432,
+      "step": 2111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7838980012218486e-05,
+      "loss": 0.7989,
+      "step": 2112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836723965097896e-05,
+      "loss": 0.7252,
+      "step": 2113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834466883781127e-05,
+      "loss": 0.7198,
+      "step": 2114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832208768566038e-05,
+      "loss": 0.7045,
+      "step": 2115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829949619750634e-05,
+      "loss": 0.6689,
+      "step": 2116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782768943763305e-05,
+      "loss": 0.797,
+      "step": 2117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782542822251155e-05,
+      "loss": 0.6601,
+      "step": 2118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823165974684556e-05,
+      "loss": 0.6803,
+      "step": 2119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820902694450597e-05,
+      "loss": 0.6674,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818638382108363e-05,
+      "loss": 0.6379,
+      "step": 2121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816373037956665e-05,
+      "loss": 0.7154,
+      "step": 2122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781410666229446e-05,
+      "loss": 0.7644,
+      "step": 2123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811839255420838e-05,
+      "loss": 0.7678,
+      "step": 2124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809570817635016e-05,
+      "loss": 0.7251,
+      "step": 2125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807301349236364e-05,
+      "loss": 0.5965,
+      "step": 2126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805030850524375e-05,
+      "loss": 0.7426,
+      "step": 2127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802759321798685e-05,
+      "loss": 0.7038,
+      "step": 2128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800486763359057e-05,
+      "loss": 0.7065,
+      "step": 2129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77982131755054e-05,
+      "loss": 0.6916,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795938558537755e-05,
+      "loss": 0.6524,
+      "step": 2131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793662912756295e-05,
+      "loss": 0.6967,
+      "step": 2132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791386238461336e-05,
+      "loss": 0.6929,
+      "step": 2133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789108535953322e-05,
+      "loss": 0.697,
+      "step": 2134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786829805532842e-05,
+      "loss": 0.8284,
+      "step": 2135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778455004750061e-05,
+      "loss": 0.7187,
+      "step": 2136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782269262157483e-05,
+      "loss": 0.7294,
+      "step": 2137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779987449804457e-05,
+      "loss": 0.818,
+      "step": 2138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777770461074265e-05,
+      "loss": 0.7469,
+      "step": 2139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775420745273323e-05,
+      "loss": 0.723,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777313585369788e-05,
+      "loss": 0.7072,
+      "step": 2141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777084993631785e-05,
+      "loss": 0.6661,
+      "step": 2142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77685629934349e-05,
+      "loss": 0.8254,
+      "step": 2143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766275025350833e-05,
+      "loss": 0.7388,
+      "step": 2144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776398603236759e-05,
+      "loss": 0.6768,
+      "step": 2145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776169601478724e-05,
+      "loss": 0.6161,
+      "step": 2146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759404972911997e-05,
+      "loss": 0.6711,
+      "step": 2147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77571129070442e-05,
+      "loss": 0.6428,
+      "step": 2148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754819817486334e-05,
+      "loss": 0.7038,
+      "step": 2149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752525704541007e-05,
+      "loss": 0.7249,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750230568510974e-05,
+      "loss": 0.8119,
+      "step": 2151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747934409699113e-05,
+      "loss": 0.65,
+      "step": 2152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774563722840845e-05,
+      "loss": 0.6911,
+      "step": 2153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743339024942135e-05,
+      "loss": 0.6741,
+      "step": 2154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774103979960346e-05,
+      "loss": 0.7515,
+      "step": 2155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738739552695846e-05,
+      "loss": 0.6785,
+      "step": 2156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736438284522855e-05,
+      "loss": 0.8126,
+      "step": 2157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773413599538818e-05,
+      "loss": 0.7788,
+      "step": 2158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731832685595643e-05,
+      "loss": 0.7321,
+      "step": 2159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729528355449214e-05,
+      "loss": 0.6761,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727223005252988e-05,
+      "loss": 0.6006,
+      "step": 2161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77249166353112e-05,
+      "loss": 0.6637,
+      "step": 2162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722609245928217e-05,
+      "loss": 0.9953,
+      "step": 2163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720300837408536e-05,
+      "loss": 0.8366,
+      "step": 2164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717991410056796e-05,
+      "loss": 0.7008,
+      "step": 2165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715680964177762e-05,
+      "loss": 0.8318,
+      "step": 2166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713369500076347e-05,
+      "loss": 0.6964,
+      "step": 2167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711057018057587e-05,
+      "loss": 0.8292,
+      "step": 2168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708743518426654e-05,
+      "loss": 0.7857,
+      "step": 2169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706429001488857e-05,
+      "loss": 1.067,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704113467549635e-05,
+      "loss": 0.6993,
+      "step": 2171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701796916914568e-05,
+      "loss": 0.8455,
+      "step": 2172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699479349889366e-05,
+      "loss": 0.675,
+      "step": 2173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769716076677987e-05,
+      "loss": 0.755,
+      "step": 2174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769484116789207e-05,
+      "loss": 0.7246,
+      "step": 2175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769252055353206e-05,
+      "loss": 0.6709,
+      "step": 2176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690198924006097e-05,
+      "loss": 0.7305,
+      "step": 2177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687876279620563e-05,
+      "loss": 0.6811,
+      "step": 2178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768555262068197e-05,
+      "loss": 0.7252,
+      "step": 2179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768322794749697e-05,
+      "loss": 0.7108,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7680902260372342e-05,
+      "loss": 0.7408,
+      "step": 2181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678575559615002e-05,
+      "loss": 0.7191,
+      "step": 2182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676247845532e-05,
+      "loss": 0.7363,
+      "step": 2183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767391911843052e-05,
+      "loss": 0.6999,
+      "step": 2184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671589378617878e-05,
+      "loss": 0.8728,
+      "step": 2185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669258626401528e-05,
+      "loss": 0.7378,
+      "step": 2186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766692686208905e-05,
+      "loss": 0.688,
+      "step": 2187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664594085988166e-05,
+      "loss": 0.6318,
+      "step": 2188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766226029840673e-05,
+      "loss": 0.6978,
+      "step": 2189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765992549965272e-05,
+      "loss": 0.775,
+      "step": 2190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657589690034257e-05,
+      "loss": 0.5863,
+      "step": 2191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655252869859593e-05,
+      "loss": 0.7288,
+      "step": 2192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652915039437118e-05,
+      "loss": 0.6554,
+      "step": 2193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650576199075346e-05,
+      "loss": 0.7524,
+      "step": 2194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648236349082928e-05,
+      "loss": 0.6837,
+      "step": 2195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645895489768653e-05,
+      "loss": 0.7707,
+      "step": 2196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643553621441436e-05,
+      "loss": 0.7332,
+      "step": 2197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764121074441033e-05,
+      "loss": 0.6985,
+      "step": 2198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763886685898452e-05,
+      "loss": 0.7769,
+      "step": 2199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636521965473324e-05,
+      "loss": 0.6941,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634176064186187e-05,
+      "loss": 0.7293,
+      "step": 2201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631829155432703e-05,
+      "loss": 0.744,
+      "step": 2202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762948123952258e-05,
+      "loss": 0.6878,
+      "step": 2203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762713231676567e-05,
+      "loss": 0.6831,
+      "step": 2204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762478238747196e-05,
+      "loss": 0.5997,
+      "step": 2205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622431451951557e-05,
+      "loss": 0.6672,
+      "step": 2206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762007951051472e-05,
+      "loss": 0.7182,
+      "step": 2207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761772656347182e-05,
+      "loss": 0.7544,
+      "step": 2208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761537261113337e-05,
+      "loss": 0.6239,
+      "step": 2209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613017653810026e-05,
+      "loss": 0.7428,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610661691812558e-05,
+      "loss": 0.6988,
+      "step": 2211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760830472545188e-05,
+      "loss": 0.7422,
+      "step": 2212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760594675503904e-05,
+      "loss": 0.6854,
+      "step": 2213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603587780885207e-05,
+      "loss": 0.803,
+      "step": 2214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601227803301694e-05,
+      "loss": 0.694,
+      "step": 2215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598866822599942e-05,
+      "loss": 0.7022,
+      "step": 2216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596504839091526e-05,
+      "loss": 0.9491,
+      "step": 2217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594141853088152e-05,
+      "loss": 0.7182,
+      "step": 2218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591777864901653e-05,
+      "loss": 0.8628,
+      "step": 2219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589412874844008e-05,
+      "loss": 0.6106,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758704688322731e-05,
+      "loss": 0.6503,
+      "step": 2221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584679890363806e-05,
+      "loss": 0.7664,
+      "step": 2222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582311896565855e-05,
+      "loss": 0.6725,
+      "step": 2223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579942902145958e-05,
+      "loss": 0.7347,
+      "step": 2224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577572907416746e-05,
+      "loss": 0.7549,
+      "step": 2225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757520191269098e-05,
+      "loss": 0.728,
+      "step": 2226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572829918281562e-05,
+      "loss": 0.7271,
+      "step": 2227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757045692450151e-05,
+      "loss": 0.8465,
+      "step": 2228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568082931663993e-05,
+      "loss": 0.8553,
+      "step": 2229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565707940082294e-05,
+      "loss": 0.7225,
+      "step": 2230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563331950069836e-05,
+      "loss": 0.7628,
+      "step": 2231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756095496194018e-05,
+      "loss": 0.7602,
+      "step": 2232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558576976007008e-05,
+      "loss": 0.8203,
+      "step": 2233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556197992584136e-05,
+      "loss": 0.7075,
+      "step": 2234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755381801198552e-05,
+      "loss": 0.7681,
+      "step": 2235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551437034525233e-05,
+      "loss": 0.7211,
+      "step": 2236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549055060517493e-05,
+      "loss": 0.6978,
+      "step": 2237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754667209027664e-05,
+      "loss": 0.7471,
+      "step": 2238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544288124117152e-05,
+      "loss": 0.684,
+      "step": 2239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541903162353638e-05,
+      "loss": 0.677,
+      "step": 2240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539517205300834e-05,
+      "loss": 0.7432,
+      "step": 2241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537130253273613e-05,
+      "loss": 0.738,
+      "step": 2242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534742306586972e-05,
+      "loss": 0.7039,
+      "step": 2243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532353365556042e-05,
+      "loss": 0.6597,
+      "step": 2244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752996343049609e-05,
+      "loss": 0.6646,
+      "step": 2245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527572501722516e-05,
+      "loss": 0.7641,
+      "step": 2246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525180579550837e-05,
+      "loss": 0.662,
+      "step": 2247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752278766429671e-05,
+      "loss": 0.6743,
+      "step": 2248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520393756275932e-05,
+      "loss": 0.5869,
+      "step": 2249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751799885580441e-05,
+      "loss": 0.6997,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515602963198204e-05,
+      "loss": 0.8273,
+      "step": 2251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513206078773488e-05,
+      "loss": 0.7189,
+      "step": 2252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751080820284658e-05,
+      "loss": 0.7574,
+      "step": 2253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750840933573392e-05,
+      "loss": 0.5589,
+      "step": 2254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750600947775208e-05,
+      "loss": 0.7335,
+      "step": 2255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750360862921777e-05,
+      "loss": 0.7512,
+      "step": 2256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501206790447813e-05,
+      "loss": 0.7696,
+      "step": 2257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498803961759187e-05,
+      "loss": 0.7932,
+      "step": 2258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496400143468984e-05,
+      "loss": 0.6397,
+      "step": 2259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749399533589443e-05,
+      "loss": 0.6458,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749158953935288e-05,
+      "loss": 0.6484,
+      "step": 2261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489182754161825e-05,
+      "loss": 0.6962,
+      "step": 2262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748677498063889e-05,
+      "loss": 0.7705,
+      "step": 2263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748436621910181e-05,
+      "loss": 0.6911,
+      "step": 2264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481956469868474e-05,
+      "loss": 0.7355,
+      "step": 2265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479545733256887e-05,
+      "loss": 0.7957,
+      "step": 2266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477134009585196e-05,
+      "loss": 0.691,
+      "step": 2267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474721299171662e-05,
+      "loss": 0.6904,
+      "step": 2268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747230760233469e-05,
+      "loss": 0.7231,
+      "step": 2269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469892919392815e-05,
+      "loss": 0.7079,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467477250664685e-05,
+      "loss": 0.6797,
+      "step": 2271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465060596469103e-05,
+      "loss": 0.7372,
+      "step": 2272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462642957124987e-05,
+      "loss": 0.7457,
+      "step": 2273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746022433295138e-05,
+      "loss": 0.7053,
+      "step": 2274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457804724267477e-05,
+      "loss": 0.6051,
+      "step": 2275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455384131392573e-05,
+      "loss": 0.7503,
+      "step": 2276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452962554646122e-05,
+      "loss": 0.6428,
+      "step": 2277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745053999434769e-05,
+      "loss": 0.6661,
+      "step": 2278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744811645081697e-05,
+      "loss": 0.7269,
+      "step": 2279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74456919243738e-05,
+      "loss": 0.7292,
+      "step": 2280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744326641533814e-05,
+      "loss": 0.6677,
+      "step": 2281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440839924030076e-05,
+      "loss": 0.7369,
+      "step": 2282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7438412450769828e-05,
+      "loss": 0.6841,
+      "step": 2283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435983995877745e-05,
+      "loss": 0.7232,
+      "step": 2284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433554559674305e-05,
+      "loss": 0.7364,
+      "step": 2285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431124142480116e-05,
+      "loss": 0.6776,
+      "step": 2286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428692744615913e-05,
+      "loss": 0.8039,
+      "step": 2287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426260366402567e-05,
+      "loss": 0.6586,
+      "step": 2288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742382700816107e-05,
+      "loss": 0.7208,
+      "step": 2289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742139267021255e-05,
+      "loss": 0.6652,
+      "step": 2290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418957352878255e-05,
+      "loss": 0.7406,
+      "step": 2291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416521056479577e-05,
+      "loss": 0.72,
+      "step": 2292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414083781338024e-05,
+      "loss": 0.7338,
+      "step": 2293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411645527775243e-05,
+      "loss": 0.7064,
+      "step": 2294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409206296113e-05,
+      "loss": 0.7994,
+      "step": 2295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406766086673196e-05,
+      "loss": 0.7335,
+      "step": 2296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404324899777863e-05,
+      "loss": 0.7137,
+      "step": 2297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401882735749157e-05,
+      "loss": 0.7599,
+      "step": 2298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399439594909367e-05,
+      "loss": 0.6217,
+      "step": 2299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396995477580912e-05,
+      "loss": 0.7004,
+      "step": 2300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394550384086332e-05,
+      "loss": 0.7619,
+      "step": 2301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73921043147483e-05,
+      "loss": 0.6728,
+      "step": 2302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389657269889623e-05,
+      "loss": 0.7744,
+      "step": 2303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387209249833228e-05,
+      "loss": 0.7378,
+      "step": 2304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384760254902182e-05,
+      "loss": 0.7031,
+      "step": 2305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738231028541967e-05,
+      "loss": 0.71,
+      "step": 2306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379859341709006e-05,
+      "loss": 0.7318,
+      "step": 2307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737740742409364e-05,
+      "loss": 0.7312,
+      "step": 2308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374954532897145e-05,
+      "loss": 0.725,
+      "step": 2309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372500668443223e-05,
+      "loss": 0.6564,
+      "step": 2310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7370045831055708e-05,
+      "loss": 0.6771,
+      "step": 2311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736759002105856e-05,
+      "loss": 0.72,
+      "step": 2312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365133238775865e-05,
+      "loss": 0.6954,
+      "step": 2313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736267548453184e-05,
+      "loss": 0.6695,
+      "step": 2314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360216758650826e-05,
+      "loss": 0.7306,
+      "step": 2315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357757061457306e-05,
+      "loss": 0.7134,
+      "step": 2316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735529639327587e-05,
+      "loss": 0.6072,
+      "step": 2317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352834754431257e-05,
+      "loss": 0.7311,
+      "step": 2318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350372145248318e-05,
+      "loss": 0.7258,
+      "step": 2319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734790856605204e-05,
+      "loss": 0.6871,
+      "step": 2320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345444017167534e-05,
+      "loss": 0.7852,
+      "step": 2321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342978498920048e-05,
+      "loss": 0.76,
+      "step": 2322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340512011634943e-05,
+      "loss": 0.6908,
+      "step": 2323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338044555637722e-05,
+      "loss": 0.8591,
+      "step": 2324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335576131254004e-05,
+      "loss": 0.7041,
+      "step": 2325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333106738809555e-05,
+      "loss": 0.6945,
+      "step": 2326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733063637863024e-05,
+      "loss": 0.776,
+      "step": 2327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732816505104208e-05,
+      "loss": 0.6988,
+      "step": 2328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325692756371196e-05,
+      "loss": 0.7748,
+      "step": 2329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732321949494387e-05,
+      "loss": 0.7058,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320745267086478e-05,
+      "loss": 0.6856,
+      "step": 2331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318270073125548e-05,
+      "loss": 0.6525,
+      "step": 2332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731579391338772e-05,
+      "loss": 0.6865,
+      "step": 2333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313316788199773e-05,
+      "loss": 0.7404,
+      "step": 2334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310838697888603e-05,
+      "loss": 0.7907,
+      "step": 2335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730835964278124e-05,
+      "loss": 0.5859,
+      "step": 2336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305879623204846e-05,
+      "loss": 0.7612,
+      "step": 2337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303398639486696e-05,
+      "loss": 0.675,
+      "step": 2338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73009166919542e-05,
+      "loss": 0.6875,
+      "step": 2339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72984337809349e-05,
+      "loss": 0.5922,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7295949906756463e-05,
+      "loss": 0.7452,
+      "step": 2341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293465069746672e-05,
+      "loss": 0.6628,
+      "step": 2342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729097927023345e-05,
+      "loss": 0.6864,
+      "step": 2343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288492508544846e-05,
+      "loss": 0.6992,
+      "step": 2344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286004785009032e-05,
+      "loss": 0.7397,
+      "step": 2345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72835160999543e-05,
+      "loss": 0.6591,
+      "step": 2346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281026453709086e-05,
+      "loss": 0.7569,
+      "step": 2347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727853584660194e-05,
+      "loss": 0.6649,
+      "step": 2348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276044278961542e-05,
+      "loss": 0.6342,
+      "step": 2349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273551751116698e-05,
+      "loss": 0.7612,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727105826339634e-05,
+      "loss": 0.791,
+      "step": 2351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268563816129534e-05,
+      "loss": 0.7659,
+      "step": 2352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266068409645465e-05,
+      "loss": 0.7436,
+      "step": 2353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263572044273445e-05,
+      "loss": 0.8156,
+      "step": 2354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726107472034292e-05,
+      "loss": 0.6756,
+      "step": 2355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258576438183444e-05,
+      "loss": 0.7051,
+      "step": 2356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256077198124723e-05,
+      "loss": 0.6656,
+      "step": 2357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253577000496565e-05,
+      "loss": 0.738,
+      "step": 2358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725107584562893e-05,
+      "loss": 0.6665,
+      "step": 2359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248573733851875e-05,
+      "loss": 0.646,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246070665495608e-05,
+      "loss": 0.6573,
+      "step": 2361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243566640890453e-05,
+      "loss": 0.735,
+      "step": 2362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724106166036686e-05,
+      "loss": 0.7803,
+      "step": 2363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238555724255404e-05,
+      "loss": 0.7338,
+      "step": 2364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236048832886793e-05,
+      "loss": 0.6963,
+      "step": 2365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723354098659185e-05,
+      "loss": 0.678,
+      "step": 2366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231032185701534e-05,
+      "loss": 0.6642,
+      "step": 2367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228522430546925e-05,
+      "loss": 0.6825,
+      "step": 2368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722601172145923e-05,
+      "loss": 0.7327,
+      "step": 2369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722350005876978e-05,
+      "loss": 0.6725,
+      "step": 2370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220987442810042e-05,
+      "loss": 0.6774,
+      "step": 2371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218473873911594e-05,
+      "loss": 0.8196,
+      "step": 2372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215959352406147e-05,
+      "loss": 0.7025,
+      "step": 2373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721344387862554e-05,
+      "loss": 0.6621,
+      "step": 2374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721092745290173e-05,
+      "loss": 0.6881,
+      "step": 2375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208410075566808e-05,
+      "loss": 0.7213,
+      "step": 2376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720589174695299e-05,
+      "loss": 0.6035,
+      "step": 2377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203372467392606e-05,
+      "loss": 0.7835,
+      "step": 2378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200852237218125e-05,
+      "loss": 0.5994,
+      "step": 2379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198331056762137e-05,
+      "loss": 0.7327,
+      "step": 2380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7195808926357357e-05,
+      "loss": 0.6951,
+      "step": 2381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193285846336628e-05,
+      "loss": 0.7381,
+      "step": 2382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719076181703291e-05,
+      "loss": 0.6431,
+      "step": 2383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188236838779297e-05,
+      "loss": 0.5567,
+      "step": 2384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185710911909e-05,
+      "loss": 0.6669,
+      "step": 2385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183184036755373e-05,
+      "loss": 0.7375,
+      "step": 2386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180656213651866e-05,
+      "loss": 0.5704,
+      "step": 2387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178127442932083e-05,
+      "loss": 0.6255,
+      "step": 2388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175597724929737e-05,
+      "loss": 0.7175,
+      "step": 2389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173067059978667e-05,
+      "loss": 0.7324,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170535448412846e-05,
+      "loss": 0.7286,
+      "step": 2391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716800289056636e-05,
+      "loss": 0.7036,
+      "step": 2392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716546938677342e-05,
+      "loss": 0.6627,
+      "step": 2393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162934937368382e-05,
+      "loss": 0.7528,
+      "step": 2394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160399542685698e-05,
+      "loss": 0.7021,
+      "step": 2395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157863203059967e-05,
+      "loss": 0.739,
+      "step": 2396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71553259188259e-05,
+      "loss": 0.7143,
+      "step": 2397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152787690318344e-05,
+      "loss": 0.7209,
+      "step": 2398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150248517872256e-05,
+      "loss": 0.7518,
+      "step": 2399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714770840182273e-05,
+      "loss": 0.7551,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145167342504975e-05,
+      "loss": 0.7151,
+      "step": 2401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142625340254337e-05,
+      "loss": 0.715,
+      "step": 2402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714008239540627e-05,
+      "loss": 0.8061,
+      "step": 2403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137538508296364e-05,
+      "loss": 0.7762,
+      "step": 2404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134993679260334e-05,
+      "loss": 0.7862,
+      "step": 2405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132447908634015e-05,
+      "loss": 0.6655,
+      "step": 2406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129901196753366e-05,
+      "loss": 0.8144,
+      "step": 2407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712735354395447e-05,
+      "loss": 0.7083,
+      "step": 2408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124804950573536e-05,
+      "loss": 0.7226,
+      "step": 2409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122255416946897e-05,
+      "loss": 0.7071,
+      "step": 2410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119704943411008e-05,
+      "loss": 0.7484,
+      "step": 2411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117153530302453e-05,
+      "loss": 0.6216,
+      "step": 2412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114601177957936e-05,
+      "loss": 0.7613,
+      "step": 2413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112047886714283e-05,
+      "loss": 0.6388,
+      "step": 2414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710949365690845e-05,
+      "loss": 0.7383,
+      "step": 2415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106938488877508e-05,
+      "loss": 0.6961,
+      "step": 2416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104382382958665e-05,
+      "loss": 0.7028,
+      "step": 2417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710182533948924e-05,
+      "loss": 0.7779,
+      "step": 2418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709926735880668e-05,
+      "loss": 0.6969,
+      "step": 2419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096708441248556e-05,
+      "loss": 0.6314,
+      "step": 2420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709414858715257e-05,
+      "loss": 0.6747,
+      "step": 2421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091587796856533e-05,
+      "loss": 0.6799,
+      "step": 2422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708902607069839e-05,
+      "loss": 0.7498,
+      "step": 2423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086463409016203e-05,
+      "loss": 0.7863,
+      "step": 2424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708389981214817e-05,
+      "loss": 0.6726,
+      "step": 2425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081335280432596e-05,
+      "loss": 0.6326,
+      "step": 2426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707876981420792e-05,
+      "loss": 0.6777,
+      "step": 2427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076203413812694e-05,
+      "loss": 0.7524,
+      "step": 2428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073636079585614e-05,
+      "loss": 0.732,
+      "step": 2429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.768,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068498610991213e-05,
+      "loss": 0.6919,
+      "step": 2431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065928477301875e-05,
+      "loss": 0.6617,
+      "step": 2432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063357411136635e-05,
+      "loss": 0.6437,
+      "step": 2433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70607854128348e-05,
+      "loss": 0.8172,
+      "step": 2434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705821248273578e-05,
+      "loss": 0.7023,
+      "step": 2435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055638621179127e-05,
+      "loss": 0.7394,
+      "step": 2436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053063828504505e-05,
+      "loss": 0.7198,
+      "step": 2437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050488105051702e-05,
+      "loss": 0.7221,
+      "step": 2438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704791145116064e-05,
+      "loss": 0.7945,
+      "step": 2439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045333867171343e-05,
+      "loss": 0.6484,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042755353423978e-05,
+      "loss": 0.6962,
+      "step": 2441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040175910258822e-05,
+      "loss": 0.7475,
+      "step": 2442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037595538016277e-05,
+      "loss": 0.7714,
+      "step": 2443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703501423703687e-05,
+      "loss": 0.6653,
+      "step": 2444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032432007661258e-05,
+      "loss": 0.6962,
+      "step": 2445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029848850230207e-05,
+      "loss": 0.7123,
+      "step": 2446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027264765084607e-05,
+      "loss": 0.7521,
+      "step": 2447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024679752565476e-05,
+      "loss": 0.7039,
+      "step": 2448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022093813013958e-05,
+      "loss": 0.7765,
+      "step": 2449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019506946771306e-05,
+      "loss": 0.6739,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7016919154178914e-05,
+      "loss": 0.6886,
+      "step": 2451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7014330435578273e-05,
+      "loss": 0.7038,
+      "step": 2452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011740791311025e-05,
+      "loss": 0.825,
+      "step": 2453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009150221718914e-05,
+      "loss": 0.7568,
+      "step": 2454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700655872714381e-05,
+      "loss": 0.6869,
+      "step": 2455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700396630792771e-05,
+      "loss": 0.7399,
+      "step": 2456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001372964412733e-05,
+      "loss": 0.6382,
+      "step": 2457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998778696941108e-05,
+      "loss": 0.641,
+      "step": 2458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996183505855207e-05,
+      "loss": 0.6848,
+      "step": 2459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69935873914975e-05,
+      "loss": 0.734,
+      "step": 2460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990990354210598e-05,
+      "loss": 0.8406,
+      "step": 2461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698839239433723e-05,
+      "loss": 0.7673,
+      "step": 2462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985793512220234e-05,
+      "loss": 0.719,
+      "step": 2463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983193708202585e-05,
+      "loss": 0.6295,
+      "step": 2464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698059298262737e-05,
+      "loss": 0.6986,
+      "step": 2465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977991335837807e-05,
+      "loss": 0.7314,
+      "step": 2466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975388768177223e-05,
+      "loss": 0.7287,
+      "step": 2467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972785279989082e-05,
+      "loss": 0.617,
+      "step": 2468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697018087161695e-05,
+      "loss": 0.6766,
+      "step": 2469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967575543404536e-05,
+      "loss": 0.6643,
+      "step": 2470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964969295695657e-05,
+      "loss": 0.8006,
+      "step": 2471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962362128834248e-05,
+      "loss": 0.7787,
+      "step": 2472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959754043164376e-05,
+      "loss": 0.7638,
+      "step": 2473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957145039030225e-05,
+      "loss": 0.7283,
+      "step": 2474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69545351167761e-05,
+      "loss": 0.7086,
+      "step": 2475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951924276746425e-05,
+      "loss": 0.7235,
+      "step": 2476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694931251928575e-05,
+      "loss": 0.6594,
+      "step": 2477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694669984473874e-05,
+      "loss": 0.7109,
+      "step": 2478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944086253450182e-05,
+      "loss": 0.6734,
+      "step": 2479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941471745764996e-05,
+      "loss": 0.6476,
+      "step": 2480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938856322028206e-05,
+      "loss": 0.7568,
+      "step": 2481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693623998258496e-05,
+      "loss": 0.5978,
+      "step": 2482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693362272778054e-05,
+      "loss": 0.6936,
+      "step": 2483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931004557960337e-05,
+      "loss": 0.7647,
+      "step": 2484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928385473469863e-05,
+      "loss": 0.5985,
+      "step": 2485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925765474654754e-05,
+      "loss": 0.7216,
+      "step": 2486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923144561860767e-05,
+      "loss": 0.7429,
+      "step": 2487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920522735433777e-05,
+      "loss": 0.7878,
+      "step": 2488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691789999571978e-05,
+      "loss": 0.7909,
+      "step": 2489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915276343064896e-05,
+      "loss": 0.7246,
+      "step": 2490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691265177781536e-05,
+      "loss": 0.6365,
+      "step": 2491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910026300317534e-05,
+      "loss": 0.7914,
+      "step": 2492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907399910917897e-05,
+      "loss": 0.7949,
+      "step": 2493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904772609963043e-05,
+      "loss": 0.6753,
+      "step": 2494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902144397799695e-05,
+      "loss": 0.6608,
+      "step": 2495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899515274774697e-05,
+      "loss": 0.6516,
+      "step": 2496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896885241235e-05,
+      "loss": 0.755,
+      "step": 2497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689425429752769e-05,
+      "loss": 0.6468,
+      "step": 2498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891622443999966e-05,
+      "loss": 0.7954,
+      "step": 2499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888989680999145e-05,
+      "loss": 0.671,
+      "step": 2500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886356008872675e-05,
+      "loss": 0.7124,
+      "step": 2501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883721427968106e-05,
+      "loss": 0.6823,
+      "step": 2502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881085938633125e-05,
+      "loss": 0.7509,
+      "step": 2503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878449541215535e-05,
+      "loss": 0.7046,
+      "step": 2504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875812236063246e-05,
+      "loss": 0.6118,
+      "step": 2505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687317402352431e-05,
+      "loss": 0.6597,
+      "step": 2506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687053490394687e-05,
+      "loss": 0.6955,
+      "step": 2507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686789487767922e-05,
+      "loss": 0.6966,
+      "step": 2508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865253945069754e-05,
+      "loss": 0.7061,
+      "step": 2509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686261210646699e-05,
+      "loss": 0.7657,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859969362219565e-05,
+      "loss": 0.7792,
+      "step": 2511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857325712676233e-05,
+      "loss": 0.7615,
+      "step": 2512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685468115818588e-05,
+      "loss": 0.7159,
+      "step": 2513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852035699097497e-05,
+      "loss": 0.7638,
+      "step": 2514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68493893357602e-05,
+      "loss": 0.6607,
+      "step": 2515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846742068523225e-05,
+      "loss": 0.7329,
+      "step": 2516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844093897735927e-05,
+      "loss": 0.6719,
+      "step": 2517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841444823747776e-05,
+      "loss": 0.6919,
+      "step": 2518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838794846908365e-05,
+      "loss": 0.6475,
+      "step": 2519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836143967567413e-05,
+      "loss": 0.654,
+      "step": 2520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833492186074743e-05,
+      "loss": 0.724,
+      "step": 2521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683083950278031e-05,
+      "loss": 0.602,
+      "step": 2522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828185918034178e-05,
+      "loss": 0.633,
+      "step": 2523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825531432186545e-05,
+      "loss": 0.6737,
+      "step": 2524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822876045587705e-05,
+      "loss": 0.7369,
+      "step": 2525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820219758588092e-05,
+      "loss": 0.6658,
+      "step": 2526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681756257153825e-05,
+      "loss": 0.7978,
+      "step": 2527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681490448478884e-05,
+      "loss": 0.652,
+      "step": 2528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812245498690645e-05,
+      "loss": 0.5623,
+      "step": 2529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809585613594568e-05,
+      "loss": 0.7516,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806924829851625e-05,
+      "loss": 0.7198,
+      "step": 2531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804263147812958e-05,
+      "loss": 0.7923,
+      "step": 2532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801600567829822e-05,
+      "loss": 0.7074,
+      "step": 2533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679893709025359e-05,
+      "loss": 0.7355,
+      "step": 2534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796272715435758e-05,
+      "loss": 0.814,
+      "step": 2535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793607443727937e-05,
+      "loss": 0.681,
+      "step": 2536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790941275481858e-05,
+      "loss": 0.7531,
+      "step": 2537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678827421104937e-05,
+      "loss": 0.7605,
+      "step": 2538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678560625078244e-05,
+      "loss": 0.7497,
+      "step": 2539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678293739503315e-05,
+      "loss": 0.7284,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678026764415371e-05,
+      "loss": 0.715,
+      "step": 2541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777596998496434e-05,
+      "loss": 0.7075,
+      "step": 2542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774925458413766e-05,
+      "loss": 0.6412,
+      "step": 2543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677225302425826e-05,
+      "loss": 0.6604,
+      "step": 2544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67695796963826e-05,
+      "loss": 0.6219,
+      "step": 2545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766905475139568e-05,
+      "loss": 0.6166,
+      "step": 2546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764230360882087e-05,
+      "loss": 0.7892,
+      "step": 2547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761554353963173e-05,
+      "loss": 0.7344,
+      "step": 2548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758877454735987e-05,
+      "loss": 0.7538,
+      "step": 2549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756199663553784e-05,
+      "loss": 0.6728,
+      "step": 2550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675352098076995e-05,
+      "loss": 0.7078,
+      "step": 2551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750841406737985e-05,
+      "loss": 0.7295,
+      "step": 2552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674816094181151e-05,
+      "loss": 0.6873,
+      "step": 2553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745479586344253e-05,
+      "loss": 0.7069,
+      "step": 2554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742797340690072e-05,
+      "loss": 0.7364,
+      "step": 2555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740114205202934e-05,
+      "loss": 0.7423,
+      "step": 2556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737430180236934e-05,
+      "loss": 0.77,
+      "step": 2557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734745266146274e-05,
+      "loss": 0.635,
+      "step": 2558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673205946328527e-05,
+      "loss": 0.7739,
+      "step": 2559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672937277200837e-05,
+      "loss": 0.6599,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726685192670125e-05,
+      "loss": 0.6872,
+      "step": 2561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723996725625215e-05,
+      "loss": 0.6893,
+      "step": 2562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721307371228426e-05,
+      "loss": 0.6591,
+      "step": 2563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671861712983467e-05,
+      "loss": 0.6737,
+      "step": 2564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715926001798974e-05,
+      "loss": 0.6858,
+      "step": 2565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713233987476476e-05,
+      "loss": 0.6982,
+      "step": 2566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671054108722244e-05,
+      "loss": 0.7352,
+      "step": 2567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707847301392237e-05,
+      "loss": 0.6494,
+      "step": 2568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705152630341363e-05,
+      "loss": 0.6756,
+      "step": 2569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702457074425435e-05,
+      "loss": 0.7178,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699760634000166e-05,
+      "loss": 0.6748,
+      "step": 2571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697063309421412e-05,
+      "loss": 0.6678,
+      "step": 2572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694365101045127e-05,
+      "loss": 0.6735,
+      "step": 2573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669166600922739e-05,
+      "loss": 0.6911,
+      "step": 2574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688966034324393e-05,
+      "loss": 0.7885,
+      "step": 2575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6686265176692446e-05,
+      "loss": 0.682,
+      "step": 2576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668356343668798e-05,
+      "loss": 0.7454,
+      "step": 2577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680860814667532e-05,
+      "loss": 0.681,
+      "step": 2578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667815731098777e-05,
+      "loss": 0.6265,
+      "step": 2579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675452926005456e-05,
+      "loss": 0.6504,
+      "step": 2580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672747660077494e-05,
+      "loss": 0.7123,
+      "step": 2581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670041513560887e-05,
+      "loss": 0.6889,
+      "step": 2582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666733448681276e-05,
+      "loss": 0.6391,
+      "step": 2583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666462658019036e-05,
+      "loss": 0.7434,
+      "step": 2584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661917794051033e-05,
+      "loss": 0.662,
+      "step": 2585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659208128752257e-05,
+      "loss": 0.7215,
+      "step": 2586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656497584651624e-05,
+      "loss": 0.7217,
+      "step": 2587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653786162106834e-05,
+      "loss": 0.7303,
+      "step": 2588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651073861475713e-05,
+      "loss": 0.6956,
+      "step": 2589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664836068311619e-05,
+      "loss": 0.7232,
+      "step": 2590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6645646627386324e-05,
+      "loss": 0.6986,
+      "step": 2591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664293169464428e-05,
+      "loss": 0.7548,
+      "step": 2592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640215885248344e-05,
+      "loss": 0.6649,
+      "step": 2593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637499199556915e-05,
+      "loss": 0.7419,
+      "step": 2594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663478163792851e-05,
+      "loss": 0.7284,
+      "step": 2595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663206320072176e-05,
+      "loss": 0.7062,
+      "step": 2596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629343888295406e-05,
+      "loss": 0.6629,
+      "step": 2597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626623701008317e-05,
+      "loss": 0.7678,
+      "step": 2598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623902639219462e-05,
+      "loss": 0.6763,
+      "step": 2599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662118070328794e-05,
+      "loss": 0.6785,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618457893572962e-05,
+      "loss": 0.6409,
+      "step": 2601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661573421043385e-05,
+      "loss": 0.7589,
+      "step": 2602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613009654230037e-05,
+      "loss": 0.6984,
+      "step": 2603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661028422532108e-05,
+      "loss": 0.7156,
+      "step": 2604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660755792406665e-05,
+      "loss": 0.7486,
+      "step": 2605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660483075082653e-05,
+      "loss": 0.6718,
+      "step": 2606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602102705960623e-05,
+      "loss": 0.7675,
+      "step": 2607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599373789828935e-05,
+      "loss": 0.742,
+      "step": 2608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596644002791604e-05,
+      "loss": 0.7032,
+      "step": 2609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659391334520887e-05,
+      "loss": 0.8326,
+      "step": 2610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659118181744109e-05,
+      "loss": 0.727,
+      "step": 2611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588449419848746e-05,
+      "loss": 0.7448,
+      "step": 2612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658571615279242e-05,
+      "loss": 0.6549,
+      "step": 2613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582982016632818e-05,
+      "loss": 0.7604,
+      "step": 2614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580247011730755e-05,
+      "loss": 0.5985,
+      "step": 2615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657751113844717e-05,
+      "loss": 0.7259,
+      "step": 2616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574774397143105e-05,
+      "loss": 0.7681,
+      "step": 2617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572036788179728e-05,
+      "loss": 0.739,
+      "step": 2618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656929831191831e-05,
+      "loss": 0.7306,
+      "step": 2619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566558968720238e-05,
+      "loss": 0.6351,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656381875894703e-05,
+      "loss": 0.7325,
+      "step": 2621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561077682960297e-05,
+      "loss": 0.7132,
+      "step": 2622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6558335741121773e-05,
+      "loss": 0.6808,
+      "step": 2623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555592933793315e-05,
+      "loss": 0.6805,
+      "step": 2624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552849261336873e-05,
+      "loss": 0.6923,
+      "step": 2625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655010472411453e-05,
+      "loss": 0.6936,
+      "step": 2626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654735932248848e-05,
+      "loss": 0.7006,
+      "step": 2627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544613056821023e-05,
+      "loss": 0.6119,
+      "step": 2628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654186592747458e-05,
+      "loss": 0.7093,
+      "step": 2629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539117934811682e-05,
+      "loss": 0.8088,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536369079194977e-05,
+      "loss": 0.6237,
+      "step": 2631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653361936098723e-05,
+      "loss": 0.7573,
+      "step": 2632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653086878055131e-05,
+      "loss": 0.6933,
+      "step": 2633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528117338250205e-05,
+      "loss": 0.722,
+      "step": 2634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652536503444702e-05,
+      "loss": 0.6488,
+      "step": 2635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522611869504972e-05,
+      "loss": 0.7029,
+      "step": 2636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519857843787388e-05,
+      "loss": 0.8206,
+      "step": 2637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517102957657714e-05,
+      "loss": 0.7019,
+      "step": 2638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65143472114795e-05,
+      "loss": 0.6736,
+      "step": 2639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511590605616423e-05,
+      "loss": 0.6491,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508833140432263e-05,
+      "loss": 0.7728,
+      "step": 2641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506074816290916e-05,
+      "loss": 0.7035,
+      "step": 2642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503315633556402e-05,
+      "loss": 0.6807,
+      "step": 2643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500555592592825e-05,
+      "loss": 0.7162,
+      "step": 2644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497794693764443e-05,
+      "loss": 0.6673,
+      "step": 2645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495032937435593e-05,
+      "loss": 0.6302,
+      "step": 2646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492270323970742e-05,
+      "loss": 0.7079,
+      "step": 2647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489506853734467e-05,
+      "loss": 0.646,
+      "step": 2648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486742527091457e-05,
+      "loss": 0.5813,
+      "step": 2649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648397734440651e-05,
+      "loss": 0.7056,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648121130604455e-05,
+      "loss": 0.6882,
+      "step": 2651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478444412370606e-05,
+      "loss": 0.7677,
+      "step": 2652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475676663749806e-05,
+      "loss": 0.7491,
+      "step": 2653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472908060547414e-05,
+      "loss": 0.7888,
+      "step": 2654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470138603128794e-05,
+      "loss": 0.7878,
+      "step": 2655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646736829185943e-05,
+      "loss": 0.6857,
+      "step": 2656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646459712710491e-05,
+      "loss": 0.8,
+      "step": 2657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461825109230937e-05,
+      "loss": 0.659,
+      "step": 2658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645905223860333e-05,
+      "loss": 0.6846,
+      "step": 2659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456278515588023e-05,
+      "loss": 0.6729,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453503940551055e-05,
+      "loss": 0.7329,
+      "step": 2661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450728513858583e-05,
+      "loss": 0.6882,
+      "step": 2662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447952235876874e-05,
+      "loss": 0.7291,
+      "step": 2663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64451751069723e-05,
+      "loss": 0.7626,
+      "step": 2664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442397127511366e-05,
+      "loss": 0.5643,
+      "step": 2665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439618297860668e-05,
+      "loss": 0.8827,
+      "step": 2666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643683861838692e-05,
+      "loss": 0.6961,
+      "step": 2667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434058089456956e-05,
+      "loss": 0.6661,
+      "step": 2668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431276711437715e-05,
+      "loss": 0.7266,
+      "step": 2669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642849448469625e-05,
+      "loss": 0.6252,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425711409599723e-05,
+      "loss": 0.6638,
+      "step": 2671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642292748651542e-05,
+      "loss": 0.8053,
+      "step": 2672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420142715810715e-05,
+      "loss": 0.6949,
+      "step": 2673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641735709785312e-05,
+      "loss": 0.6467,
+      "step": 2674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641457063301024e-05,
+      "loss": 0.6671,
+      "step": 2675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411783321649806e-05,
+      "loss": 0.6894,
+      "step": 2676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640899516413965e-05,
+      "loss": 0.6467,
+      "step": 2677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406206160847717e-05,
+      "loss": 0.6599,
+      "step": 2678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403416312142066e-05,
+      "loss": 0.7315,
+      "step": 2679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400625618390872e-05,
+      "loss": 0.7262,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397834079962415e-05,
+      "loss": 0.574,
+      "step": 2681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395041697225088e-05,
+      "loss": 0.6961,
+      "step": 2682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392248470547395e-05,
+      "loss": 1.0841,
+      "step": 2683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389454400297956e-05,
+      "loss": 0.6638,
+      "step": 2684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386659486845493e-05,
+      "loss": 0.7051,
+      "step": 2685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638386373055885e-05,
+      "loss": 0.7813,
+      "step": 2686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381067131806974e-05,
+      "loss": 0.693,
+      "step": 2687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378269690958925e-05,
+      "loss": 0.7868,
+      "step": 2688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375471408383884e-05,
+      "loss": 0.6613,
+      "step": 2689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372672284451126e-05,
+      "loss": 0.591,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369872319530048e-05,
+      "loss": 0.7234,
+      "step": 2691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367071513990155e-05,
+      "loss": 0.767,
+      "step": 2692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364269868201063e-05,
+      "loss": 0.6347,
+      "step": 2693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361467382532502e-05,
+      "loss": 0.7347,
+      "step": 2694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358664057354313e-05,
+      "loss": 0.7644,
+      "step": 2695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355859893036433e-05,
+      "loss": 0.6693,
+      "step": 2696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353054889948937e-05,
+      "loss": 0.6796,
+      "step": 2697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350249048461983e-05,
+      "loss": 0.6944,
+      "step": 2698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347442368945855e-05,
+      "loss": 0.6479,
+      "step": 2699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344634851770954e-05,
+      "loss": 0.7014,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634182649730777e-05,
+      "loss": 0.6785,
+      "step": 2701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633901730592692e-05,
+      "loss": 0.6265,
+      "step": 2702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336207277999127e-05,
+      "loss": 0.7955,
+      "step": 2703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333396413895227e-05,
+      "loss": 0.7185,
+      "step": 2704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330584713986164e-05,
+      "loss": 0.839,
+      "step": 2705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6327772178642986e-05,
+      "loss": 0.6506,
+      "step": 2706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324958808236865e-05,
+      "loss": 0.8058,
+      "step": 2707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632214460313907e-05,
+      "loss": 0.6807,
+      "step": 2708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631932956372099e-05,
+      "loss": 0.6912,
+      "step": 2709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631651369035412e-05,
+      "loss": 0.7025,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631369698341006e-05,
+      "loss": 0.6629,
+      "step": 2711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631087944326053e-05,
+      "loss": 0.6996,
+      "step": 2712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630806107027735e-05,
+      "loss": 0.7428,
+      "step": 2713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630524186483246e-05,
+      "loss": 0.6899,
+      "step": 2714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302421827297905e-05,
+      "loss": 0.782,
+      "step": 2715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6299600958045833e-05,
+      "loss": 0.7325,
+      "step": 2716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296779257448514e-05,
+      "loss": 0.6822,
+      "step": 2717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293956725878318e-05,
+      "loss": 0.7363,
+      "step": 2718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291133363707736e-05,
+      "loss": 0.7089,
+      "step": 2719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628830917130935e-05,
+      "loss": 0.7327,
+      "step": 2720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285484149055872e-05,
+      "loss": 0.7612,
+      "step": 2721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282658297320112e-05,
+      "loss": 0.7234,
+      "step": 2722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279831616474985e-05,
+      "loss": 0.6773,
+      "step": 2723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277004106893533e-05,
+      "loss": 0.6549,
+      "step": 2724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274175768948892e-05,
+      "loss": 0.6166,
+      "step": 2725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271346603014312e-05,
+      "loss": 0.7412,
+      "step": 2726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626851660946315e-05,
+      "loss": 0.71,
+      "step": 2727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265685788668875e-05,
+      "loss": 0.6814,
+      "step": 2728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626285414100507e-05,
+      "loss": 0.7716,
+      "step": 2729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260021666845415e-05,
+      "loss": 0.6505,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257188366563704e-05,
+      "loss": 0.7257,
+      "step": 2731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254354240533852e-05,
+      "loss": 0.6799,
+      "step": 2732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251519289129866e-05,
+      "loss": 0.7172,
+      "step": 2733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6248683512725864e-05,
+      "loss": 0.6631,
+      "step": 2734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245846911696088e-05,
+      "loss": 0.6797,
+      "step": 2735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243009486414872e-05,
+      "loss": 0.5875,
+      "step": 2736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240171237256664e-05,
+      "loss": 0.688,
+      "step": 2737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237332164596024e-05,
+      "loss": 0.6828,
+      "step": 2738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234492268807617e-05,
+      "loss": 0.7136,
+      "step": 2739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231651550266225e-05,
+      "loss": 0.7672,
+      "step": 2740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228810009346724e-05,
+      "loss": 0.7224,
+      "step": 2741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225967646424107e-05,
+      "loss": 0.6159,
+      "step": 2742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6223124461873476e-05,
+      "loss": 1.0289,
+      "step": 2743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220280456070043e-05,
+      "loss": 0.6832,
+      "step": 2744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217435629389123e-05,
+      "loss": 0.7325,
+      "step": 2745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621458998220614e-05,
+      "loss": 0.7446,
+      "step": 2746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621174351489663e-05,
+      "loss": 0.6974,
+      "step": 2747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208896227836236e-05,
+      "loss": 0.645,
+      "step": 2748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620604812140071e-05,
+      "loss": 0.6984,
+      "step": 2749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620319919596591e-05,
+      "loss": 0.5836,
+      "step": 2750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200349451907794e-05,
+      "loss": 0.6614,
+      "step": 2751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619749888960245e-05,
+      "loss": 0.7108,
+      "step": 2752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194647509426053e-05,
+      "loss": 0.6466,
+      "step": 2753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191795311754897e-05,
+      "loss": 0.7083,
+      "step": 2754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618894229696538e-05,
+      "loss": 0.7571,
+      "step": 2755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186088465434008e-05,
+      "loss": 0.6925,
+      "step": 2756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618323381753739e-05,
+      "loss": 0.6406,
+      "step": 2757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180378353652256e-05,
+      "loss": 0.6083,
+      "step": 2758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177522074155436e-05,
+      "loss": 0.6947,
+      "step": 2759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174664979423863e-05,
+      "loss": 0.7164,
+      "step": 2760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617180706983458e-05,
+      "loss": 0.6365,
+      "step": 2761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168948345764743e-05,
+      "loss": 0.6876,
+      "step": 2762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166088807591612e-05,
+      "loss": 0.654,
+      "step": 2763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616322845569255e-05,
+      "loss": 0.7628,
+      "step": 2764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160367290445038e-05,
+      "loss": 0.7279,
+      "step": 2765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157505312226654e-05,
+      "loss": 0.7266,
+      "step": 2766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154642521415086e-05,
+      "loss": 0.696,
+      "step": 2767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151778918388138e-05,
+      "loss": 0.7152,
+      "step": 2768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148914503523708e-05,
+      "loss": 0.618,
+      "step": 2769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146049277199804e-05,
+      "loss": 0.6848,
+      "step": 2770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143183239794548e-05,
+      "loss": 0.7231,
+      "step": 2771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6140316391686167e-05,
+      "loss": 0.7596,
+      "step": 2772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613744873325299e-05,
+      "loss": 0.6867,
+      "step": 2773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134580264873455e-05,
+      "loss": 0.69,
+      "step": 2774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613171098692611e-05,
+      "loss": 0.7228,
+      "step": 2775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128840899789605e-05,
+      "loss": 0.7529,
+      "step": 2776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125970003842703e-05,
+      "loss": 0.7207,
+      "step": 2777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612309829946427e-05,
+      "loss": 0.5753,
+      "step": 2778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120225787033274e-05,
+      "loss": 0.6487,
+      "step": 2779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61173524669288e-05,
+      "loss": 0.797,
+      "step": 2780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114478339530032e-05,
+      "loss": 0.7001,
+      "step": 2781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611160340521626e-05,
+      "loss": 0.5073,
+      "step": 2782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108727664366888e-05,
+      "loss": 0.7506,
+      "step": 2783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105851117361416e-05,
+      "loss": 0.6557,
+      "step": 2784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102973764579464e-05,
+      "loss": 0.7278,
+      "step": 2785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100095606400745e-05,
+      "loss": 0.6991,
+      "step": 2786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609721664320508e-05,
+      "loss": 0.6485,
+      "step": 2787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094336875372408e-05,
+      "loss": 0.6649,
+      "step": 2788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609145630328276e-05,
+      "loss": 0.6173,
+      "step": 2789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088574927316278e-05,
+      "loss": 0.7215,
+      "step": 2790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085692747853218e-05,
+      "loss": 0.6816,
+      "step": 2791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082809765273928e-05,
+      "loss": 0.6977,
+      "step": 2792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079925979958875e-05,
+      "loss": 0.679,
+      "step": 2793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077041392288623e-05,
+      "loss": 0.7925,
+      "step": 2794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074156002643846e-05,
+      "loss": 0.6341,
+      "step": 2795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607126981140532e-05,
+      "loss": 0.6307,
+      "step": 2796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068382818953935e-05,
+      "loss": 0.7072,
+      "step": 2797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065495025670675e-05,
+      "loss": 0.7895,
+      "step": 2798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6062606431936646e-05,
+      "loss": 0.7205,
+      "step": 2799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059717038133038e-05,
+      "loss": 0.6899,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6056826844641163e-05,
+      "loss": 0.7274,
+      "step": 2801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053935851842436e-05,
+      "loss": 0.6976,
+      "step": 2802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6051044060118377e-05,
+      "loss": 0.6972,
+      "step": 2803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048151469850605e-05,
+      "loss": 0.6184,
+      "step": 2804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604525808142085e-05,
+      "loss": 0.6617,
+      "step": 2805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042363895210948e-05,
+      "loss": 0.678,
+      "step": 2806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039468911602835e-05,
+      "loss": 0.6995,
+      "step": 2807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036573130978562e-05,
+      "loss": 0.7375,
+      "step": 2808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033676553720274e-05,
+      "loss": 0.7418,
+      "step": 2809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603077918021023e-05,
+      "loss": 0.6937,
+      "step": 2810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027881010830788e-05,
+      "loss": 0.6773,
+      "step": 2811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024982045964413e-05,
+      "loss": 0.6667,
+      "step": 2812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602208228599368e-05,
+      "loss": 0.7374,
+      "step": 2813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6019181731301256e-05,
+      "loss": 0.6449,
+      "step": 2814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601628038226993e-05,
+      "loss": 0.603,
+      "step": 2815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013378239282577e-05,
+      "loss": 0.7375,
+      "step": 2816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60104753027222e-05,
+      "loss": 0.6913,
+      "step": 2817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600757157297188e-05,
+      "loss": 0.7319,
+      "step": 2818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004667050414828e-05,
+      "loss": 0.775,
+      "step": 2819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001761735434337e-05,
+      "loss": 0.6544,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998855628413822e-05,
+      "loss": 0.7411,
+      "step": 2821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5995948729736795e-05,
+      "loss": 0.7495,
+      "step": 2822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993041039786875e-05,
+      "loss": 0.6643,
+      "step": 2823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599013255894778e-05,
+      "loss": 0.7208,
+      "step": 2824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987223287603337e-05,
+      "loss": 0.6337,
+      "step": 2825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598431322613748e-05,
+      "loss": 0.7438,
+      "step": 2826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981402374934236e-05,
+      "loss": 0.8287,
+      "step": 2827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978490734377754e-05,
+      "loss": 0.6735,
+      "step": 2828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597557830485227e-05,
+      "loss": 0.5392,
+      "step": 2829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972665086742135e-05,
+      "loss": 0.6716,
+      "step": 2830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969751080431796e-05,
+      "loss": 0.6358,
+      "step": 2831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966836286305813e-05,
+      "loss": 0.7667,
+      "step": 2832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963920704748847e-05,
+      "loss": 0.6838,
+      "step": 2833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596100433614565e-05,
+      "loss": 0.7378,
+      "step": 2834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958087180881104e-05,
+      "loss": 0.6834,
+      "step": 2835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595516923934017e-05,
+      "loss": 0.6677,
+      "step": 2836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952250511907922e-05,
+      "loss": 0.6567,
+      "step": 2837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949330998969545e-05,
+      "loss": 0.7004,
+      "step": 2838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594641070091032e-05,
+      "loss": 0.6809,
+      "step": 2839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594348961811563e-05,
+      "loss": 0.7651,
+      "step": 2840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940567750970962e-05,
+      "loss": 0.6614,
+      "step": 2841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937645099861913e-05,
+      "loss": 0.7283,
+      "step": 2842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934721665174176e-05,
+      "loss": 0.6944,
+      "step": 2843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931797447293553e-05,
+      "loss": 0.7239,
+      "step": 2844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928872446605947e-05,
+      "loss": 0.6003,
+      "step": 2845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925946663497366e-05,
+      "loss": 0.6481,
+      "step": 2846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923020098353913e-05,
+      "loss": 0.7812,
+      "step": 2847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920092751561806e-05,
+      "loss": 0.6704,
+      "step": 2848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591716462350736e-05,
+      "loss": 0.6652,
+      "step": 2849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914235714576998e-05,
+      "loss": 0.6774,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591130602515723e-05,
+      "loss": 0.6679,
+      "step": 2851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590837555563469e-05,
+      "loss": 0.7257,
+      "step": 2852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590544430639611e-05,
+      "loss": 0.7554,
+      "step": 2853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5902512277828313e-05,
+      "loss": 0.8038,
+      "step": 2854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899579470318234e-05,
+      "loss": 0.653,
+      "step": 2855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896645884252913e-05,
+      "loss": 0.6518,
+      "step": 2856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5893711520019483e-05,
+      "loss": 0.7874,
+      "step": 2857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589077637800519e-05,
+      "loss": 0.6435,
+      "step": 2858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887840458597383e-05,
+      "loss": 0.6899,
+      "step": 2859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884903762183505e-05,
+      "loss": 0.6654,
+      "step": 2860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881966289151104e-05,
+      "loss": 0.7095,
+      "step": 2861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587902803988783e-05,
+      "loss": 0.7019,
+      "step": 2862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876089014781445e-05,
+      "loss": 0.6917,
+      "step": 2863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873149214219802e-05,
+      "loss": 0.6715,
+      "step": 2864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587020863859086e-05,
+      "loss": 0.6062,
+      "step": 2865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5867267288282682e-05,
+      "loss": 0.7158,
+      "step": 2866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586432516368343e-05,
+      "loss": 0.6854,
+      "step": 2867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586138226518137e-05,
+      "loss": 0.5906,
+      "step": 2868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585843859316488e-05,
+      "loss": 0.6412,
+      "step": 2869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855494148022414e-05,
+      "loss": 0.7239,
+      "step": 2870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852548930142552e-05,
+      "loss": 0.709,
+      "step": 2871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849602939913973e-05,
+      "loss": 0.6033,
+      "step": 2872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846656177725447e-05,
+      "loss": 0.6728,
+      "step": 2873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584370864396585e-05,
+      "loss": 0.707,
+      "step": 2874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5840760339024168e-05,
+      "loss": 0.6784,
+      "step": 2875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583781126328948e-05,
+      "loss": 0.7813,
+      "step": 2876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834861417150973e-05,
+      "loss": 0.6992,
+      "step": 2877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831910800997924e-05,
+      "loss": 0.6434,
+      "step": 2878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828959415219727e-05,
+      "loss": 0.6639,
+      "step": 2879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5826007260205868e-05,
+      "loss": 0.5913,
+      "step": 2880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823054336345936e-05,
+      "loss": 0.6995,
+      "step": 2881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582010064402962e-05,
+      "loss": 0.5651,
+      "step": 2882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817146183646718e-05,
+      "loss": 0.6595,
+      "step": 2883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814190955587126e-05,
+      "loss": 0.6545,
+      "step": 2884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811234960240827e-05,
+      "loss": 0.7129,
+      "step": 2885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808278197997928e-05,
+      "loss": 0.7338,
+      "step": 2886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805320669248627e-05,
+      "loss": 0.7378,
+      "step": 2887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802362374383212e-05,
+      "loss": 0.6354,
+      "step": 2888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799403313792098e-05,
+      "loss": 0.6506,
+      "step": 2889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796443487865774e-05,
+      "loss": 0.6805,
+      "step": 2890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579348289699485e-05,
+      "loss": 0.7616,
+      "step": 2891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579052154157003e-05,
+      "loss": 0.6277,
+      "step": 2892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578755942198211e-05,
+      "loss": 0.7294,
+      "step": 2893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5784596538621997e-05,
+      "loss": 0.713,
+      "step": 2894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781632891880698e-05,
+      "loss": 0.7398,
+      "step": 2895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778668482149325e-05,
+      "loss": 0.7379,
+      "step": 2896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775703309819075e-05,
+      "loss": 0.8199,
+      "step": 2897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577273737528126e-05,
+      "loss": 0.6036,
+      "step": 2898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769770678927293e-05,
+      "loss": 0.6002,
+      "step": 2899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766803221148676e-05,
+      "loss": 0.729,
+      "step": 2900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5763835002337017e-05,
+      "loss": 0.7448,
+      "step": 2901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760866022884032e-05,
+      "loss": 0.6906,
+      "step": 2902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757896283181528e-05,
+      "loss": 0.7517,
+      "step": 2903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754925783621417e-05,
+      "loss": 0.6405,
+      "step": 2904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751954524595705e-05,
+      "loss": 0.6593,
+      "step": 2905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5748982506496505e-05,
+      "loss": 0.7262,
+      "step": 2906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746009729716037e-05,
+      "loss": 0.7155,
+      "step": 2907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57430361946466e-05,
+      "loss": 0.7869,
+      "step": 2908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740061901680608e-05,
+      "loss": 0.7382,
+      "step": 2909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737086851210575e-05,
+      "loss": 0.6369,
+      "step": 2910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573411104362911e-05,
+      "loss": 0.7489,
+      "step": 2911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731134479328927e-05,
+      "loss": 0.614,
+      "step": 2912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5728157158702836e-05,
+      "loss": 0.7182,
+      "step": 2913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725179082143743e-05,
+      "loss": 0.7418,
+      "step": 2914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722200250044665e-05,
+      "loss": 0.6616,
+      "step": 2915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571922066279871e-05,
+      "loss": 0.7209,
+      "step": 2916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571624032079909e-05,
+      "loss": 0.6507,
+      "step": 2917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571325922443911e-05,
+      "loss": 0.7594,
+      "step": 2918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571027737411218e-05,
+      "loss": 0.7032,
+      "step": 2919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570729477021181e-05,
+      "loss": 0.6149,
+      "step": 2920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570431141313161e-05,
+      "loss": 0.6409,
+      "step": 2921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570132730326529e-05,
+      "loss": 0.729,
+      "step": 2922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698342441006644e-05,
+      "loss": 0.6359,
+      "step": 2923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569535682674959e-05,
+      "loss": 0.7704,
+      "step": 2924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569237046088813e-05,
+      "loss": 0.7452,
+      "step": 2925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689383343816368e-05,
+      "loss": 0.7916,
+      "step": 2926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568639547592851e-05,
+      "loss": 0.6975,
+      "step": 2927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683406857618853e-05,
+      "loss": 0.6774,
+      "step": 2928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5680417489281802e-05,
+      "loss": 0.7413,
+      "step": 2929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567742737131186e-05,
+      "loss": 0.6933,
+      "step": 2930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5674436504103624e-05,
+      "loss": 0.7188,
+      "step": 2931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671444888051796e-05,
+      "loss": 0.6493,
+      "step": 2932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668452523551166e-05,
+      "loss": 0.7498,
+      "step": 2933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665459410996638e-05,
+      "loss": 0.6634,
+      "step": 2934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56624655507832e-05,
+      "loss": 0.7581,
+      "step": 2935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659470943305956e-05,
+      "loss": 0.6222,
+      "step": 2936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656475588960084e-05,
+      "loss": 0.5648,
+      "step": 2937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653479488140884e-05,
+      "loss": 0.7001,
+      "step": 2938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650482641243744e-05,
+      "loss": 0.6046,
+      "step": 2939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647485048664148e-05,
+      "loss": 0.5989,
+      "step": 2940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644486710797687e-05,
+      "loss": 0.6191,
+      "step": 2941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641487628040043e-05,
+      "loss": 0.639,
+      "step": 2942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638487800787002e-05,
+      "loss": 0.7347,
+      "step": 2943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635487229434436e-05,
+      "loss": 0.7471,
+      "step": 2944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632485914378333e-05,
+      "loss": 0.7678,
+      "step": 2945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629483856014766e-05,
+      "loss": 0.654,
+      "step": 2946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5626481054739916e-05,
+      "loss": 0.663,
+      "step": 2947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623477510950048e-05,
+      "loss": 0.7556,
+      "step": 2948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562047322504154e-05,
+      "loss": 0.7474,
+      "step": 2949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617468197410857e-05,
+      "loss": 0.6799,
+      "step": 2950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614462428454568e-05,
+      "loss": 0.7276,
+      "step": 2951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561145591856934e-05,
+      "loss": 0.693,
+      "step": 2952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608448668151934e-05,
+      "loss": 0.717,
+      "step": 2953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560544067759921e-05,
+      "loss": 0.6786,
+      "step": 2954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560243194730813e-05,
+      "loss": 0.6523,
+      "step": 2955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599422477675742e-05,
+      "loss": 0.6023,
+      "step": 2956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596412269099208e-05,
+      "loss": 0.6953,
+      "step": 2957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593401321975775e-05,
+      "loss": 0.6929,
+      "step": 2958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559038963670279e-05,
+      "loss": 0.7748,
+      "step": 2959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587377213677705e-05,
+      "loss": 0.6946,
+      "step": 2960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584364053298055e-05,
+      "loss": 0.5815,
+      "step": 2961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5581350155961486e-05,
+      "loss": 0.7124,
+      "step": 2962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557833552206573e-05,
+      "loss": 0.708,
+      "step": 2963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575320152008632e-05,
+      "loss": 0.7365,
+      "step": 2964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572304046188115e-05,
+      "loss": 0.6657,
+      "step": 2965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569287205002212e-05,
+      "loss": 0.795,
+      "step": 2966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556626962884905e-05,
+      "loss": 0.5808,
+      "step": 2967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563251318126845e-05,
+      "loss": 0.6314,
+      "step": 2968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560232273233928e-05,
+      "loss": 0.7263,
+      "step": 2969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557212494568715e-05,
+      "loss": 0.7218,
+      "step": 2970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555419198252971e-05,
+      "loss": 0.6777,
+      "step": 2971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551170737515533e-05,
+      "loss": 0.6864,
+      "step": 2972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5548148759924884e-05,
+      "loss": 0.7469,
+      "step": 2973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545126050156577e-05,
+      "loss": 0.6955,
+      "step": 2974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542102608609507e-05,
+      "loss": 0.6267,
+      "step": 2975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539078435682667e-05,
+      "loss": 0.7098,
+      "step": 2976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536053531775158e-05,
+      "loss": 0.7146,
+      "step": 2977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533027897286168e-05,
+      "loss": 0.7058,
+      "step": 2978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530001532614983e-05,
+      "loss": 0.7705,
+      "step": 2979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526974438160983e-05,
+      "loss": 0.6953,
+      "step": 2980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523946614323653e-05,
+      "loss": 0.6762,
+      "step": 2981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552091806150257e-05,
+      "loss": 0.6213,
+      "step": 2982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517888780097395e-05,
+      "loss": 0.6429,
+      "step": 2983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551485877050791e-05,
+      "loss": 0.7125,
+      "step": 2984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511828033133967e-05,
+      "loss": 0.6557,
+      "step": 2985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508796568375535e-05,
+      "loss": 0.6985,
+      "step": 2986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5505764376632663e-05,
+      "loss": 0.7163,
+      "step": 2987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502731458305507e-05,
+      "loss": 0.7405,
+      "step": 2988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499697813794316e-05,
+      "loss": 0.6464,
+      "step": 2989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549666344349943e-05,
+      "loss": 0.6959,
+      "step": 2990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493628347821293e-05,
+      "loss": 0.6878,
+      "step": 2991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490592527160437e-05,
+      "loss": 0.7281,
+      "step": 2992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487555981917493e-05,
+      "loss": 0.6084,
+      "step": 2993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5484518712493188e-05,
+      "loss": 0.7073,
+      "step": 2994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481480719288347e-05,
+      "loss": 0.6879,
+      "step": 2995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478442002703884e-05,
+      "loss": 0.6825,
+      "step": 2996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475402563140812e-05,
+      "loss": 0.7203,
+      "step": 2997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547236240100024e-05,
+      "loss": 0.6636,
+      "step": 2998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469321516683374e-05,
+      "loss": 0.7026,
+      "step": 2999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466279910591512e-05,
+      "loss": 0.6885,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463237583126052e-05,
+      "loss": 0.7965,
+      "step": 3001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546019453468848e-05,
+      "loss": 0.6471,
+      "step": 3002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457150765680374e-05,
+      "loss": 0.5962,
+      "step": 3003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454106276503424e-05,
+      "loss": 0.744,
+      "step": 3004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54510610675594e-05,
+      "loss": 0.6354,
+      "step": 3005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448015139250176e-05,
+      "loss": 0.6019,
+      "step": 3006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444968491977713e-05,
+      "loss": 0.6238,
+      "step": 3007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544192112614407e-05,
+      "loss": 0.7054,
+      "step": 3008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543887304215141e-05,
+      "loss": 0.7283,
+      "step": 3009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543582424040197e-05,
+      "loss": 0.6136,
+      "step": 3010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432774721298104e-05,
+      "loss": 0.6613,
+      "step": 3011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429724485242243e-05,
+      "loss": 0.7135,
+      "step": 3012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542667353263692e-05,
+      "loss": 0.6852,
+      "step": 3013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423621863884776e-05,
+      "loss": 0.6326,
+      "step": 3014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420569479388516e-05,
+      "loss": 0.6819,
+      "step": 3015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541751637955097e-05,
+      "loss": 0.7095,
+      "step": 3016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414462564775042e-05,
+      "loss": 0.6622,
+      "step": 3017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411408035463743e-05,
+      "loss": 0.656,
+      "step": 3018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408352792020168e-05,
+      "loss": 0.6369,
+      "step": 3019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540529683484751e-05,
+      "loss": 0.5475,
+      "step": 3020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5402240164349063e-05,
+      "loss": 0.6715,
+      "step": 3021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399182780928206e-05,
+      "loss": 0.686,
+      "step": 3022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396124684988415e-05,
+      "loss": 0.6508,
+      "step": 3023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393065876933262e-05,
+      "loss": 0.6626,
+      "step": 3024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539000635716641e-05,
+      "loss": 0.5927,
+      "step": 3025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538694612609162e-05,
+      "loss": 0.6194,
+      "step": 3026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383885184112748e-05,
+      "loss": 0.6692,
+      "step": 3027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380823531633727e-05,
+      "loss": 0.6743,
+      "step": 3028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377761169058612e-05,
+      "loss": 0.6989,
+      "step": 3029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374698096791524e-05,
+      "loss": 0.7458,
+      "step": 3030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371634315236698e-05,
+      "loss": 0.6864,
+      "step": 3031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368569824798453e-05,
+      "loss": 0.6045,
+      "step": 3032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365504625881205e-05,
+      "loss": 0.5871,
+      "step": 3033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362438718889457e-05,
+      "loss": 0.6877,
+      "step": 3034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359372104227815e-05,
+      "loss": 0.6992,
+      "step": 3035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356304782300977e-05,
+      "loss": 0.7378,
+      "step": 3036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535323675351372e-05,
+      "loss": 0.7007,
+      "step": 3037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350168018270934e-05,
+      "loss": 0.6843,
+      "step": 3038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347098576977592e-05,
+      "loss": 0.6291,
+      "step": 3039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344028430038764e-05,
+      "loss": 0.6961,
+      "step": 3040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340957577859605e-05,
+      "loss": 0.6594,
+      "step": 3041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533788602084537e-05,
+      "loss": 0.6802,
+      "step": 3042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533481375940141e-05,
+      "loss": 0.7776,
+      "step": 3043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331740793933164e-05,
+      "loss": 0.6477,
+      "step": 3044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328667124846166e-05,
+      "loss": 0.6735,
+      "step": 3045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532559275254604e-05,
+      "loss": 0.6371,
+      "step": 3046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53225176774385e-05,
+      "loss": 0.6304,
+      "step": 3047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531944189992936e-05,
+      "loss": 0.7047,
+      "step": 3048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531636542042453e-05,
+      "loss": 0.7131,
+      "step": 3049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531328823933e-05,
+      "loss": 0.742,
+      "step": 3050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310210357051864e-05,
+      "loss": 0.6709,
+      "step": 3051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307131773996297e-05,
+      "loss": 0.7678,
+      "step": 3052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530405249056958e-05,
+      "loss": 0.6987,
+      "step": 3053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530097250717808e-05,
+      "loss": 0.7205,
+      "step": 3054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529789182422825e-05,
+      "loss": 0.6977,
+      "step": 3055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529481044212664e-05,
+      "loss": 0.727,
+      "step": 3056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52917283612799e-05,
+      "loss": 0.6166,
+      "step": 3057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288645582094764e-05,
+      "loss": 0.565,
+      "step": 3058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285562104978063e-05,
+      "loss": 0.7407,
+      "step": 3059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528247793033671e-05,
+      "loss": 0.6744,
+      "step": 3060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527939305857772e-05,
+      "loss": 0.6721,
+      "step": 3061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52763074901082e-05,
+      "loss": 0.7894,
+      "step": 3062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273221225335344e-05,
+      "loss": 0.6469,
+      "step": 3063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527013426466644e-05,
+      "loss": 0.7013,
+      "step": 3064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267046608508866e-05,
+      "loss": 0.7319,
+      "step": 3065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5263958257270096e-05,
+      "loss": 0.7252,
+      "step": 3066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260869211357695e-05,
+      "loss": 0.7576,
+      "step": 3067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257779471179313e-05,
+      "loss": 0.7691,
+      "step": 3068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254689037142698e-05,
+      "loss": 0.816,
+      "step": 3069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525159790965569e-05,
+      "loss": 0.6358,
+      "step": 3070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524850608912622e-05,
+      "loss": 0.6852,
+      "step": 3071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245413575962306e-05,
+      "loss": 0.5677,
+      "step": 3072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242320370572062e-05,
+      "loss": 0.74,
+      "step": 3073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523922647336369e-05,
+      "loss": 0.6433,
+      "step": 3074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236131884745486e-05,
+      "loss": 0.7382,
+      "step": 3075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523303660512584e-05,
+      "loss": 0.6395,
+      "step": 3076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229940634913223e-05,
+      "loss": 0.689,
+      "step": 3077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522684397451621e-05,
+      "loss": 0.687,
+      "step": 3078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223746624343458e-05,
+      "loss": 0.6407,
+      "step": 3079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220648584803721e-05,
+      "loss": 0.6317,
+      "step": 3080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217549856305836e-05,
+      "loss": 0.7235,
+      "step": 3081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214450439258731e-05,
+      "loss": 0.696,
+      "step": 3082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211350334071444e-05,
+      "loss": 0.6303,
+      "step": 3083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208249541153079e-05,
+      "loss": 0.7004,
+      "step": 3084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205148060912847e-05,
+      "loss": 0.6261,
+      "step": 3085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202045893760037e-05,
+      "loss": 0.6824,
+      "step": 3086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198943040104044e-05,
+      "loss": 0.7687,
+      "step": 3087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195839500354337e-05,
+      "loss": 0.6733,
+      "step": 3088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519273527492049e-05,
+      "loss": 0.7024,
+      "step": 3089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189630364212158e-05,
+      "loss": 0.5988,
+      "step": 3090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186524768639087e-05,
+      "loss": 0.6936,
+      "step": 3091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183418488611122e-05,
+      "loss": 0.6988,
+      "step": 3092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180311524538188e-05,
+      "loss": 0.6859,
+      "step": 3093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177203876830307e-05,
+      "loss": 0.6333,
+      "step": 3094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174095545897587e-05,
+      "loss": 0.7799,
+      "step": 3095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170986532150229e-05,
+      "loss": 0.7477,
+      "step": 3096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167876835998522e-05,
+      "loss": 0.6507,
+      "step": 3097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516476645785285e-05,
+      "loss": 0.6403,
+      "step": 3098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161655398123677e-05,
+      "loss": 0.7513,
+      "step": 3099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158543657221571e-05,
+      "loss": 0.7343,
+      "step": 3100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155431235557171e-05,
+      "loss": 0.5836,
+      "step": 3101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152318133541225e-05,
+      "loss": 0.7427,
+      "step": 3102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149204351584564e-05,
+      "loss": 0.6326,
+      "step": 3103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146089890098102e-05,
+      "loss": 0.6395,
+      "step": 3104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142974749492849e-05,
+      "loss": 0.6209,
+      "step": 3105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139858930179908e-05,
+      "loss": 0.6396,
+      "step": 3106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513674243257046e-05,
+      "loss": 0.7307,
+      "step": 3107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133625257075787e-05,
+      "loss": 0.8065,
+      "step": 3108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513050740410726e-05,
+      "loss": 0.6637,
+      "step": 3109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127388874076325e-05,
+      "loss": 0.7098,
+      "step": 3110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124269667394536e-05,
+      "loss": 0.7123,
+      "step": 3111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121149784473527e-05,
+      "loss": 0.7222,
+      "step": 3112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118029225725023e-05,
+      "loss": 0.778,
+      "step": 3113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114907991560838e-05,
+      "loss": 0.6893,
+      "step": 3114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111786082392872e-05,
+      "loss": 0.699,
+      "step": 3115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108663498633118e-05,
+      "loss": 0.6972,
+      "step": 3116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105540240693657e-05,
+      "loss": 0.6552,
+      "step": 3117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510241630898666e-05,
+      "loss": 0.6721,
+      "step": 3118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099291703924381e-05,
+      "loss": 0.6631,
+      "step": 3119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096166425919176e-05,
+      "loss": 0.748,
+      "step": 3120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093040475383474e-05,
+      "loss": 0.6528,
+      "step": 3121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089913852729803e-05,
+      "loss": 0.6983,
+      "step": 3122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086786558370781e-05,
+      "loss": 0.6675,
+      "step": 3123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083658592719104e-05,
+      "loss": 0.7207,
+      "step": 3124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080529956187566e-05,
+      "loss": 0.688,
+      "step": 3125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077400649189042e-05,
+      "loss": 0.7089,
+      "step": 3126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507427067213651e-05,
+      "loss": 0.6428,
+      "step": 3127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071140025443018e-05,
+      "loss": 0.7015,
+      "step": 3128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068008709521716e-05,
+      "loss": 0.7494,
+      "step": 3129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064876724785835e-05,
+      "loss": 0.7278,
+      "step": 3130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061744071648691e-05,
+      "loss": 0.7264,
+      "step": 3131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058610750523705e-05,
+      "loss": 0.7196,
+      "step": 3132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055476761824373e-05,
+      "loss": 0.6167,
+      "step": 3133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052342105964272e-05,
+      "loss": 0.5479,
+      "step": 3134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049206783357082e-05,
+      "loss": 0.6831,
+      "step": 3135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046070794416563e-05,
+      "loss": 0.7049,
+      "step": 3136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042934139556567e-05,
+      "loss": 0.6953,
+      "step": 3137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5039796819191032e-05,
+      "loss": 0.7492,
+      "step": 3138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036658833733984e-05,
+      "loss": 0.7491,
+      "step": 3139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033520183599532e-05,
+      "loss": 0.7761,
+      "step": 3140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030380869201881e-05,
+      "loss": 0.6016,
+      "step": 3141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502724089095532e-05,
+      "loss": 0.7168,
+      "step": 3142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024100249274228e-05,
+      "loss": 0.7476,
+      "step": 3143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5020958944573061e-05,
+      "loss": 0.6924,
+      "step": 3144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017816977266373e-05,
+      "loss": 0.6381,
+      "step": 3145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501467434776881e-05,
+      "loss": 0.7875,
+      "step": 3146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011531056495088e-05,
+      "loss": 0.6786,
+      "step": 3147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008387103860029e-05,
+      "loss": 0.7302,
+      "step": 3148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005242490278531e-05,
+      "loss": 0.6576,
+      "step": 3149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002097216165581e-05,
+      "loss": 0.6647,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998951281936254e-05,
+      "loss": 0.727,
+      "step": 3151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995804688005715e-05,
+      "loss": 0.7277,
+      "step": 3152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499265743478921e-05,
+      "loss": 0.7083,
+      "step": 3153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498950952270208e-05,
+      "loss": 0.6221,
+      "step": 3154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986360952159743e-05,
+      "loss": 0.7116,
+      "step": 3155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983211723577711e-05,
+      "loss": 0.7077,
+      "step": 3156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980061837371587e-05,
+      "loss": 0.7439,
+      "step": 3157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976911293957045e-05,
+      "loss": 0.7713,
+      "step": 3158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973760093749867e-05,
+      "loss": 0.6142,
+      "step": 3159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970608237165899e-05,
+      "loss": 0.6685,
+      "step": 3160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967455724621091e-05,
+      "loss": 0.6765,
+      "step": 3161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964302556531474e-05,
+      "loss": 0.6899,
+      "step": 3162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961148733313164e-05,
+      "loss": 0.6668,
+      "step": 3163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957994255382361e-05,
+      "loss": 0.723,
+      "step": 3164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495483912315536e-05,
+      "loss": 0.6744,
+      "step": 3165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951683337048536e-05,
+      "loss": 0.6911,
+      "step": 3166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948526897478352e-05,
+      "loss": 0.7207,
+      "step": 3167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945369804861357e-05,
+      "loss": 0.7077,
+      "step": 3168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494221205961418e-05,
+      "loss": 0.7724,
+      "step": 3169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939053662153548e-05,
+      "loss": 0.7778,
+      "step": 3170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935894612896267e-05,
+      "loss": 0.7666,
+      "step": 3171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932734912259233e-05,
+      "loss": 0.7026,
+      "step": 3172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929574560659416e-05,
+      "loss": 0.6711,
+      "step": 3173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926413558513892e-05,
+      "loss": 0.7185,
+      "step": 3174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923251906239804e-05,
+      "loss": 0.708,
+      "step": 3175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920089604254395e-05,
+      "loss": 0.6607,
+      "step": 3176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916926652974984e-05,
+      "loss": 0.6241,
+      "step": 3177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913763052818979e-05,
+      "loss": 0.6897,
+      "step": 3178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910598804203874e-05,
+      "loss": 0.7159,
+      "step": 3179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907433907547243e-05,
+      "loss": 0.6419,
+      "step": 3180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904268363266762e-05,
+      "loss": 0.6896,
+      "step": 3181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901102171780175e-05,
+      "loss": 0.6931,
+      "step": 3182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897935333505316e-05,
+      "loss": 0.556,
+      "step": 3183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489476784886011e-05,
+      "loss": 0.7452,
+      "step": 3184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891599718262559e-05,
+      "loss": 0.7165,
+      "step": 3185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888430942130758e-05,
+      "loss": 0.7158,
+      "step": 3186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885261520882885e-05,
+      "loss": 0.6203,
+      "step": 3187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882091454937201e-05,
+      "loss": 0.6713,
+      "step": 3188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487892074471205e-05,
+      "loss": 0.6482,
+      "step": 3189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875749390625864e-05,
+      "loss": 0.7797,
+      "step": 3190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872577393097165e-05,
+      "loss": 0.5723,
+      "step": 3191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486940475254455e-05,
+      "loss": 0.7553,
+      "step": 3192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486623146938671e-05,
+      "loss": 0.6686,
+      "step": 3193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863057544042412e-05,
+      "loss": 0.6974,
+      "step": 3194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859882976930516e-05,
+      "loss": 0.6543,
+      "step": 3195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485670776846996e-05,
+      "loss": 0.71,
+      "step": 3196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853531919079775e-05,
+      "loss": 0.788,
+      "step": 3197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850355429179066e-05,
+      "loss": 0.6628,
+      "step": 3198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847178299187027e-05,
+      "loss": 0.6406,
+      "step": 3199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844000529522942e-05,
+      "loss": 0.7284,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840822120606169e-05,
+      "loss": 0.7,
+      "step": 3201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837643072856163e-05,
+      "loss": 0.6477,
+      "step": 3202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483446338669245e-05,
+      "loss": 0.6715,
+      "step": 3203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483128306253465e-05,
+      "loss": 0.627,
+      "step": 3204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828102100802464e-05,
+      "loss": 0.7109,
+      "step": 3205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824920501915676e-05,
+      "loss": 0.6918,
+      "step": 3206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821738266294155e-05,
+      "loss": 0.7893,
+      "step": 3207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818555394357852e-05,
+      "loss": 0.6886,
+      "step": 3208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815371886526807e-05,
+      "loss": 0.6655,
+      "step": 3209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812187743221138e-05,
+      "loss": 0.6471,
+      "step": 3210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809002964861052e-05,
+      "loss": 0.7729,
+      "step": 3211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805817551866839e-05,
+      "loss": 0.6368,
+      "step": 3212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4802631504658868e-05,
+      "loss": 0.6121,
+      "step": 3213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799444823657596e-05,
+      "loss": 0.6253,
+      "step": 3214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796257509283567e-05,
+      "loss": 0.6667,
+      "step": 3215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793069561957398e-05,
+      "loss": 0.6962,
+      "step": 3216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789880982099799e-05,
+      "loss": 0.728,
+      "step": 3217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786691770131557e-05,
+      "loss": 0.744,
+      "step": 3218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478350192647355e-05,
+      "loss": 0.6778,
+      "step": 3219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478031145154673e-05,
+      "loss": 0.6691,
+      "step": 3220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777120345772145e-05,
+      "loss": 0.669,
+      "step": 3221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477392860957091e-05,
+      "loss": 0.6792,
+      "step": 3222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477073624336424e-05,
+      "loss": 0.6554,
+      "step": 3223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767543247573419e-05,
+      "loss": 0.6794,
+      "step": 3224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764349622619822e-05,
+      "loss": 0.6937,
+      "step": 3225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761155368924906e-05,
+      "loss": 0.5875,
+      "step": 3226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757960486910206e-05,
+      "loss": 0.7146,
+      "step": 3227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754764976997349e-05,
+      "loss": 0.7736,
+      "step": 3228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751568839608036e-05,
+      "loss": 0.7267,
+      "step": 3229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474837207516406e-05,
+      "loss": 0.7375,
+      "step": 3230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745174684087284e-05,
+      "loss": 0.6379,
+      "step": 3231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741976666799667e-05,
+      "loss": 0.685,
+      "step": 3232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738778023723243e-05,
+      "loss": 0.674,
+      "step": 3233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735578755280126e-05,
+      "loss": 0.6904,
+      "step": 3234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732378861892524e-05,
+      "loss": 0.7448,
+      "step": 3235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729178343982717e-05,
+      "loss": 0.6943,
+      "step": 3236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725977201973068e-05,
+      "loss": 0.6712,
+      "step": 3237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722775436286028e-05,
+      "loss": 0.7679,
+      "step": 3238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719573047344127e-05,
+      "loss": 0.7058,
+      "step": 3239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716370035569978e-05,
+      "loss": 0.7315,
+      "step": 3240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713166401386276e-05,
+      "loss": 0.7204,
+      "step": 3241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709962145215797e-05,
+      "loss": 0.7288,
+      "step": 3242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706757267481399e-05,
+      "loss": 0.6804,
+      "step": 3243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703551768606026e-05,
+      "loss": 0.5965,
+      "step": 3244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700345649012699e-05,
+      "loss": 0.7655,
+      "step": 3245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697138909124522e-05,
+      "loss": 0.6075,
+      "step": 3246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693931549364683e-05,
+      "loss": 0.7101,
+      "step": 3247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690723570156453e-05,
+      "loss": 0.5495,
+      "step": 3248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687514971923178e-05,
+      "loss": 0.6462,
+      "step": 3249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684305755088295e-05,
+      "loss": 0.7058,
+      "step": 3250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681095920075313e-05,
+      "loss": 0.6439,
+      "step": 3251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677885467307829e-05,
+      "loss": 0.5285,
+      "step": 3252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467467439720952e-05,
+      "loss": 0.6735,
+      "step": 3253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671462710204142e-05,
+      "loss": 0.7289,
+      "step": 3254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466825040671554e-05,
+      "loss": 0.7463,
+      "step": 3255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665037487167629e-05,
+      "loss": 0.735,
+      "step": 3256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661823951984418e-05,
+      "loss": 0.6217,
+      "step": 3257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658609801589982e-05,
+      "loss": 0.6443,
+      "step": 3258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4655395036408493e-05,
+      "loss": 0.6568,
+      "step": 3259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652179656864194e-05,
+      "loss": 0.6384,
+      "step": 3260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648963663381412e-05,
+      "loss": 0.724,
+      "step": 3261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645747056384555e-05,
+      "loss": 0.6552,
+      "step": 3262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642529836298111e-05,
+      "loss": 0.842,
+      "step": 3263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639312003546652e-05,
+      "loss": 0.7233,
+      "step": 3264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463609355855483e-05,
+      "loss": 0.6905,
+      "step": 3265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632874501747372e-05,
+      "loss": 0.7103,
+      "step": 3266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629654833549093e-05,
+      "loss": 0.6927,
+      "step": 3267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626434554384882e-05,
+      "loss": 0.7565,
+      "step": 3268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623213664679723e-05,
+      "loss": 0.8172,
+      "step": 3269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619992164858658e-05,
+      "loss": 0.6929,
+      "step": 3270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461677005534683e-05,
+      "loss": 0.6448,
+      "step": 3271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613547336569446e-05,
+      "loss": 0.583,
+      "step": 3272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461032400895181e-05,
+      "loss": 0.6796,
+      "step": 3273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460710007291929e-05,
+      "loss": 0.7046,
+      "step": 3274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603875528897351e-05,
+      "loss": 0.7032,
+      "step": 3275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600650377311523e-05,
+      "loss": 0.7156,
+      "step": 3276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597424618587424e-05,
+      "loss": 0.7201,
+      "step": 3277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594198253150751e-05,
+      "loss": 0.8081,
+      "step": 3278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459097128142728e-05,
+      "loss": 0.6964,
+      "step": 3279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458774370384287e-05,
+      "loss": 0.7366,
+      "step": 3280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584515520823453e-05,
+      "loss": 0.7629,
+      "step": 3281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458128673279505e-05,
+      "loss": 0.5762,
+      "step": 3282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578057340183755e-05,
+      "loss": 0.5862,
+      "step": 3283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457482734341575e-05,
+      "loss": 0.7434,
+      "step": 3284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571596742917282e-05,
+      "loss": 0.7512,
+      "step": 3285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568365539114692e-05,
+      "loss": 0.6996,
+      "step": 3286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565133732434396e-05,
+      "loss": 0.711,
+      "step": 3287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561901323302886e-05,
+      "loss": 0.747,
+      "step": 3288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558668312146736e-05,
+      "loss": 0.716,
+      "step": 3289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555434699392603e-05,
+      "loss": 0.7363,
+      "step": 3290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552200485467216e-05,
+      "loss": 0.6779,
+      "step": 3291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454896567079739e-05,
+      "loss": 0.6959,
+      "step": 3292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545730255810017e-05,
+      "loss": 0.6851,
+      "step": 3293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542494240932068e-05,
+      "loss": 0.5991,
+      "step": 3294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539257626590592e-05,
+      "loss": 0.6316,
+      "step": 3295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536020413212722e-05,
+      "loss": 0.6658,
+      "step": 3296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532782601225661e-05,
+      "loss": 0.793,
+      "step": 3297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529544191056696e-05,
+      "loss": 0.6983,
+      "step": 3298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526305183133201e-05,
+      "loss": 0.7085,
+      "step": 3299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523065577882612e-05,
+      "loss": 0.684,
+      "step": 3300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519825375732456e-05,
+      "loss": 0.6636,
+      "step": 3301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516584577110339e-05,
+      "loss": 0.6709,
+      "step": 3302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513343182443939e-05,
+      "loss": 0.8008,
+      "step": 3303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451010119216102e-05,
+      "loss": 0.6661,
+      "step": 3304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506858606689415e-05,
+      "loss": 0.7122,
+      "step": 3305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503615426457047e-05,
+      "loss": 0.7571,
+      "step": 3306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450037165189191e-05,
+      "loss": 0.6561,
+      "step": 3307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497127283422072e-05,
+      "loss": 0.7595,
+      "step": 3308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493882321475691e-05,
+      "loss": 0.7299,
+      "step": 3309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490636766481002e-05,
+      "loss": 0.6405,
+      "step": 3310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487390618866309e-05,
+      "loss": 0.7005,
+      "step": 3311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484143879059998e-05,
+      "loss": 0.6239,
+      "step": 3312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480896547490538e-05,
+      "loss": 0.6606,
+      "step": 3313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447764862458647e-05,
+      "loss": 0.71,
+      "step": 3314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447440011077642e-05,
+      "loss": 0.7365,
+      "step": 3315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471151006489084e-05,
+      "loss": 0.6921,
+      "step": 3316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4467901312153236e-05,
+      "loss": 0.6656,
+      "step": 3317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464651028197739e-05,
+      "loss": 0.6155,
+      "step": 3318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446140015505152e-05,
+      "loss": 0.6948,
+      "step": 3319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458148693143596e-05,
+      "loss": 0.7111,
+      "step": 3320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454896642903052e-05,
+      "loss": 0.6869,
+      "step": 3321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445164400475905e-05,
+      "loss": 0.6546,
+      "step": 3322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448390779140844e-05,
+      "loss": 0.674,
+      "step": 3323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445136966477749e-05,
+      "loss": 0.6134,
+      "step": 3324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441882567199164e-05,
+      "loss": 0.7434,
+      "step": 3325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438627581734566e-05,
+      "loss": 0.7213,
+      "step": 3326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443537201051351e-05,
+      "loss": 0.6312,
+      "step": 3327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432115853965622e-05,
+      "loss": 0.6275,
+      "step": 3328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428859112520617e-05,
+      "loss": 0.6764,
+      "step": 3329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425601786608279e-05,
+      "loss": 0.7003,
+      "step": 3330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422343876658467e-05,
+      "loss": 0.7035,
+      "step": 3331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419085383101121e-05,
+      "loss": 0.722,
+      "step": 3332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415826306366263e-05,
+      "loss": 0.7491,
+      "step": 3333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412566646883982e-05,
+      "loss": 0.6821,
+      "step": 3334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409306405084445e-05,
+      "loss": 0.6257,
+      "step": 3335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406045581397908e-05,
+      "loss": 0.6433,
+      "step": 3336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402784176254686e-05,
+      "loss": 0.5957,
+      "step": 3337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439952219008519e-05,
+      "loss": 0.7499,
+      "step": 3338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439625962331989e-05,
+      "loss": 0.5996,
+      "step": 3339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392996476389344e-05,
+      "loss": 0.5924,
+      "step": 3340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438973274972418e-05,
+      "loss": 0.7195,
+      "step": 3341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386468443755104e-05,
+      "loss": 0.6742,
+      "step": 3342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383203558912904e-05,
+      "loss": 0.642,
+      "step": 3343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379938095628439e-05,
+      "loss": 0.6376,
+      "step": 3344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376672054332641e-05,
+      "loss": 0.686,
+      "step": 3345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373405435456529e-05,
+      "loss": 0.6999,
+      "step": 3346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370138239431187e-05,
+      "loss": 0.7518,
+      "step": 3347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4366870466687784e-05,
+      "loss": 0.5816,
+      "step": 3348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363602117657558e-05,
+      "loss": 0.6484,
+      "step": 3349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436033319277183e-05,
+      "loss": 0.7443,
+      "step": 3350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357063692461988e-05,
+      "loss": 0.7316,
+      "step": 3351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353793617159504e-05,
+      "loss": 0.7785,
+      "step": 3352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350522967295923e-05,
+      "loss": 0.6988,
+      "step": 3353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4347251743302865e-05,
+      "loss": 0.7966,
+      "step": 3354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434397994561203e-05,
+      "loss": 0.7154,
+      "step": 3355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340707574655185e-05,
+      "loss": 0.6737,
+      "step": 3356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433743463086418e-05,
+      "loss": 0.6575,
+      "step": 3357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334161114670943e-05,
+      "loss": 0.6731,
+      "step": 3358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330887026507467e-05,
+      "loss": 0.7338,
+      "step": 3359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327612366805832e-05,
+      "loss": 0.6262,
+      "step": 3360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324337135998183e-05,
+      "loss": 0.6726,
+      "step": 3361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321061334516746e-05,
+      "loss": 0.7608,
+      "step": 3362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317784962793823e-05,
+      "loss": 0.744,
+      "step": 3363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314508021261793e-05,
+      "loss": 0.612,
+      "step": 3364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311230510353102e-05,
+      "loss": 0.7537,
+      "step": 3365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430795243050028e-05,
+      "loss": 0.7225,
+      "step": 3366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304673782135927e-05,
+      "loss": 0.6546,
+      "step": 3367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430139456569272e-05,
+      "loss": 0.7331,
+      "step": 3368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298114781603409e-05,
+      "loss": 0.662,
+      "step": 3369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294834430300822e-05,
+      "loss": 0.6688,
+      "step": 3370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291553512217857e-05,
+      "loss": 0.6169,
+      "step": 3371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428827202778749e-05,
+      "loss": 0.6143,
+      "step": 3372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428498997744278e-05,
+      "loss": 0.6811,
+      "step": 3373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281707361616844e-05,
+      "loss": 0.6796,
+      "step": 3374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278424180742885e-05,
+      "loss": 0.6584,
+      "step": 3375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275140435254176e-05,
+      "loss": 0.6915,
+      "step": 3376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271856125584065e-05,
+      "loss": 0.7406,
+      "step": 3377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426857125216598e-05,
+      "loss": 0.6179,
+      "step": 3378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265285815433419e-05,
+      "loss": 0.6102,
+      "step": 3379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261999815819945e-05,
+      "loss": 0.6597,
+      "step": 3380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258713253759213e-05,
+      "loss": 0.6144,
+      "step": 3381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255426129684944e-05,
+      "loss": 0.634,
+      "step": 3382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425213844403093e-05,
+      "loss": 0.6189,
+      "step": 3383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424885019723104e-05,
+      "loss": 0.6992,
+      "step": 3384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424556138971922e-05,
+      "loss": 0.6795,
+      "step": 3385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242272021929482e-05,
+      "loss": 0.72,
+      "step": 3386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238982094295924e-05,
+      "loss": 0.8305,
+      "step": 3387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235691607252704e-05,
+      "loss": 0.6946,
+      "step": 3388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232400561234066e-05,
+      "loss": 0.6497,
+      "step": 3389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229108956674318e-05,
+      "loss": 0.7606,
+      "step": 3390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225816794007847e-05,
+      "loss": 0.6748,
+      "step": 3391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222524073669117e-05,
+      "loss": 0.633,
+      "step": 3392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421923079609266e-05,
+      "loss": 0.8106,
+      "step": 3393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421593696171308e-05,
+      "loss": 0.7627,
+      "step": 3394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212642570965059e-05,
+      "loss": 0.6954,
+      "step": 3395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209347624283352e-05,
+      "loss": 0.7923,
+      "step": 3396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206052122102784e-05,
+      "loss": 0.6066,
+      "step": 3397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202756064858258e-05,
+      "loss": 0.7105,
+      "step": 3398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199459452984748e-05,
+      "loss": 0.6616,
+      "step": 3399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196162286917297e-05,
+      "loss": 0.6638,
+      "step": 3400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192864567091029e-05,
+      "loss": 0.698,
+      "step": 3401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189566293941137e-05,
+      "loss": 0.6924,
+      "step": 3402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186267467902887e-05,
+      "loss": 0.5886,
+      "step": 3403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418296808941162e-05,
+      "loss": 0.6385,
+      "step": 3404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179668158902742e-05,
+      "loss": 0.7675,
+      "step": 3405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176367676811743e-05,
+      "loss": 0.6243,
+      "step": 3406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173066643574178e-05,
+      "loss": 0.6313,
+      "step": 3407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416976505962568e-05,
+      "loss": 0.6194,
+      "step": 3408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166462925401952e-05,
+      "loss": 0.7881,
+      "step": 3409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416316024133877e-05,
+      "loss": 0.6666,
+      "step": 3410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159857007871977e-05,
+      "loss": 0.624,
+      "step": 3411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156553225437505e-05,
+      "loss": 0.6572,
+      "step": 3412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4153248894471336e-05,
+      "loss": 0.6279,
+      "step": 3413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414994401540954e-05,
+      "loss": 0.5962,
+      "step": 3414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414663858868826e-05,
+      "loss": 0.7458,
+      "step": 3415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143332614743694e-05,
+      "loss": 0.5865,
+      "step": 3416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140026094012136e-05,
+      "loss": 0.705,
+      "step": 3417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136719026929937e-05,
+      "loss": 0.6987,
+      "step": 3418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133411413933525e-05,
+      "loss": 0.7076,
+      "step": 3419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130103255459396e-05,
+      "loss": 0.6938,
+      "step": 3420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412679455194412e-05,
+      "loss": 0.6385,
+      "step": 3421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123485303824342e-05,
+      "loss": 0.6023,
+      "step": 3422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412017551153678e-05,
+      "loss": 0.7138,
+      "step": 3423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116865175518215e-05,
+      "loss": 0.6353,
+      "step": 3424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113554296205506e-05,
+      "loss": 0.691,
+      "step": 3425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110242874035584e-05,
+      "loss": 0.7527,
+      "step": 3426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106930909445451e-05,
+      "loss": 0.7502,
+      "step": 3427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103618402872181e-05,
+      "loss": 0.6534,
+      "step": 3428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100305354752917e-05,
+      "loss": 0.7067,
+      "step": 3429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096991765524872e-05,
+      "loss": 0.7325,
+      "step": 3430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409367763562534e-05,
+      "loss": 0.7093,
+      "step": 3431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090362965491675e-05,
+      "loss": 0.6667,
+      "step": 3432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408704775556131e-05,
+      "loss": 0.7047,
+      "step": 3433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083732006271742e-05,
+      "loss": 0.6854,
+      "step": 3434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080415718060549e-05,
+      "loss": 0.7495,
+      "step": 3435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407709889136537e-05,
+      "loss": 0.6369,
+      "step": 3436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073781526623924e-05,
+      "loss": 0.6807,
+      "step": 3437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070463624273996e-05,
+      "loss": 0.6191,
+      "step": 3438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067145184753438e-05,
+      "loss": 0.7095,
+      "step": 3439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063826208500182e-05,
+      "loss": 0.7493,
+      "step": 3440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060506695952226e-05,
+      "loss": 0.6455,
+      "step": 3441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057186647547638e-05,
+      "loss": 0.6851,
+      "step": 3442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405386606372456e-05,
+      "loss": 0.5458,
+      "step": 3443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050544944921197e-05,
+      "loss": 0.6746,
+      "step": 3444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047223291575835e-05,
+      "loss": 0.7068,
+      "step": 3445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4043901104126826e-05,
+      "loss": 0.6531,
+      "step": 3446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4040578383012594e-05,
+      "loss": 0.7011,
+      "step": 3447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037255128671625e-05,
+      "loss": 0.6878,
+      "step": 3448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033931341542486e-05,
+      "loss": 0.6843,
+      "step": 3449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030607022063809e-05,
+      "loss": 0.7306,
+      "step": 3450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40272821706743e-05,
+      "loss": 0.685,
+      "step": 3451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023956787812731e-05,
+      "loss": 0.7287,
+      "step": 3452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020630873917944e-05,
+      "loss": 0.658,
+      "step": 3453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017304429428858e-05,
+      "loss": 0.6236,
+      "step": 3454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013977454784448e-05,
+      "loss": 0.6374,
+      "step": 3455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010649950423778e-05,
+      "loss": 0.6116,
+      "step": 3456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007321916785968e-05,
+      "loss": 0.5935,
+      "step": 3457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400399335431021e-05,
+      "loss": 0.6871,
+      "step": 3458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400066426343577e-05,
+      "loss": 0.6355,
+      "step": 3459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997334644601978e-05,
+      "loss": 0.6313,
+      "step": 3460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994004498248242e-05,
+      "loss": 0.6978,
+      "step": 3461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990673824814027e-05,
+      "loss": 0.6979,
+      "step": 3462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987342624738882e-05,
+      "loss": 0.6865,
+      "step": 3463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984010898462417e-05,
+      "loss": 0.688,
+      "step": 3464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3980678646424308e-05,
+      "loss": 0.6596,
+      "step": 3465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977345869064314e-05,
+      "loss": 0.6404,
+      "step": 3466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974012566822248e-05,
+      "loss": 0.7064,
+      "step": 3467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970678740137999e-05,
+      "loss": 0.6431,
+      "step": 3468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967344389451532e-05,
+      "loss": 0.7411,
+      "step": 3469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964009515202866e-05,
+      "loss": 0.6025,
+      "step": 3470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960674117832099e-05,
+      "loss": 0.7148,
+      "step": 3471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39573381977794e-05,
+      "loss": 0.7178,
+      "step": 3472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954001755485003e-05,
+      "loss": 0.7222,
+      "step": 3473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395066479138921e-05,
+      "loss": 0.6407,
+      "step": 3474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947327305932392e-05,
+      "loss": 0.5957,
+      "step": 3475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943989299554995e-05,
+      "loss": 0.6381,
+      "step": 3476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940650772697522e-05,
+      "loss": 0.745,
+      "step": 3477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937311725800555e-05,
+      "loss": 0.8197,
+      "step": 3478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393397215930474e-05,
+      "loss": 0.6352,
+      "step": 3479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930632073650796e-05,
+      "loss": 0.5922,
+      "step": 3480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39272914692795e-05,
+      "loss": 0.713,
+      "step": 3481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923950346631712e-05,
+      "loss": 0.5936,
+      "step": 3482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920608706148351e-05,
+      "loss": 0.6999,
+      "step": 3483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917266548270405e-05,
+      "loss": 0.7017,
+      "step": 3484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391392387343893e-05,
+      "loss": 0.6651,
+      "step": 3485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910580682095059e-05,
+      "loss": 0.7451,
+      "step": 3486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907236974679977e-05,
+      "loss": 0.705,
+      "step": 3487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903892751634949e-05,
+      "loss": 0.5882,
+      "step": 3488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390054801340131e-05,
+      "loss": 0.6959,
+      "step": 3489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897202760420449e-05,
+      "loss": 0.7374,
+      "step": 3490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893856993133842e-05,
+      "loss": 0.6795,
+      "step": 3491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890510711983018e-05,
+      "loss": 0.775,
+      "step": 3492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887163917409581e-05,
+      "loss": 0.7369,
+      "step": 3493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3883816609855194e-05,
+      "loss": 0.612,
+      "step": 3494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880468789761603e-05,
+      "loss": 0.7068,
+      "step": 3495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877120457570608e-05,
+      "loss": 0.6523,
+      "step": 3496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873771613724083e-05,
+      "loss": 0.6784,
+      "step": 3497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870422258663964e-05,
+      "loss": 0.6704,
+      "step": 3498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3867072392832261e-05,
+      "loss": 0.6656,
+      "step": 3499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863722016671054e-05,
+      "loss": 0.6842,
+      "step": 3500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860371130622482e-05,
+      "loss": 0.6195,
+      "step": 3501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385701973512875e-05,
+      "loss": 0.7222,
+      "step": 3502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853667830632139e-05,
+      "loss": 0.5934,
+      "step": 3503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850315417574986e-05,
+      "loss": 0.6915,
+      "step": 3504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846962496399714e-05,
+      "loss": 0.6052,
+      "step": 3505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843609067548793e-05,
+      "loss": 0.6949,
+      "step": 3506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384025513146477e-05,
+      "loss": 0.7245,
+      "step": 3507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836900688590259e-05,
+      "loss": 0.6397,
+      "step": 3508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833545739367931e-05,
+      "loss": 0.6316,
+      "step": 3509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830190284240543e-05,
+      "loss": 0.7109,
+      "step": 3510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.7232,
+      "step": 3511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823477858041883e-05,
+      "loss": 0.7281,
+      "step": 3512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820120887856439e-05,
+      "loss": 0.7193,
+      "step": 3513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816763413537575e-05,
+      "loss": 0.6591,
+      "step": 3514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813405435528376e-05,
+      "loss": 0.7341,
+      "step": 3515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810046954271985e-05,
+      "loss": 0.7211,
+      "step": 3516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806687970211615e-05,
+      "loss": 0.594,
+      "step": 3517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803328483790544e-05,
+      "loss": 0.6923,
+      "step": 3518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799968495452115e-05,
+      "loss": 0.7214,
+      "step": 3519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796608005639738e-05,
+      "loss": 0.7306,
+      "step": 3520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793247014796894e-05,
+      "loss": 0.6612,
+      "step": 3521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789885523367122e-05,
+      "loss": 0.7323,
+      "step": 3522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786523531794034e-05,
+      "loss": 0.6842,
+      "step": 3523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783161040521298e-05,
+      "loss": 0.7255,
+      "step": 3524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779798049992663e-05,
+      "loss": 0.6713,
+      "step": 3525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776434560651935e-05,
+      "loss": 0.6771,
+      "step": 3526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773070572942984e-05,
+      "loss": 0.6908,
+      "step": 3527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769706087309752e-05,
+      "loss": 0.6865,
+      "step": 3528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766341104196234e-05,
+      "loss": 0.7104,
+      "step": 3529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762975624046511e-05,
+      "loss": 0.6534,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759609647304715e-05,
+      "loss": 0.575,
+      "step": 3531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756243174415044e-05,
+      "loss": 0.6871,
+      "step": 3532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752876205821765e-05,
+      "loss": 0.7024,
+      "step": 3533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749508741969213e-05,
+      "loss": 0.7039,
+      "step": 3534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374614078330178e-05,
+      "loss": 0.6783,
+      "step": 3535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742772330263936e-05,
+      "loss": 0.7211,
+      "step": 3536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739403383300202e-05,
+      "loss": 0.754,
+      "step": 3537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736033942855176e-05,
+      "loss": 0.6284,
+      "step": 3538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373266400937351e-05,
+      "loss": 0.695,
+      "step": 3539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729293583299932e-05,
+      "loss": 0.6445,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372592266507923e-05,
+      "loss": 0.672,
+      "step": 3541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722551255156253e-05,
+      "loss": 0.668,
+      "step": 3542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719179353975924e-05,
+      "loss": 0.6886,
+      "step": 3543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715806961983222e-05,
+      "loss": 0.682,
+      "step": 3544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712434079623198e-05,
+      "loss": 0.7283,
+      "step": 3545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709060707340963e-05,
+      "loss": 0.6305,
+      "step": 3546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705686845581692e-05,
+      "loss": 0.6374,
+      "step": 3547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702312494790626e-05,
+      "loss": 0.6892,
+      "step": 3548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698937655413076e-05,
+      "loss": 0.605,
+      "step": 3549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369556232789441e-05,
+      "loss": 0.7494,
+      "step": 3550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692186512680058e-05,
+      "loss": 0.6274,
+      "step": 3551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688810210215526e-05,
+      "loss": 0.7402,
+      "step": 3552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685433420946374e-05,
+      "loss": 0.6922,
+      "step": 3553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682056145318235e-05,
+      "loss": 0.6162,
+      "step": 3554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678678383776796e-05,
+      "loss": 0.5809,
+      "step": 3555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675300136767816e-05,
+      "loss": 0.7105,
+      "step": 3556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671921404737109e-05,
+      "loss": 0.5937,
+      "step": 3557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668542188130567e-05,
+      "loss": 0.6872,
+      "step": 3558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665162487394135e-05,
+      "loss": 0.6849,
+      "step": 3559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661782302973825e-05,
+      "loss": 0.6432,
+      "step": 3560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365840163531571e-05,
+      "loss": 0.7069,
+      "step": 3561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655020484865937e-05,
+      "loss": 0.649,
+      "step": 3562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651638852070703e-05,
+      "loss": 0.6249,
+      "step": 3563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648256737376278e-05,
+      "loss": 0.6928,
+      "step": 3564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644874141228993e-05,
+      "loss": 0.7046,
+      "step": 3565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364149106407524e-05,
+      "loss": 0.7576,
+      "step": 3566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363810750636148e-05,
+      "loss": 0.6589,
+      "step": 3567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634723468534228e-05,
+      "loss": 0.6046,
+      "step": 3568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631338951040074e-05,
+      "loss": 0.7503,
+      "step": 3569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627953954325664e-05,
+      "loss": 0.6724,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624568478837707e-05,
+      "loss": 0.6887,
+      "step": 3571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621182525022984e-05,
+      "loss": 0.6756,
+      "step": 3572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617796093328322e-05,
+      "loss": 0.6564,
+      "step": 3573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361440918420063e-05,
+      "loss": 0.6571,
+      "step": 3574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361102179808687e-05,
+      "loss": 0.6448,
+      "step": 3575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607633935434063e-05,
+      "loss": 0.7099,
+      "step": 3576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604245596689304e-05,
+      "loss": 0.7459,
+      "step": 3577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360085678229974e-05,
+      "loss": 0.7124,
+      "step": 3578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597467492712589e-05,
+      "loss": 0.7583,
+      "step": 3579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594077728375129e-05,
+      "loss": 0.6566,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35906874897347e-05,
+      "loss": 0.585,
+      "step": 3581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587296777238703e-05,
+      "loss": 0.7177,
+      "step": 3582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583905591334603e-05,
+      "loss": 0.6296,
+      "step": 3583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580513932469932e-05,
+      "loss": 0.6437,
+      "step": 3584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577121801092274e-05,
+      "loss": 0.6656,
+      "step": 3585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573729197649288e-05,
+      "loss": 0.7232,
+      "step": 3586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570336122588681e-05,
+      "loss": 0.6617,
+      "step": 3587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356694257635824e-05,
+      "loss": 0.7017,
+      "step": 3588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563548559405793e-05,
+      "loss": 0.6503,
+      "step": 3589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356015407217925e-05,
+      "loss": 0.7024,
+      "step": 3590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556759115126572e-05,
+      "loss": 0.728,
+      "step": 3591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553363688695784e-05,
+      "loss": 0.6316,
+      "step": 3592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549967793334972e-05,
+      "loss": 0.7093,
+      "step": 3593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546571429492288e-05,
+      "loss": 0.7,
+      "step": 3594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543174597615942e-05,
+      "loss": 0.7272,
+      "step": 3595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539777298154206e-05,
+      "loss": 0.7419,
+      "step": 3596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536379531555413e-05,
+      "loss": 0.6273,
+      "step": 3597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532981298267962e-05,
+      "loss": 0.6983,
+      "step": 3598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529582598740312e-05,
+      "loss": 0.7345,
+      "step": 3599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352618343342098e-05,
+      "loss": 0.7323,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522783802758547e-05,
+      "loss": 0.7099,
+      "step": 3601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519383707201655e-05,
+      "loss": 0.7552,
+      "step": 3602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515983147199008e-05,
+      "loss": 0.6747,
+      "step": 3603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512582123199371e-05,
+      "loss": 0.7788,
+      "step": 3604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350918063565157e-05,
+      "loss": 0.7129,
+      "step": 3605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505778685004491e-05,
+      "loss": 0.5679,
+      "step": 3606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502376271707082e-05,
+      "loss": 0.6936,
+      "step": 3607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498973396208357e-05,
+      "loss": 0.68,
+      "step": 3608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495570058957384e-05,
+      "loss": 0.7242,
+      "step": 3609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492166260403293e-05,
+      "loss": 0.5512,
+      "step": 3610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3488762000995275e-05,
+      "loss": 0.6461,
+      "step": 3611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485357281182586e-05,
+      "loss": 0.596,
+      "step": 3612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481952101414538e-05,
+      "loss": 0.668,
+      "step": 3613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478546462140507e-05,
+      "loss": 0.7425,
+      "step": 3614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475140363809924e-05,
+      "loss": 0.6661,
+      "step": 3615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471733806872292e-05,
+      "loss": 0.6354,
+      "step": 3616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346832679177716e-05,
+      "loss": 0.7002,
+      "step": 3617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3464919318974152e-05,
+      "loss": 0.6316,
+      "step": 3618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346151138891294e-05,
+      "loss": 0.6291,
+      "step": 3619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458103002043261e-05,
+      "loss": 0.7592,
+      "step": 3620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454694158814915e-05,
+      "loss": 0.68,
+      "step": 3621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451284859677757e-05,
+      "loss": 0.7134,
+      "step": 3622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447875105081712e-05,
+      "loss": 0.7663,
+      "step": 3623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344446489547675e-05,
+      "loss": 0.7015,
+      "step": 3624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3441054231312914e-05,
+      "loss": 0.7121,
+      "step": 3625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437643113040302e-05,
+      "loss": 0.74,
+      "step": 3626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434231541109069e-05,
+      "loss": 0.6817,
+      "step": 3627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430819515969436e-05,
+      "loss": 0.6569,
+      "step": 3628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427407038071682e-05,
+      "loss": 0.6802,
+      "step": 3629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342399410786614e-05,
+      "loss": 0.6368,
+      "step": 3630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420580725803214e-05,
+      "loss": 0.6524,
+      "step": 3631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417166892333353e-05,
+      "loss": 0.7214,
+      "step": 3632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413752607907078e-05,
+      "loss": 0.6164,
+      "step": 3633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410337872974965e-05,
+      "loss": 0.6916,
+      "step": 3634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406922687987653e-05,
+      "loss": 0.65,
+      "step": 3635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340350705339583e-05,
+      "loss": 0.6687,
+      "step": 3636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400090969650252e-05,
+      "loss": 0.7512,
+      "step": 3637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396674437201735e-05,
+      "loss": 0.6027,
+      "step": 3638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393257456501151e-05,
+      "loss": 0.731,
+      "step": 3639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338984002799943e-05,
+      "loss": 0.6737,
+      "step": 3640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3386422152147564e-05,
+      "loss": 0.7165,
+      "step": 3641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383003829396604e-05,
+      "loss": 0.6455,
+      "step": 3642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379585060197658e-05,
+      "loss": 0.6856,
+      "step": 3643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376165845001891e-05,
+      "loss": 0.719,
+      "step": 3644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372746184260536e-05,
+      "loss": 0.7685,
+      "step": 3645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369326078424875e-05,
+      "loss": 0.649,
+      "step": 3646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336590552794625e-05,
+      "loss": 0.6799,
+      "step": 3647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336248453327607e-05,
+      "loss": 0.7652,
+      "step": 3648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359063094865792e-05,
+      "loss": 0.6446,
+      "step": 3649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355641213166934e-05,
+      "loss": 0.7046,
+      "step": 3650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352218888631079e-05,
+      "loss": 0.6277,
+      "step": 3651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3348796121709862e-05,
+      "loss": 0.758,
+      "step": 3652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334537291285498e-05,
+      "loss": 0.6377,
+      "step": 3653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341949262518188e-05,
+      "loss": 0.7249,
+      "step": 3654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338525171151294e-05,
+      "loss": 0.6754,
+      "step": 3655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335100639206169e-05,
+      "loss": 0.5696,
+      "step": 3656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331675667134745e-05,
+      "loss": 0.7279,
+      "step": 3657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328250255389003e-05,
+      "loss": 0.7163,
+      "step": 3658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324824404420996e-05,
+      "loss": 0.6289,
+      "step": 3659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321398114682813e-05,
+      "loss": 0.6524,
+      "step": 3660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317971386626624e-05,
+      "loss": 0.588,
+      "step": 3661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314544220704648e-05,
+      "loss": 0.7025,
+      "step": 3662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331111661736916e-05,
+      "loss": 0.5886,
+      "step": 3663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3307688577072488e-05,
+      "loss": 0.7336,
+      "step": 3664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304260100267032e-05,
+      "loss": 0.6624,
+      "step": 3665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300831187405233e-05,
+      "loss": 0.6035,
+      "step": 3666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32974018389396e-05,
+      "loss": 0.7033,
+      "step": 3667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293972055322702e-05,
+      "loss": 0.7169,
+      "step": 3668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329054183700715e-05,
+      "loss": 0.8189,
+      "step": 3669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287111184445634e-05,
+      "loss": 0.6529,
+      "step": 3670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328368009809088e-05,
+      "loss": 0.66,
+      "step": 3671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328024857839569e-05,
+      "loss": 0.6578,
+      "step": 3672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276816625812914e-05,
+      "loss": 0.6647,
+      "step": 3673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273384240795455e-05,
+      "loss": 0.6157,
+      "step": 3674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269951423796277e-05,
+      "loss": 0.6517,
+      "step": 3675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266518175268403e-05,
+      "loss": 0.6901,
+      "step": 3676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263084495664917e-05,
+      "loss": 0.6922,
+      "step": 3677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259650385438947e-05,
+      "loss": 0.7313,
+      "step": 3678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325621584504369e-05,
+      "loss": 0.696,
+      "step": 3679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252780874932395e-05,
+      "loss": 0.6518,
+      "step": 3680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249345475558361e-05,
+      "loss": 0.5915,
+      "step": 3681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245909647374963e-05,
+      "loss": 0.7439,
+      "step": 3682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242473390835613e-05,
+      "loss": 0.6984,
+      "step": 3683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239036706393784e-05,
+      "loss": 0.7711,
+      "step": 3684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235599594503015e-05,
+      "loss": 0.7663,
+      "step": 3685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232162055616886e-05,
+      "loss": 0.6772,
+      "step": 3686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322872409018905e-05,
+      "loss": 0.7332,
+      "step": 3687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225285698673198e-05,
+      "loss": 0.5686,
+      "step": 3688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32218468815231e-05,
+      "loss": 0.7239,
+      "step": 3689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218407639192563e-05,
+      "loss": 0.6747,
+      "step": 3690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214967972135457e-05,
+      "loss": 0.6627,
+      "step": 3691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211527880805709e-05,
+      "loss": 0.7508,
+      "step": 3692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208087365657299e-05,
+      "loss": 0.7623,
+      "step": 3693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204646427144265e-05,
+      "loss": 0.7306,
+      "step": 3694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201205065720699e-05,
+      "loss": 0.6843,
+      "step": 3695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197763281840752e-05,
+      "loss": 0.7785,
+      "step": 3696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194321075958629e-05,
+      "loss": 0.7175,
+      "step": 3697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190878448528595e-05,
+      "loss": 0.7092,
+      "step": 3698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318743540000496e-05,
+      "loss": 0.7023,
+      "step": 3699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3183991930842099e-05,
+      "loss": 0.6403,
+      "step": 3700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180548041494436e-05,
+      "loss": 0.672,
+      "step": 3701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3177103732416462e-05,
+      "loss": 0.6737,
+      "step": 3702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173659004062707e-05,
+      "loss": 0.6211,
+      "step": 3703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170213856887766e-05,
+      "loss": 0.6024,
+      "step": 3704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3166768291346291e-05,
+      "loss": 0.705,
+      "step": 3705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163322307892983e-05,
+      "loss": 0.6537,
+      "step": 3706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159875906982604e-05,
+      "loss": 0.6363,
+      "step": 3707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156429089069969e-05,
+      "loss": 0.6231,
+      "step": 3708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152981854609946e-05,
+      "loss": 0.6145,
+      "step": 3709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149534204057457e-05,
+      "loss": 0.6396,
+      "step": 3710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146086137867481e-05,
+      "loss": 0.6816,
+      "step": 3711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314263765649506e-05,
+      "loss": 0.6318,
+      "step": 3712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139188760395277e-05,
+      "loss": 0.7802,
+      "step": 3713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313573945002327e-05,
+      "loss": 0.6646,
+      "step": 3714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132289725834246e-05,
+      "loss": 0.7406,
+      "step": 3715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128839588283457e-05,
+      "loss": 0.7347,
+      "step": 3716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125389037826211e-05,
+      "loss": 0.7074,
+      "step": 3717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121938074917866e-05,
+      "loss": 0.6957,
+      "step": 3718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311848670001384e-05,
+      "loss": 0.6355,
+      "step": 3719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115034913569601e-05,
+      "loss": 0.685,
+      "step": 3720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311158271604068e-05,
+      "loss": 0.6035,
+      "step": 3721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3108130107882652e-05,
+      "loss": 0.7453,
+      "step": 3722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104677089551153e-05,
+      "loss": 0.6601,
+      "step": 3723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310122366150187e-05,
+      "loss": 0.6329,
+      "step": 3724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097769824190541e-05,
+      "loss": 0.7801,
+      "step": 3725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309431557807297e-05,
+      "loss": 0.6271,
+      "step": 3726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090860923605e-05,
+      "loss": 0.6002,
+      "step": 3727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087405861242535e-05,
+      "loss": 0.7228,
+      "step": 3728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083950391441535e-05,
+      "loss": 0.7348,
+      "step": 3729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080494514658008e-05,
+      "loss": 0.6549,
+      "step": 3730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077038231348023e-05,
+      "loss": 0.7924,
+      "step": 3731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073581541967693e-05,
+      "loss": 0.6332,
+      "step": 3732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070124446973196e-05,
+      "loss": 0.7096,
+      "step": 3733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066666946820754e-05,
+      "loss": 0.6611,
+      "step": 3734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063209041966647e-05,
+      "loss": 0.6525,
+      "step": 3735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059750732867209e-05,
+      "loss": 0.7117,
+      "step": 3736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056292019978823e-05,
+      "loss": 0.737,
+      "step": 3737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052832903757931e-05,
+      "loss": 0.602,
+      "step": 3738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049373384661024e-05,
+      "loss": 0.6788,
+      "step": 3739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045913463144646e-05,
+      "loss": 0.6497,
+      "step": 3740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042453139665398e-05,
+      "loss": 0.7171,
+      "step": 3741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303899241467993e-05,
+      "loss": 0.7033,
+      "step": 3742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303553128864495e-05,
+      "loss": 0.7065,
+      "step": 3743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032069762017213e-05,
+      "loss": 0.7582,
+      "step": 3744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028607835253526e-05,
+      "loss": 0.6853,
+      "step": 3745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302514550881076e-05,
+      "loss": 0.7511,
+      "step": 3746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021682783145826e-05,
+      "loss": 0.7845,
+      "step": 3747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018219658715694e-05,
+      "loss": 0.6044,
+      "step": 3748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014756135977383e-05,
+      "loss": 0.68,
+      "step": 3749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301129221538797e-05,
+      "loss": 0.7029,
+      "step": 3750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007827897404579e-05,
+      "loss": 0.7491,
+      "step": 3751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004363182484391e-05,
+      "loss": 0.7181,
+      "step": 3752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300089807108464e-05,
+      "loss": 0.6612,
+      "step": 3753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997432563662604e-05,
+      "loss": 0.5977,
+      "step": 3754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993966660675619e-05,
+      "loss": 0.6651,
+      "step": 3755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990500362581076e-05,
+      "loss": 0.7033,
+      "step": 3756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987033669836416e-05,
+      "loss": 0.7431,
+      "step": 3757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983566582899125e-05,
+      "loss": 0.6004,
+      "step": 3758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980099102226753e-05,
+      "loss": 0.705,
+      "step": 3759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976631228276894e-05,
+      "loss": 0.6504,
+      "step": 3760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29731629615072e-05,
+      "loss": 0.6962,
+      "step": 3761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969694302375366e-05,
+      "loss": 0.6635,
+      "step": 3762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966225251339146e-05,
+      "loss": 0.7716,
+      "step": 3763
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962755808856341e-05,
+      "loss": 0.7382,
+      "step": 3764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295928597538481e-05,
+      "loss": 0.7756,
+      "step": 3765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955815751382456e-05,
+      "loss": 0.5957,
+      "step": 3766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952345137307244e-05,
+      "loss": 0.699,
+      "step": 3767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294887413361717e-05,
+      "loss": 0.5671,
+      "step": 3768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294540274077031e-05,
+      "loss": 0.6796,
+      "step": 3769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294193095922477e-05,
+      "loss": 0.6682,
+      "step": 3770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938458789438713e-05,
+      "loss": 0.6409,
+      "step": 3771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934986231870357e-05,
+      "loss": 0.6908,
+      "step": 3772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931513286977966e-05,
+      "loss": 0.6013,
+      "step": 3773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928039955219855e-05,
+      "loss": 0.6817,
+      "step": 3774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924566237054398e-05,
+      "loss": 0.6332,
+      "step": 3775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921092132940012e-05,
+      "loss": 0.6542,
+      "step": 3776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917617643335164e-05,
+      "loss": 0.7087,
+      "step": 3777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914142768698379e-05,
+      "loss": 0.6659,
+      "step": 3778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910667509488227e-05,
+      "loss": 0.6819,
+      "step": 3779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907191866163333e-05,
+      "loss": 0.7099,
+      "step": 3780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290371583918237e-05,
+      "loss": 0.7854,
+      "step": 3781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900239429004062e-05,
+      "loss": 0.6081,
+      "step": 3782
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289676263608718e-05,
+      "loss": 0.7585,
+      "step": 3783
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893285460890555e-05,
+      "loss": 0.6587,
+      "step": 3784
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889807903873057e-05,
+      "loss": 0.6438,
+      "step": 3785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2886329965493615e-05,
+      "loss": 0.7983,
+      "step": 3786
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882851646211206e-05,
+      "loss": 0.8366,
+      "step": 3787
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879372946484857e-05,
+      "loss": 0.7736,
+      "step": 3788
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2875893866773637e-05,
+      "loss": 0.6732,
+      "step": 3789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872414407536684e-05,
+      "loss": 0.656,
+      "step": 3790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868934569233169e-05,
+      "loss": 0.6983,
+      "step": 3791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286545435232232e-05,
+      "loss": 0.7709,
+      "step": 3792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861973757263416e-05,
+      "loss": 0.624,
+      "step": 3793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858492784515779e-05,
+      "loss": 0.6713,
+      "step": 3794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855011434538791e-05,
+      "loss": 0.7078,
+      "step": 3795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851529707791874e-05,
+      "loss": 0.6572,
+      "step": 3796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848047604734509e-05,
+      "loss": 0.6868,
+      "step": 3797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844565125826219e-05,
+      "loss": 0.656,
+      "step": 3798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841082271526574e-05,
+      "loss": 0.6563,
+      "step": 3799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837599042295212e-05,
+      "loss": 0.6994,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28341154385918e-05,
+      "loss": 0.6896,
+      "step": 3801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830631460876061e-05,
+      "loss": 0.6254,
+      "step": 3802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827147109607769e-05,
+      "loss": 0.6843,
+      "step": 3803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823662385246749e-05,
+      "loss": 0.7232,
+      "step": 3804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282017728825287e-05,
+      "loss": 0.73,
+      "step": 3805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816691819086053e-05,
+      "loss": 0.6772,
+      "step": 3806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813205978206274e-05,
+      "loss": 0.7371,
+      "step": 3807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809719766073542e-05,
+      "loss": 0.6683,
+      "step": 3808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806233183147934e-05,
+      "loss": 0.6876,
+      "step": 3809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802746229889563e-05,
+      "loss": 0.6421,
+      "step": 3810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799258906758595e-05,
+      "loss": 0.6277,
+      "step": 3811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795771214215247e-05,
+      "loss": 0.7455,
+      "step": 3812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792283152719777e-05,
+      "loss": 0.7777,
+      "step": 3813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788794722732504e-05,
+      "loss": 0.6873,
+      "step": 3814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785305924713786e-05,
+      "loss": 0.6482,
+      "step": 3815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781816759124035e-05,
+      "loss": 0.7224,
+      "step": 3816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778327226423706e-05,
+      "loss": 0.686,
+      "step": 3817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774837327073307e-05,
+      "loss": 0.6103,
+      "step": 3818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771347061533385e-05,
+      "loss": 0.6922,
+      "step": 3819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767856430264556e-05,
+      "loss": 0.611,
+      "step": 3820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764365433727463e-05,
+      "loss": 0.7286,
+      "step": 3821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760874072382805e-05,
+      "loss": 0.7443,
+      "step": 3822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757382346691336e-05,
+      "loss": 0.6907,
+      "step": 3823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753890257113845e-05,
+      "loss": 0.7647,
+      "step": 3824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750397804111183e-05,
+      "loss": 0.687,
+      "step": 3825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746904988144238e-05,
+      "loss": 0.63,
+      "step": 3826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743411809673947e-05,
+      "loss": 0.7115,
+      "step": 3827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739918269161298e-05,
+      "loss": 0.6458,
+      "step": 3828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736424367067331e-05,
+      "loss": 0.8365,
+      "step": 3829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732930103853126e-05,
+      "loss": 0.6101,
+      "step": 3830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272943547997981e-05,
+      "loss": 0.6957,
+      "step": 3831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272594049590857e-05,
+      "loss": 0.604,
+      "step": 3832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722445152100624e-05,
+      "loss": 0.7842,
+      "step": 3833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718949449017251e-05,
+      "loss": 0.65,
+      "step": 3834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715453387119768e-05,
+      "loss": 0.6259,
+      "step": 3835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711956966869543e-05,
+      "loss": 0.5946,
+      "step": 3836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708460188727991e-05,
+      "loss": 0.6576,
+      "step": 3837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704963053156576e-05,
+      "loss": 0.6743,
+      "step": 3838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701465560616808e-05,
+      "loss": 0.6128,
+      "step": 3839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697967711570243e-05,
+      "loss": 0.7263,
+      "step": 3840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694469506478486e-05,
+      "loss": 0.6243,
+      "step": 3841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690970945803186e-05,
+      "loss": 0.7871,
+      "step": 3842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687472030006042e-05,
+      "loss": 0.6509,
+      "step": 3843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683972759548799e-05,
+      "loss": 0.7075,
+      "step": 3844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680473134893247e-05,
+      "loss": 0.7161,
+      "step": 3845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676973156501226e-05,
+      "loss": 0.5799,
+      "step": 3846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673472824834621e-05,
+      "loss": 0.6075,
+      "step": 3847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669972140355359e-05,
+      "loss": 0.7243,
+      "step": 3848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666471103525424e-05,
+      "loss": 0.6668,
+      "step": 3849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662969714806834e-05,
+      "loss": 0.6978,
+      "step": 3850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265946797466167e-05,
+      "loss": 0.6258,
+      "step": 3851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655965883552039e-05,
+      "loss": 0.7106,
+      "step": 3852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2652463441940111e-05,
+      "loss": 0.7052,
+      "step": 3853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648960650288094e-05,
+      "loss": 0.7284,
+      "step": 3854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645457509058243e-05,
+      "loss": 0.6241,
+      "step": 3855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641954018712863e-05,
+      "loss": 0.6308,
+      "step": 3856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26384501797143e-05,
+      "loss": 0.6017,
+      "step": 3857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634945992524947e-05,
+      "loss": 0.6974,
+      "step": 3858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631441457607247e-05,
+      "loss": 0.7406,
+      "step": 3859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627936575423688e-05,
+      "loss": 0.5918,
+      "step": 3860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624431346436798e-05,
+      "loss": 0.7072,
+      "step": 3861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620925771109159e-05,
+      "loss": 0.5967,
+      "step": 3862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617419849903387e-05,
+      "loss": 0.7351,
+      "step": 3863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261391358328216e-05,
+      "loss": 0.6656,
+      "step": 3864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261040697170819e-05,
+      "loss": 0.6294,
+      "step": 3865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606900015644232e-05,
+      "loss": 0.6781,
+      "step": 3866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2603392715553095e-05,
+      "loss": 0.5725,
+      "step": 3867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599885071897632e-05,
+      "loss": 0.7624,
+      "step": 3868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259637708514074e-05,
+      "loss": 0.6897,
+      "step": 3869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259286875574536e-05,
+      "loss": 0.7049,
+      "step": 3870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589360084174476e-05,
+      "loss": 0.773,
+      "step": 3871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585851070891122e-05,
+      "loss": 0.6439,
+      "step": 3872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582341716358374e-05,
+      "loss": 0.7516,
+      "step": 3873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578832021039358e-05,
+      "loss": 0.7294,
+      "step": 3874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575321985397238e-05,
+      "loss": 0.6329,
+      "step": 3875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571811609895224e-05,
+      "loss": 0.6862,
+      "step": 3876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568300894996579e-05,
+      "loss": 0.7538,
+      "step": 3877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25647898411646e-05,
+      "loss": 0.7286,
+      "step": 3878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561278448862635e-05,
+      "loss": 0.6475,
+      "step": 3879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557766718554074e-05,
+      "loss": 0.6589,
+      "step": 3880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554254650702355e-05,
+      "loss": 0.6323,
+      "step": 3881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550742245770956e-05,
+      "loss": 0.6825,
+      "step": 3882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547229504223406e-05,
+      "loss": 0.5979,
+      "step": 3883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254371642652327e-05,
+      "loss": 0.6299,
+      "step": 3884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254020301313416e-05,
+      "loss": 0.6154,
+      "step": 3885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253668926451974e-05,
+      "loss": 0.7158,
+      "step": 3886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533175181143704e-05,
+      "loss": 0.6649,
+      "step": 3887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529660763469807e-05,
+      "loss": 0.6672,
+      "step": 3888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526146011961834e-05,
+      "loss": 0.7699,
+      "step": 3889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522630927083622e-05,
+      "loss": 0.6527,
+      "step": 3890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519115509299047e-05,
+      "loss": 0.695,
+      "step": 3891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515599759072033e-05,
+      "loss": 0.8118,
+      "step": 3892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512083676866546e-05,
+      "loss": 0.7593,
+      "step": 3893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508567263146593e-05,
+      "loss": 0.6635,
+      "step": 3894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505050518376234e-05,
+      "loss": 0.6449,
+      "step": 3895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501533443019562e-05,
+      "loss": 0.7422,
+      "step": 3896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498016037540719e-05,
+      "loss": 0.7895,
+      "step": 3897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494498302403893e-05,
+      "loss": 0.6935,
+      "step": 3898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490980238073307e-05,
+      "loss": 0.6985,
+      "step": 3899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2487461845013236e-05,
+      "loss": 0.8383,
+      "step": 3900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483943123687991e-05,
+      "loss": 0.7293,
+      "step": 3901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480424074561934e-05,
+      "loss": 0.7775,
+      "step": 3902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476904698099468e-05,
+      "loss": 0.6999,
+      "step": 3903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473384994765034e-05,
+      "loss": 0.6901,
+      "step": 3904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469864965023121e-05,
+      "loss": 0.7265,
+      "step": 3905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466344609338263e-05,
+      "loss": 0.6558,
+      "step": 3906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462823928175029e-05,
+      "loss": 0.6215,
+      "step": 3907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245930292199804e-05,
+      "loss": 0.7052,
+      "step": 3908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455781591271955e-05,
+      "loss": 0.7066,
+      "step": 3909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452259936461477e-05,
+      "loss": 0.7009,
+      "step": 3910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244873795803135e-05,
+      "loss": 0.726,
+      "step": 3911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445215656446358e-05,
+      "loss": 0.6308,
+      "step": 3912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244169303217134e-05,
+      "loss": 0.7208,
+      "step": 3913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438170085671169e-05,
+      "loss": 0.687,
+      "step": 3914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434646817410755e-05,
+      "loss": 0.6169,
+      "step": 3915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243112322785506e-05,
+      "loss": 0.7611,
+      "step": 3916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427599317469084e-05,
+      "loss": 0.7054,
+      "step": 3917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424075086717871e-05,
+      "loss": 0.7069,
+      "step": 3918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420550536066504e-05,
+      "loss": 0.7741,
+      "step": 3919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417025665980114e-05,
+      "loss": 0.7392,
+      "step": 3920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2413500476923868e-05,
+      "loss": 0.7242,
+      "step": 3921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2409974969362977e-05,
+      "loss": 0.6173,
+      "step": 3922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24064491437627e-05,
+      "loss": 0.6006,
+      "step": 3923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240292300058833e-05,
+      "loss": 0.6466,
+      "step": 3924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2399396540305205e-05,
+      "loss": 0.6806,
+      "step": 3925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395869763378702e-05,
+      "loss": 0.6901,
+      "step": 3926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392342670274244e-05,
+      "loss": 0.6733,
+      "step": 3927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388815261457297e-05,
+      "loss": 0.7313,
+      "step": 3928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385287537393363e-05,
+      "loss": 0.6541,
+      "step": 3929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381759498547985e-05,
+      "loss": 0.7083,
+      "step": 3930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237823114538676e-05,
+      "loss": 0.7116,
+      "step": 3931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374702478375309e-05,
+      "loss": 0.6754,
+      "step": 3932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237117349797931e-05,
+      "loss": 0.8628,
+      "step": 3933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367644204664468e-05,
+      "loss": 0.7845,
+      "step": 3934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364114598896543e-05,
+      "loss": 0.7277,
+      "step": 3935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360584681141325e-05,
+      "loss": 0.7217,
+      "step": 3936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235705445186465e-05,
+      "loss": 0.7736,
+      "step": 3937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23535239115324e-05,
+      "loss": 0.7408,
+      "step": 3938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349993060610486e-05,
+      "loss": 0.7046,
+      "step": 3939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346461899564873e-05,
+      "loss": 0.6666,
+      "step": 3940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342930428861557e-05,
+      "loss": 0.7196,
+      "step": 3941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339398648966583e-05,
+      "loss": 0.7374,
+      "step": 3942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335866560346029e-05,
+      "loss": 0.6455,
+      "step": 3943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233233416346602e-05,
+      "loss": 0.6919,
+      "step": 3944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328801458792717e-05,
+      "loss": 0.6704,
+      "step": 3945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325268446792322e-05,
+      "loss": 0.6394,
+      "step": 3946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321735127931084e-05,
+      "loss": 0.6747,
+      "step": 3947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318201502675285e-05,
+      "loss": 0.6488,
+      "step": 3948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231466757149125e-05,
+      "loss": 0.6733,
+      "step": 3949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311133334845345e-05,
+      "loss": 0.7788,
+      "step": 3950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307598793203975e-05,
+      "loss": 0.7511,
+      "step": 3951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230406394703359e-05,
+      "loss": 0.6982,
+      "step": 3952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300528796800673e-05,
+      "loss": 0.8141,
+      "step": 3953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229699334297175e-05,
+      "loss": 0.6159,
+      "step": 3954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229345758601339e-05,
+      "loss": 0.6207,
+      "step": 3955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289921526392194e-05,
+      "loss": 0.7137,
+      "step": 3956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286385164574814e-05,
+      "loss": 0.6384,
+      "step": 3957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282848501027933e-05,
+      "loss": 0.6945,
+      "step": 3958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279311536218281e-05,
+      "loss": 0.6586,
+      "step": 3959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275774270612625e-05,
+      "loss": 0.6724,
+      "step": 3960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227223670467776e-05,
+      "loss": 0.7363,
+      "step": 3961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268698838880544e-05,
+      "loss": 0.7316,
+      "step": 3962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265160673687854e-05,
+      "loss": 0.6036,
+      "step": 3963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261622209566617e-05,
+      "loss": 0.7769,
+      "step": 3964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258083446983796e-05,
+      "loss": 0.6162,
+      "step": 3965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254544386406395e-05,
+      "loss": 0.6753,
+      "step": 3966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251005028301455e-05,
+      "loss": 0.6122,
+      "step": 3967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247465373136059e-05,
+      "loss": 0.7194,
+      "step": 3968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243925421377327e-05,
+      "loss": 0.6849,
+      "step": 3969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2240385173492422e-05,
+      "loss": 0.6469,
+      "step": 3970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236844629948538e-05,
+      "loss": 0.647,
+      "step": 3971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233303791212919e-05,
+      "loss": 0.6688,
+      "step": 3972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229762657752838e-05,
+      "loss": 0.7044,
+      "step": 3973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226221230035613e-05,
+      "loss": 0.7132,
+      "step": 3974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222679508528595e-05,
+      "loss": 0.6231,
+      "step": 3975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221913749369918e-05,
+      "loss": 0.7062,
+      "step": 3976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215595186014807e-05,
+      "loss": 0.5967,
+      "step": 3977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212052585942935e-05,
+      "loss": 0.7622,
+      "step": 3978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208509693951083e-05,
+      "loss": 0.723,
+      "step": 3979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204966510506794e-05,
+      "loss": 0.6432,
+      "step": 3980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201423036077657e-05,
+      "loss": 0.6748,
+      "step": 3981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197879271131297e-05,
+      "loss": 0.6021,
+      "step": 3982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194335216135376e-05,
+      "loss": 0.7315,
+      "step": 3983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190790871557593e-05,
+      "loss": 0.5737,
+      "step": 3984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187246237865695e-05,
+      "loss": 0.6656,
+      "step": 3985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183701315527451e-05,
+      "loss": 0.644,
+      "step": 3986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180156105010688e-05,
+      "loss": 0.6462,
+      "step": 3987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217661060678325e-05,
+      "loss": 0.67,
+      "step": 3988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173064821313034e-05,
+      "loss": 0.6666,
+      "step": 3989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216951874906797e-05,
+      "loss": 0.6536,
+      "step": 3990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165972390516023e-05,
+      "loss": 0.72,
+      "step": 3991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21624257461252e-05,
+      "loss": 0.6881,
+      "step": 3992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158878816363544e-05,
+      "loss": 0.6607,
+      "step": 3993
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155331601699136e-05,
+      "loss": 0.6691,
+      "step": 3994
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151784102600095e-05,
+      "loss": 0.7253,
+      "step": 3995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214823631953458e-05,
+      "loss": 0.5621,
+      "step": 3996
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214468825297078e-05,
+      "loss": 0.7303,
+      "step": 3997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141139903376928e-05,
+      "loss": 0.6993,
+      "step": 3998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137591271221294e-05,
+      "loss": 0.7242,
+      "step": 3999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134042356972175e-05,
+      "loss": 0.6679,
+      "step": 4000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130493161097927e-05,
+      "loss": 0.687,
+      "step": 4001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212694368406692e-05,
+      "loss": 0.6398,
+      "step": 4002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123393926347575e-05,
+      "loss": 0.6321,
+      "step": 4003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119843888408346e-05,
+      "loss": 0.701,
+      "step": 4004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116293570717724e-05,
+      "loss": 0.5979,
+      "step": 4005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112742973744236e-05,
+      "loss": 0.7076,
+      "step": 4006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109192097956449e-05,
+      "loss": 0.6746,
+      "step": 4007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105640943822963e-05,
+      "loss": 0.7241,
+      "step": 4008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102089511812418e-05,
+      "loss": 0.7297,
+      "step": 4009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098537802393485e-05,
+      "loss": 0.6051,
+      "step": 4010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094985816034881e-05,
+      "loss": 0.6752,
+      "step": 4011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091433553205349e-05,
+      "loss": 0.7352,
+      "step": 4012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208788101437368e-05,
+      "loss": 0.7481,
+      "step": 4013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084328200008691e-05,
+      "loss": 0.5966,
+      "step": 4014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208077511057924e-05,
+      "loss": 0.6896,
+      "step": 4015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077221746554219e-05,
+      "loss": 0.639,
+      "step": 4016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073668108402565e-05,
+      "loss": 0.6166,
+      "step": 4017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070114196593237e-05,
+      "loss": 0.6507,
+      "step": 4018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206656001159524e-05,
+      "loss": 0.7589,
+      "step": 4019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063005553877612e-05,
+      "loss": 0.7183,
+      "step": 4020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059450823909424e-05,
+      "loss": 0.7178,
+      "step": 4021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055895822159795e-05,
+      "loss": 0.7008,
+      "step": 4022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052340549097867e-05,
+      "loss": 0.6758,
+      "step": 4023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204878500519282e-05,
+      "loss": 0.5424,
+      "step": 4024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204522919091387e-05,
+      "loss": 0.7744,
+      "step": 4025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041673106730276e-05,
+      "loss": 0.6016,
+      "step": 4026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038116753111325e-05,
+      "loss": 0.6313,
+      "step": 4027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034560130526341e-05,
+      "loss": 0.6173,
+      "step": 4028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031003239444681e-05,
+      "loss": 0.6238,
+      "step": 4029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027446080335746e-05,
+      "loss": 0.6568,
+      "step": 4030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023888653668963e-05,
+      "loss": 0.6342,
+      "step": 4031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020330959913802e-05,
+      "loss": 0.6653,
+      "step": 4032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201677299953976e-05,
+      "loss": 0.6776,
+      "step": 4033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201321477301638e-05,
+      "loss": 0.568,
+      "step": 4034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009656280813224e-05,
+      "loss": 0.7003,
+      "step": 4035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006097523399907e-05,
+      "loss": 0.7141,
+      "step": 4036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002538501246068e-05,
+      "loss": 0.6705,
+      "step": 4037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998979214821385e-05,
+      "loss": 0.693,
+      "step": 4038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1995419664595566e-05,
+      "loss": 0.6475,
+      "step": 4039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991859851038362e-05,
+      "loss": 0.6044,
+      "step": 4040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1988299774619552e-05,
+      "loss": 0.6524,
+      "step": 4041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984739435808952e-05,
+      "loss": 0.7611,
+      "step": 4042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981178835076414e-05,
+      "loss": 0.7061,
+      "step": 4043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977617972891821e-05,
+      "loss": 0.6703,
+      "step": 4044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974056849725091e-05,
+      "loss": 0.6746,
+      "step": 4045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970495466046184e-05,
+      "loss": 0.6924,
+      "step": 4046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196693382232508e-05,
+      "loss": 0.7207,
+      "step": 4047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196337191903181e-05,
+      "loss": 0.6347,
+      "step": 4048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959809756636429e-05,
+      "loss": 0.7014,
+      "step": 4049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956247335609022e-05,
+      "loss": 0.6328,
+      "step": 4050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952684656419722e-05,
+      "loss": 0.6458,
+      "step": 4051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949121719538686e-05,
+      "loss": 0.6433,
+      "step": 4052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945558525436106e-05,
+      "loss": 0.6806,
+      "step": 4053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941995074582206e-05,
+      "loss": 0.6637,
+      "step": 4054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938431367447257e-05,
+      "loss": 0.6187,
+      "step": 4055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934867404501544e-05,
+      "loss": 0.6847,
+      "step": 4056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931303186215401e-05,
+      "loss": 0.7056,
+      "step": 4057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927738713059191e-05,
+      "loss": 0.6946,
+      "step": 4058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924173985503306e-05,
+      "loss": 0.6479,
+      "step": 4059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920609004018182e-05,
+      "loss": 0.7797,
+      "step": 4060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917043769074277e-05,
+      "loss": 0.6861,
+      "step": 4061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913478281142089e-05,
+      "loss": 0.68,
+      "step": 4062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909912540692147e-05,
+      "loss": 0.6897,
+      "step": 4063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906346548195016e-05,
+      "loss": 0.6774,
+      "step": 4064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902780304121294e-05,
+      "loss": 0.6445,
+      "step": 4065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899213808941605e-05,
+      "loss": 0.6652,
+      "step": 4066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895647063126618e-05,
+      "loss": 0.64,
+      "step": 4067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892080067147029e-05,
+      "loss": 0.674,
+      "step": 4068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888512821473561e-05,
+      "loss": 0.5764,
+      "step": 4069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884945326576985e-05,
+      "loss": 0.7369,
+      "step": 4070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881377582928086e-05,
+      "loss": 0.6015,
+      "step": 4071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18778095909977e-05,
+      "loss": 0.6655,
+      "step": 4072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874241351256685e-05,
+      "loss": 0.6597,
+      "step": 4073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187067286417593e-05,
+      "loss": 0.649,
+      "step": 4074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867104130226363e-05,
+      "loss": 0.6566,
+      "step": 4075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863535149878946e-05,
+      "loss": 0.6387,
+      "step": 4076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859965923604669e-05,
+      "loss": 0.5836,
+      "step": 4077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856396451874552e-05,
+      "loss": 0.6501,
+      "step": 4078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185282673515965e-05,
+      "loss": 0.6559,
+      "step": 4079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849256773931058e-05,
+      "loss": 0.6309,
+      "step": 4080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845686568659892e-05,
+      "loss": 0.7197,
+      "step": 4081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842116119817301e-05,
+      "loss": 0.7476,
+      "step": 4082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838545427874475e-05,
+      "loss": 0.6935,
+      "step": 4083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834974493302629e-05,
+      "loss": 0.6797,
+      "step": 4084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183140331657301e-05,
+      "loss": 0.6443,
+      "step": 4085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827831898156905e-05,
+      "loss": 0.6087,
+      "step": 4086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824260238525622e-05,
+      "loss": 0.7303,
+      "step": 4087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1820688338150503e-05,
+      "loss": 0.6946,
+      "step": 4088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817116197502928e-05,
+      "loss": 0.6762,
+      "step": 4089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813543817054305e-05,
+      "loss": 0.8532,
+      "step": 4090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809971197276073e-05,
+      "loss": 0.5921,
+      "step": 4091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18063983386397e-05,
+      "loss": 0.6504,
+      "step": 4092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802825241616695e-05,
+      "loss": 0.5376,
+      "step": 4093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799251906678587e-05,
+      "loss": 0.6875,
+      "step": 4094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795678334296943e-05,
+      "loss": 0.6763,
+      "step": 4095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792104524943363e-05,
+      "loss": 0.6063,
+      "step": 4096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788530479089473e-05,
+      "loss": 0.6592,
+      "step": 4097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784956197206929e-05,
+      "loss": 0.7988,
+      "step": 4098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781381679767425e-05,
+      "loss": 0.6163,
+      "step": 4099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777806927242684e-05,
+      "loss": 0.6647,
+      "step": 4100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774231940104453e-05,
+      "loss": 0.6795,
+      "step": 4101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770656718824523e-05,
+      "loss": 0.6707,
+      "step": 4102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767081263874705e-05,
+      "loss": 0.6027,
+      "step": 4103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763505575726841e-05,
+      "loss": 0.7386,
+      "step": 4104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759929654852813e-05,
+      "loss": 0.665,
+      "step": 4105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756353501724526e-05,
+      "loss": 0.7692,
+      "step": 4106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752777116813916e-05,
+      "loss": 0.7999,
+      "step": 4107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749200500592954e-05,
+      "loss": 0.7093,
+      "step": 4108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745623653533633e-05,
+      "loss": 0.6101,
+      "step": 4109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742046576107988e-05,
+      "loss": 0.6983,
+      "step": 4110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738469268788074e-05,
+      "loss": 0.5837,
+      "step": 4111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734891732045984e-05,
+      "loss": 0.7247,
+      "step": 4112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731313966353834e-05,
+      "loss": 0.6232,
+      "step": 4113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727735972183783e-05,
+      "loss": 0.6674,
+      "step": 4114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724157750008003e-05,
+      "loss": 0.653,
+      "step": 4115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720579300298706e-05,
+      "loss": 0.6532,
+      "step": 4116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717000623528137e-05,
+      "loss": 0.639,
+      "step": 4117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171342172016856e-05,
+      "loss": 0.5614,
+      "step": 4118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709842590692282e-05,
+      "loss": 0.6159,
+      "step": 4119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706263235571627e-05,
+      "loss": 0.7349,
+      "step": 4120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170268365527896e-05,
+      "loss": 0.6737,
+      "step": 4121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699103850286668e-05,
+      "loss": 0.6566,
+      "step": 4122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695523821067172e-05,
+      "loss": 0.706,
+      "step": 4123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169194356809292e-05,
+      "loss": 0.7387,
+      "step": 4124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688363091836391e-05,
+      "loss": 0.7163,
+      "step": 4125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684782392770095e-05,
+      "loss": 0.6555,
+      "step": 4126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681201471366566e-05,
+      "loss": 0.6852,
+      "step": 4127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167762032809837e-05,
+      "loss": 0.6376,
+      "step": 4128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674038963438106e-05,
+      "loss": 0.5606,
+      "step": 4129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16704573778584e-05,
+      "loss": 0.6672,
+      "step": 4130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666875571831904e-05,
+      "loss": 0.6878,
+      "step": 4131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663293545831302e-05,
+      "loss": 0.685,
+      "step": 4132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659711300329306e-05,
+      "loss": 0.7316,
+      "step": 4133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656128835798661e-05,
+      "loss": 0.6675,
+      "step": 4134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652546152712132e-05,
+      "loss": 0.646,
+      "step": 4135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648963251542524e-05,
+      "loss": 0.6115,
+      "step": 4136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1645380132762657e-05,
+      "loss": 0.5896,
+      "step": 4137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641796796845396e-05,
+      "loss": 0.7165,
+      "step": 4138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638213244263623e-05,
+      "loss": 0.6822,
+      "step": 4139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634629475490254e-05,
+      "loss": 0.6675,
+      "step": 4140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631045490998227e-05,
+      "loss": 0.5522,
+      "step": 4141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1627461291260517e-05,
+      "loss": 0.6516,
+      "step": 4142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1623876876750122e-05,
+      "loss": 0.6564,
+      "step": 4143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620292247940068e-05,
+      "loss": 0.575,
+      "step": 4144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616707405303416e-05,
+      "loss": 0.6999,
+      "step": 4145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1613122349313244e-05,
+      "loss": 0.6869,
+      "step": 4146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609537080442667e-05,
+      "loss": 0.5724,
+      "step": 4147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605951599164826e-05,
+      "loss": 0.6822,
+      "step": 4148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602365905952893e-05,
+      "loss": 0.7073,
+      "step": 4149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598780001280058e-05,
+      "loss": 0.6026,
+      "step": 4150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159519388561955e-05,
+      "loss": 0.6805,
+      "step": 4151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1591607559444618e-05,
+      "loss": 0.5476,
+      "step": 4152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588021023228544e-05,
+      "loss": 0.6694,
+      "step": 4153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584434277444634e-05,
+      "loss": 0.6527,
+      "step": 4154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580847322566224e-05,
+      "loss": 0.6903,
+      "step": 4155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577260159066681e-05,
+      "loss": 0.7341,
+      "step": 4156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157367278741939e-05,
+      "loss": 0.7363,
+      "step": 4157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157008520809777e-05,
+      "loss": 0.7029,
+      "step": 4158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566497421575271e-05,
+      "loss": 0.6638,
+      "step": 4159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156290942832536e-05,
+      "loss": 0.6636,
+      "step": 4160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559321228821541e-05,
+      "loss": 0.7189,
+      "step": 4161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555732823537339e-05,
+      "loss": 0.7203,
+      "step": 4162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552144212946305e-05,
+      "loss": 0.6644,
+      "step": 4163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548555397522027e-05,
+      "loss": 0.6904,
+      "step": 4164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544966377738111e-05,
+      "loss": 0.6857,
+      "step": 4165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541377154068191e-05,
+      "loss": 0.6078,
+      "step": 4166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537787726985932e-05,
+      "loss": 0.672,
+      "step": 4167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534198096965023e-05,
+      "loss": 0.5536,
+      "step": 4168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153060826447918e-05,
+      "loss": 0.7001,
+      "step": 4169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527018230002143e-05,
+      "loss": 0.7383,
+      "step": 4170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523427994007683e-05,
+      "loss": 0.6803,
+      "step": 4171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519837556969593e-05,
+      "loss": 0.7417,
+      "step": 4172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516246919361703e-05,
+      "loss": 0.6519,
+      "step": 4173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512656081657854e-05,
+      "loss": 0.7434,
+      "step": 4174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1509065044331928e-05,
+      "loss": 0.694,
+      "step": 4175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505473807857824e-05,
+      "loss": 0.6384,
+      "step": 4176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501882372709467e-05,
+      "loss": 0.5918,
+      "step": 4177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498290739360815e-05,
+      "loss": 0.6954,
+      "step": 4178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149469890828585e-05,
+      "loss": 0.6685,
+      "step": 4179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491106879958574e-05,
+      "loss": 0.6404,
+      "step": 4180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487514654853023e-05,
+      "loss": 0.7012,
+      "step": 4181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148392223344325e-05,
+      "loss": 0.675,
+      "step": 4182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1480329616203344e-05,
+      "loss": 0.7412,
+      "step": 4183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476736803607416e-05,
+      "loss": 0.6258,
+      "step": 4184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14731437961296e-05,
+      "loss": 0.7061,
+      "step": 4185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146955059424406e-05,
+      "loss": 0.6783,
+      "step": 4186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465957198424977e-05,
+      "loss": 0.6727,
+      "step": 4187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462363609146572e-05,
+      "loss": 0.7347,
+      "step": 4188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145876982688308e-05,
+      "loss": 0.6421,
+      "step": 4189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455175852108767e-05,
+      "loss": 0.663,
+      "step": 4190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451581685297915e-05,
+      "loss": 0.6698,
+      "step": 4191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447987326924845e-05,
+      "loss": 0.6738,
+      "step": 4192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14443927774639e-05,
+      "loss": 0.6788,
+      "step": 4193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144079803738944e-05,
+      "loss": 0.6817,
+      "step": 4194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437203107175857e-05,
+      "loss": 0.6354,
+      "step": 4195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433607987297567e-05,
+      "loss": 0.5927,
+      "step": 4196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430012678229008e-05,
+      "loss": 0.7288,
+      "step": 4197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426417180444646e-05,
+      "loss": 0.6655,
+      "step": 4198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422821494418977e-05,
+      "loss": 0.8405,
+      "step": 4199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419225620626506e-05,
+      "loss": 0.6276,
+      "step": 4200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415629559541783e-05,
+      "loss": 0.6582,
+      "step": 4201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412033311639366e-05,
+      "loss": 0.6921,
+      "step": 4202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140843687739385e-05,
+      "loss": 0.6653,
+      "step": 4203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404840257279843e-05,
+      "loss": 0.7303,
+      "step": 4204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1401243451771988e-05,
+      "loss": 0.6596,
+      "step": 4205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397646461344946e-05,
+      "loss": 0.6411,
+      "step": 4206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394049286473402e-05,
+      "loss": 0.7086,
+      "step": 4207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1390451927632076e-05,
+      "loss": 0.7561,
+      "step": 4208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386854385295694e-05,
+      "loss": 0.6365,
+      "step": 4209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383256659939019e-05,
+      "loss": 0.7454,
+      "step": 4210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379658752036842e-05,
+      "loss": 0.6369,
+      "step": 4211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376060662063962e-05,
+      "loss": 0.642,
+      "step": 4212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372462390495215e-05,
+      "loss": 0.6294,
+      "step": 4213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368863937805461e-05,
+      "loss": 0.6321,
+      "step": 4214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136526530446958e-05,
+      "loss": 0.658,
+      "step": 4215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361666490962468e-05,
+      "loss": 0.6317,
+      "step": 4216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358067497759059e-05,
+      "loss": 0.5729,
+      "step": 4217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354468325334307e-05,
+      "loss": 0.5661,
+      "step": 4218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350868974163183e-05,
+      "loss": 0.6901,
+      "step": 4219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347269444720687e-05,
+      "loss": 0.723,
+      "step": 4220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343669737481844e-05,
+      "loss": 0.7005,
+      "step": 4221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340069852921693e-05,
+      "loss": 0.696,
+      "step": 4222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336469791515312e-05,
+      "loss": 0.5774,
+      "step": 4223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133286955373779e-05,
+      "loss": 0.7281,
+      "step": 4224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329269140064244e-05,
+      "loss": 0.7237,
+      "step": 4225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325668550969806e-05,
+      "loss": 0.7008,
+      "step": 4226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322067786929647e-05,
+      "loss": 0.6822,
+      "step": 4227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318466848418948e-05,
+      "loss": 0.6148,
+      "step": 4228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131486573591292e-05,
+      "loss": 0.6402,
+      "step": 4229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311264449886794e-05,
+      "loss": 0.6968,
+      "step": 4230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130766299081582e-05,
+      "loss": 0.7177,
+      "step": 4231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130406135917528e-05,
+      "loss": 0.5893,
+      "step": 4232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300459555440474e-05,
+      "loss": 0.6529,
+      "step": 4233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129685758008672e-05,
+      "loss": 0.6712,
+      "step": 4234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293255433589367e-05,
+      "loss": 0.619,
+      "step": 4235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289653116423778e-05,
+      "loss": 0.6129,
+      "step": 4236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286050629065351e-05,
+      "loss": 0.6043,
+      "step": 4237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282447971989491e-05,
+      "loss": 0.6495,
+      "step": 4238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278845145671641e-05,
+      "loss": 0.6643,
+      "step": 4239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275242150587254e-05,
+      "loss": 0.6815,
+      "step": 4240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271638987211809e-05,
+      "loss": 0.6286,
+      "step": 4241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268035656020807e-05,
+      "loss": 0.6882,
+      "step": 4242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126443215748978e-05,
+      "loss": 0.7116,
+      "step": 4243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260828492094264e-05,
+      "loss": 0.6354,
+      "step": 4244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257224660309832e-05,
+      "loss": 0.682,
+      "step": 4245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253620662612072e-05,
+      "loss": 0.6175,
+      "step": 4246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250016499476602e-05,
+      "loss": 0.6934,
+      "step": 4247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124641217137905e-05,
+      "loss": 0.7663,
+      "step": 4248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242807678795076e-05,
+      "loss": 0.8171,
+      "step": 4249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239203022200353e-05,
+      "loss": 0.5852,
+      "step": 4250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235598202070581e-05,
+      "loss": 0.7066,
+      "step": 4251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231993218881485e-05,
+      "loss": 0.6916,
+      "step": 4252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228388073108802e-05,
+      "loss": 0.6566,
+      "step": 4253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224782765228295e-05,
+      "loss": 0.6643,
+      "step": 4254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221177295715755e-05,
+      "loss": 0.6861,
+      "step": 4255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121757166504698e-05,
+      "loss": 0.7147,
+      "step": 4256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213965873697808e-05,
+      "loss": 0.5872,
+      "step": 4257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1210359922144078e-05,
+      "loss": 0.6946,
+      "step": 4258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206753810861668e-05,
+      "loss": 0.7181,
+      "step": 4259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203147540326462e-05,
+      "loss": 0.7329,
+      "step": 4260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199541111014373e-05,
+      "loss": 0.6465,
+      "step": 4261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195934523401342e-05,
+      "loss": 0.6671,
+      "step": 4262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192327777963313e-05,
+      "loss": 0.7445,
+      "step": 4263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188720875176264e-05,
+      "loss": 0.6257,
+      "step": 4264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185113815516193e-05,
+      "loss": 0.6587,
+      "step": 4265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181506599459113e-05,
+      "loss": 0.7223,
+      "step": 4266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177899227481061e-05,
+      "loss": 0.7493,
+      "step": 4267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11742917000581e-05,
+      "loss": 0.6504,
+      "step": 4268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11706840176663e-05,
+      "loss": 0.7769,
+      "step": 4269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167076180781764e-05,
+      "loss": 0.7525,
+      "step": 4270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163468189880606e-05,
+      "loss": 0.6334,
+      "step": 4271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159860045438973e-05,
+      "loss": 0.7018,
+      "step": 4272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156251747933016e-05,
+      "loss": 0.6232,
+      "step": 4273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115264329783892e-05,
+      "loss": 0.6193,
+      "step": 4274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149034695632885e-05,
+      "loss": 0.6579,
+      "step": 4275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145425941791126e-05,
+      "loss": 0.6876,
+      "step": 4276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141817036789889e-05,
+      "loss": 0.6804,
+      "step": 4277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138207981105427e-05,
+      "loss": 0.6715,
+      "step": 4278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1134598775214023e-05,
+      "loss": 0.6867,
+      "step": 4279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130989419591976e-05,
+      "loss": 0.7474,
+      "step": 4280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127379914715607e-05,
+      "loss": 0.6693,
+      "step": 4281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123770261061251e-05,
+      "loss": 0.6123,
+      "step": 4282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120160459105269e-05,
+      "loss": 0.6853,
+      "step": 4283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111655050932404e-05,
+      "loss": 0.6638,
+      "step": 4284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111294041219396e-05,
+      "loss": 0.7297,
+      "step": 4285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1109330168191445e-05,
+      "loss": 0.6588,
+      "step": 4286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105719777792937e-05,
+      "loss": 0.6648,
+      "step": 4287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102109241474883e-05,
+      "loss": 0.6325,
+      "step": 4288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098498559713766e-05,
+      "loss": 0.706,
+      "step": 4289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094887732986072e-05,
+      "loss": 0.6196,
+      "step": 4290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109127676176832e-05,
+      "loss": 0.6703,
+      "step": 4291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087665646537043e-05,
+      "loss": 0.5866,
+      "step": 4292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108405438776879e-05,
+      "loss": 0.757,
+      "step": 4293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080442985940135e-05,
+      "loss": 0.7161,
+      "step": 4294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076831441527657e-05,
+      "loss": 0.6799,
+      "step": 4295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073219755007977e-05,
+      "loss": 0.7285,
+      "step": 4296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1069607926857716e-05,
+      "loss": 0.5825,
+      "step": 4297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106599595755352e-05,
+      "loss": 0.7306,
+      "step": 4298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062383847572048e-05,
+      "loss": 0.6601,
+      "step": 4299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105877159738999e-05,
+      "loss": 0.6828,
+      "step": 4300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105515920748404e-05,
+      "loss": 0.6527,
+      "step": 4301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1051546678330927e-05,
+      "loss": 0.6457,
+      "step": 4302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047934010407385e-05,
+      "loss": 0.7145,
+      "step": 4303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044321204190165e-05,
+      "loss": 0.6246,
+      "step": 4304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040708260156045e-05,
+      "loss": 0.6538,
+      "step": 4305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037095178781822e-05,
+      "loss": 0.6751,
+      "step": 4306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033481960544302e-05,
+      "loss": 0.5936,
+      "step": 4307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029868605920313e-05,
+      "loss": 0.6687,
+      "step": 4308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026255115386706e-05,
+      "loss": 0.7068,
+      "step": 4309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022641489420342e-05,
+      "loss": 0.6464,
+      "step": 4310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101902772849811e-05,
+      "loss": 0.7093,
+      "step": 4311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015413833096902e-05,
+      "loss": 0.6665,
+      "step": 4312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011799803693642e-05,
+      "loss": 0.7121,
+      "step": 4313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008185640765262e-05,
+      "loss": 0.6719,
+      "step": 4314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004571344788719e-05,
+      "loss": 0.7281,
+      "step": 4315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000956916240985e-05,
+      "loss": 0.6595,
+      "step": 4316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997342355599045e-05,
+      "loss": 0.6386,
+      "step": 4317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993727663339905e-05,
+      "loss": 0.6764,
+      "step": 4318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990112839940594e-05,
+      "loss": 0.7306,
+      "step": 4319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986497885878145e-05,
+      "loss": 0.6882,
+      "step": 4320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982882801629622e-05,
+      "loss": 0.6091,
+      "step": 4321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0979267587672097e-05,
+      "loss": 0.7492,
+      "step": 4322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975652244482666e-05,
+      "loss": 0.6373,
+      "step": 4323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972036772538432e-05,
+      "loss": 0.6572,
+      "step": 4324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968421172316526e-05,
+      "loss": 0.7639,
+      "step": 4325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964805444294091e-05,
+      "loss": 0.6869,
+      "step": 4326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961189588948284e-05,
+      "loss": 0.6162,
+      "step": 4327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957573606756286e-05,
+      "loss": 0.5561,
+      "step": 4328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953957498195289e-05,
+      "loss": 0.6941,
+      "step": 4329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950341263742502e-05,
+      "loss": 0.6355,
+      "step": 4330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946724903875155e-05,
+      "loss": 0.7119,
+      "step": 4331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943108419070488e-05,
+      "loss": 0.6757,
+      "step": 4332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939491809805762e-05,
+      "loss": 0.628,
+      "step": 4333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935875076558258e-05,
+      "loss": 0.6504,
+      "step": 4334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932258219805259e-05,
+      "loss": 0.6449,
+      "step": 4335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092864124002408e-05,
+      "loss": 0.7051,
+      "step": 4336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925024137692052e-05,
+      "loss": 0.66,
+      "step": 4337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921406913286506e-05,
+      "loss": 0.6982,
+      "step": 4338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917789567284805e-05,
+      "loss": 0.6259,
+      "step": 4339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914172100164322e-05,
+      "loss": 0.7391,
+      "step": 4340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910554512402447e-05,
+      "loss": 0.5785,
+      "step": 4341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906936804476586e-05,
+      "loss": 0.715,
+      "step": 4342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903318976864159e-05,
+      "loss": 0.7227,
+      "step": 4343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08997010300426e-05,
+      "loss": 0.6773,
+      "step": 4344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896082964489368e-05,
+      "loss": 0.6481,
+      "step": 4345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892464780681928e-05,
+      "loss": 0.669,
+      "step": 4346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888846479097768e-05,
+      "loss": 0.5907,
+      "step": 4347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885228060214383e-05,
+      "loss": 0.6785,
+      "step": 4348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881609524509292e-05,
+      "loss": 0.6577,
+      "step": 4349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877990872460022e-05,
+      "loss": 0.577,
+      "step": 4350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874372104544122e-05,
+      "loss": 0.5349,
+      "step": 4351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870753221239154e-05,
+      "loss": 0.7206,
+      "step": 4352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867134223022688e-05,
+      "loss": 0.6596,
+      "step": 4353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863515110372325e-05,
+      "loss": 0.6808,
+      "step": 4354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859895883765663e-05,
+      "loss": 0.6644,
+      "step": 4355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085627654368033e-05,
+      "loss": 0.6666,
+      "step": 4356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852657090593961e-05,
+      "loss": 0.669,
+      "step": 4357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0849037524984209e-05,
+      "loss": 0.7271,
+      "step": 4358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845417847328736e-05,
+      "loss": 0.6022,
+      "step": 4359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841798058105228e-05,
+      "loss": 0.6987,
+      "step": 4360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083817815779138e-05,
+      "loss": 0.673,
+      "step": 4361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08345581468649e-05,
+      "loss": 0.6115,
+      "step": 4362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830938025803518e-05,
+      "loss": 0.5906,
+      "step": 4363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827317795084966e-05,
+      "loss": 0.6492,
+      "step": 4364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823697455187008e-05,
+      "loss": 0.7038,
+      "step": 4365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820077006587408e-05,
+      "loss": 0.6306,
+      "step": 4366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081645644976395e-05,
+      "loss": 0.5605,
+      "step": 4367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812835785194431e-05,
+      "loss": 0.6993,
+      "step": 4368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080921501335666e-05,
+      "loss": 0.689,
+      "step": 4369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805594134728468e-05,
+      "loss": 0.6154,
+      "step": 4370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801973149787692e-05,
+      "loss": 0.6702,
+      "step": 4371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798352059012183e-05,
+      "loss": 0.6534,
+      "step": 4372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794730862879813e-05,
+      "loss": 0.6301,
+      "step": 4373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079110956186846e-05,
+      "loss": 0.6213,
+      "step": 4374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787488156456025e-05,
+      "loss": 0.623,
+      "step": 4375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783866647120416e-05,
+      "loss": 0.629,
+      "step": 4376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780245034339551e-05,
+      "loss": 0.6191,
+      "step": 4377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776623318591373e-05,
+      "loss": 0.7182,
+      "step": 4378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773001500353825e-05,
+      "loss": 0.637,
+      "step": 4379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0769379580104878e-05,
+      "loss": 0.6983,
+      "step": 4380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765757558322504e-05,
+      "loss": 0.6634,
+      "step": 4381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762135435484699e-05,
+      "loss": 0.6641,
+      "step": 4382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758513212069463e-05,
+      "loss": 0.706,
+      "step": 4383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754890888554814e-05,
+      "loss": 0.6423,
+      "step": 4384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751268465418784e-05,
+      "loss": 0.6841,
+      "step": 4385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747645943139418e-05,
+      "loss": 0.6626,
+      "step": 4386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744023322194767e-05,
+      "loss": 0.6383,
+      "step": 4387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740400603062906e-05,
+      "loss": 0.6778,
+      "step": 4388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736777786221913e-05,
+      "loss": 0.6454,
+      "step": 4389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733154872149886e-05,
+      "loss": 0.6405,
+      "step": 4390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729531861324939e-05,
+      "loss": 0.6378,
+      "step": 4391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725908754225186e-05,
+      "loss": 0.6293,
+      "step": 4392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072228555132876e-05,
+      "loss": 0.7149,
+      "step": 4393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718662253113815e-05,
+      "loss": 0.7194,
+      "step": 4394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715038860058502e-05,
+      "loss": 0.6279,
+      "step": 4395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711415372640998e-05,
+      "loss": 0.6896,
+      "step": 4396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707791791339487e-05,
+      "loss": 0.7162,
+      "step": 4397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704168116632162e-05,
+      "loss": 0.6897,
+      "step": 4398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700544348997233e-05,
+      "loss": 0.6405,
+      "step": 4399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696920488912923e-05,
+      "loss": 0.6149,
+      "step": 4400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693296536857465e-05,
+      "loss": 0.6323,
+      "step": 4401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689672493309104e-05,
+      "loss": 0.7279,
+      "step": 4402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686048358746096e-05,
+      "loss": 0.6992,
+      "step": 4403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682424133646712e-05,
+      "loss": 0.729,
+      "step": 4404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678799818489232e-05,
+      "loss": 0.6162,
+      "step": 4405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675175413751953e-05,
+      "loss": 0.6453,
+      "step": 4406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671550919913174e-05,
+      "loss": 0.6843,
+      "step": 4407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667926337451217e-05,
+      "loss": 0.6392,
+      "step": 4408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664301666844408e-05,
+      "loss": 0.7229,
+      "step": 4409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066067690857109e-05,
+      "loss": 0.6652,
+      "step": 4410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657052063109615e-05,
+      "loss": 0.6648,
+      "step": 4411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065342713093834e-05,
+      "loss": 0.6275,
+      "step": 4412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649802112535647e-05,
+      "loss": 0.669,
+      "step": 4413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646177008379919e-05,
+      "loss": 0.6463,
+      "step": 4414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642551818949556e-05,
+      "loss": 0.5984,
+      "step": 4415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063892654472296e-05,
+      "loss": 0.6433,
+      "step": 4416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635301186178559e-05,
+      "loss": 0.6069,
+      "step": 4417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631675743794778e-05,
+      "loss": 0.5808,
+      "step": 4418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628050218050063e-05,
+      "loss": 0.5887,
+      "step": 4419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624424609422867e-05,
+      "loss": 0.6681,
+      "step": 4420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0620798918391652e-05,
+      "loss": 0.6844,
+      "step": 4421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617173145434893e-05,
+      "loss": 0.7104,
+      "step": 4422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613547291031077e-05,
+      "loss": 0.6367,
+      "step": 4423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0609921355658698e-05,
+      "loss": 0.6324,
+      "step": 4424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606295339796269e-05,
+      "loss": 0.6931,
+      "step": 4425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602669243922298e-05,
+      "loss": 0.5787,
+      "step": 4426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599043068515325e-05,
+      "loss": 0.6856,
+      "step": 4427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595416814053878e-05,
+      "loss": 0.705,
+      "step": 4428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591790481016514e-05,
+      "loss": 0.6014,
+      "step": 4429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058816406988179e-05,
+      "loss": 0.7764,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584537581128276e-05,
+      "loss": 0.5735,
+      "step": 4431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580911015234552e-05,
+      "loss": 0.5951,
+      "step": 4432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577284372679207e-05,
+      "loss": 0.6245,
+      "step": 4433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573657653940842e-05,
+      "loss": 0.7558,
+      "step": 4434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570030859498071e-05,
+      "loss": 0.6504,
+      "step": 4435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056640398982951e-05,
+      "loss": 0.6347,
+      "step": 4436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562777045413792e-05,
+      "loss": 0.6632,
+      "step": 4437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559150026729557e-05,
+      "loss": 0.6816,
+      "step": 4438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555522934255456e-05,
+      "loss": 0.6879,
+      "step": 4439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551895768470146e-05,
+      "loss": 0.6512,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05482685298523e-05,
+      "loss": 0.7758,
+      "step": 4441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0544641218880594e-05,
+      "loss": 0.6428,
+      "step": 4442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541013836033716e-05,
+      "loss": 0.7805,
+      "step": 4443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537386381790367e-05,
+      "loss": 0.6638,
+      "step": 4444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533758856629255e-05,
+      "loss": 0.5686,
+      "step": 4445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530131261029097e-05,
+      "loss": 0.596,
+      "step": 4446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526503595468617e-05,
+      "loss": 0.6354,
+      "step": 4447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052287586042655e-05,
+      "loss": 0.6246,
+      "step": 4448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519248056381644e-05,
+      "loss": 0.5668,
+      "step": 4449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051562018381265e-05,
+      "loss": 0.7431,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511992243198335e-05,
+      "loss": 0.5643,
+      "step": 4451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050836423501746e-05,
+      "loss": 0.7219,
+      "step": 4452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504736159748818e-05,
+      "loss": 0.7214,
+      "step": 4453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501108017871193e-05,
+      "loss": 0.6113,
+      "step": 4454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497479809863386e-05,
+      "loss": 0.7293,
+      "step": 4455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493851536204203e-05,
+      "loss": 0.569,
+      "step": 4456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490223197372454e-05,
+      "loss": 0.7851,
+      "step": 4457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486594793846972e-05,
+      "loss": 0.7278,
+      "step": 4458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482966326106585e-05,
+      "loss": 0.6669,
+      "step": 4459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479337794630135e-05,
+      "loss": 0.7305,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475709199896474e-05,
+      "loss": 0.6854,
+      "step": 4461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472080542384457e-05,
+      "loss": 0.6959,
+      "step": 4462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468451822572949e-05,
+      "loss": 0.7003,
+      "step": 4463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464823040940832e-05,
+      "loss": 0.619,
+      "step": 4464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461194197966985e-05,
+      "loss": 0.5019,
+      "step": 4465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457565294130294e-05,
+      "loss": 0.7199,
+      "step": 4466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453936329909662e-05,
+      "loss": 0.6187,
+      "step": 4467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450307305783998e-05,
+      "loss": 0.6837,
+      "step": 4468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446678222232214e-05,
+      "loss": 0.6811,
+      "step": 4469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443049079733231e-05,
+      "loss": 0.5661,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439419878765984e-05,
+      "loss": 0.641,
+      "step": 4471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435790619809405e-05,
+      "loss": 0.6309,
+      "step": 4472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432161303342448e-05,
+      "loss": 0.5668,
+      "step": 4473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428531929844062e-05,
+      "loss": 0.6521,
+      "step": 4474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424902499793206e-05,
+      "loss": 0.5575,
+      "step": 4475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042127301366885e-05,
+      "loss": 0.7246,
+      "step": 4476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417643471949968e-05,
+      "loss": 0.7549,
+      "step": 4477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414013875115545e-05,
+      "loss": 0.629,
+      "step": 4478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410384223644575e-05,
+      "loss": 0.674,
+      "step": 4479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0406754518016047e-05,
+      "loss": 0.6873,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403124758708974e-05,
+      "loss": 0.6872,
+      "step": 4481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039949494620236e-05,
+      "loss": 0.6777,
+      "step": 4482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039586508097523e-05,
+      "loss": 0.7159,
+      "step": 4483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039223516350661e-05,
+      "loss": 0.7316,
+      "step": 4484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388605194275529e-05,
+      "loss": 0.6305,
+      "step": 4485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038497517376103e-05,
+      "loss": 0.5983,
+      "step": 4486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381345102442153e-05,
+      "loss": 0.5987,
+      "step": 4487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037771498079796e-05,
+      "loss": 0.6716,
+      "step": 4488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374084809307502e-05,
+      "loss": 0.5732,
+      "step": 4489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370454588449849e-05,
+      "loss": 0.7085,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366824318704076e-05,
+      "loss": 0.731,
+      "step": 4491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363194000549256e-05,
+      "loss": 0.682,
+      "step": 4492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359563634464485e-05,
+      "loss": 0.676,
+      "step": 4493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355933220928845e-05,
+      "loss": 0.7044,
+      "step": 4494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352302760421438e-05,
+      "loss": 0.7804,
+      "step": 4495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348672253421366e-05,
+      "loss": 0.6967,
+      "step": 4496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345041700407741e-05,
+      "loss": 0.7127,
+      "step": 4497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034141110185968e-05,
+      "loss": 0.626,
+      "step": 4498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337780458256305e-05,
+      "loss": 0.6043,
+      "step": 4499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334149770076747e-05,
+      "loss": 0.6501,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330519037800137e-05,
+      "loss": 0.7173,
+      "step": 4501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326888261905616e-05,
+      "loss": 0.5792,
+      "step": 4502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323257442872332e-05,
+      "loss": 0.7457,
+      "step": 4503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319626581179434e-05,
+      "loss": 0.6978,
+      "step": 4504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315995677306078e-05,
+      "loss": 0.6508,
+      "step": 4505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031236473173143e-05,
+      "loss": 0.7847,
+      "step": 4506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308733744934657e-05,
+      "loss": 0.7158,
+      "step": 4507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305102717394935e-05,
+      "loss": 0.5835,
+      "step": 4508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301471649591441e-05,
+      "loss": 0.6632,
+      "step": 4509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297840542003362e-05,
+      "loss": 0.6291,
+      "step": 4510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294209395109882e-05,
+      "loss": 0.7208,
+      "step": 4511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290578209390201e-05,
+      "loss": 0.7192,
+      "step": 4512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286946985323516e-05,
+      "loss": 0.5516,
+      "step": 4513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283315723389035e-05,
+      "loss": 0.6596,
+      "step": 4514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279684424065965e-05,
+      "loss": 0.6562,
+      "step": 4515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276053087833527e-05,
+      "loss": 0.6926,
+      "step": 4516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027242171517093e-05,
+      "loss": 0.6224,
+      "step": 4517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026879030655741e-05,
+      "loss": 0.8227,
+      "step": 4518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265158862472191e-05,
+      "loss": 0.6191,
+      "step": 4519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261527383394509e-05,
+      "loss": 0.6822,
+      "step": 4520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257895869803599e-05,
+      "loss": 0.715,
+      "step": 4521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254264322178708e-05,
+      "loss": 0.5991,
+      "step": 4522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0250632740999085e-05,
+      "loss": 0.6832,
+      "step": 4523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247001126743976e-05,
+      "loss": 0.7427,
+      "step": 4524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243369479892641e-05,
+      "loss": 0.6329,
+      "step": 4525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0239737800924344e-05,
+      "loss": 0.6717,
+      "step": 4526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236106090318347e-05,
+      "loss": 0.6484,
+      "step": 4527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232474348553922e-05,
+      "loss": 0.6987,
+      "step": 4528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228842576110338e-05,
+      "loss": 0.6607,
+      "step": 4529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225210773466876e-05,
+      "loss": 0.7125,
+      "step": 4530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221578941102814e-05,
+      "loss": 0.6361,
+      "step": 4531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0217947079497442e-05,
+      "loss": 0.6379,
+      "step": 4532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214315189130046e-05,
+      "loss": 0.696,
+      "step": 4533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210683270479918e-05,
+      "loss": 0.673,
+      "step": 4534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207051324026357e-05,
+      "loss": 0.5039,
+      "step": 4535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203419350248664e-05,
+      "loss": 0.575,
+      "step": 4536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199787349626142e-05,
+      "loss": 0.6635,
+      "step": 4537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196155322638101e-05,
+      "loss": 0.6546,
+      "step": 4538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192523269763848e-05,
+      "loss": 0.6624,
+      "step": 4539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01888911914827e-05,
+      "loss": 0.6765,
+      "step": 4540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185259088273969e-05,
+      "loss": 0.6711,
+      "step": 4541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181626960616986e-05,
+      "loss": 0.7115,
+      "step": 4542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177994808991068e-05,
+      "loss": 0.6127,
+      "step": 4543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174362633875546e-05,
+      "loss": 0.7367,
+      "step": 4544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017073043574975e-05,
+      "loss": 0.6308,
+      "step": 4545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016709821509301e-05,
+      "loss": 0.6415,
+      "step": 4546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016346597238467e-05,
+      "loss": 0.6916,
+      "step": 4547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159833708104063e-05,
+      "loss": 0.8036,
+      "step": 4548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156201422730534e-05,
+      "loss": 0.6724,
+      "step": 4549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015256911674343e-05,
+      "loss": 0.7169,
+      "step": 4550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148936790622093e-05,
+      "loss": 0.6579,
+      "step": 4551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145304444845876e-05,
+      "loss": 0.645,
+      "step": 4552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141672079894138e-05,
+      "loss": 0.7311,
+      "step": 4553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013803969624623e-05,
+      "loss": 0.7358,
+      "step": 4554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134407294381508e-05,
+      "loss": 0.6805,
+      "step": 4555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130774874779333e-05,
+      "loss": 0.7094,
+      "step": 4556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127142437919071e-05,
+      "loss": 0.6425,
+      "step": 4557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123509984280086e-05,
+      "loss": 0.591,
+      "step": 4558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119877514341748e-05,
+      "loss": 0.6302,
+      "step": 4559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116245028583418e-05,
+      "loss": 0.7325,
+      "step": 4560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112612527484476e-05,
+      "loss": 0.6701,
+      "step": 4561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010898001152429e-05,
+      "loss": 0.612,
+      "step": 4562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105347481182241e-05,
+      "loss": 0.6894,
+      "step": 4563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101714936937705e-05,
+      "loss": 0.5989,
+      "step": 4564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098082379270057e-05,
+      "loss": 0.6734,
+      "step": 4565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094449808658684e-05,
+      "loss": 0.628,
+      "step": 4566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090817225582965e-05,
+      "loss": 0.7038,
+      "step": 4567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087184630522287e-05,
+      "loss": 0.6904,
+      "step": 4568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083552023956031e-05,
+      "loss": 0.5973,
+      "step": 4569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079919406363589e-05,
+      "loss": 0.6947,
+      "step": 4570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076286778224348e-05,
+      "loss": 0.7811,
+      "step": 4571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0072654140017704e-05,
+      "loss": 0.5999,
+      "step": 4572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069021492223042e-05,
+      "loss": 0.5956,
+      "step": 4573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065388835319756e-05,
+      "loss": 0.6147,
+      "step": 4574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006175616978724e-05,
+      "loss": 0.6821,
+      "step": 4575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058123496104892e-05,
+      "loss": 0.6306,
+      "step": 4576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005449081475211e-05,
+      "loss": 0.6437,
+      "step": 4577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050858126208284e-05,
+      "loss": 0.6289,
+      "step": 4578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004722543095282e-05,
+      "loss": 0.7634,
+      "step": 4579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043592729465108e-05,
+      "loss": 0.6398,
+      "step": 4580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003996002222456e-05,
+      "loss": 0.6708,
+      "step": 4581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003632730971057e-05,
+      "loss": 0.6316,
+      "step": 4582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032694592402541e-05,
+      "loss": 0.7091,
+      "step": 4583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029061870779875e-05,
+      "loss": 0.6808,
+      "step": 4584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002542914532197e-05,
+      "loss": 0.5517,
+      "step": 4585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021796416508237e-05,
+      "loss": 0.6871,
+      "step": 4586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018163684818076e-05,
+      "loss": 0.6763,
+      "step": 4587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014530950730889e-05,
+      "loss": 0.6395,
+      "step": 4588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010898214726087e-05,
+      "loss": 0.6392,
+      "step": 4589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0007265477283063e-05,
+      "loss": 0.5733,
+      "step": 4590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003632738881235e-05,
+      "loss": 0.7219,
+      "step": 4591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7119,
+      "step": 4592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996367261118768e-06,
+      "loss": 0.7636,
+      "step": 4593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992734522716939e-06,
+      "loss": 0.594,
+      "step": 4594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98910178527392e-06,
+      "loss": 0.6649,
+      "step": 4595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985469049269114e-06,
+      "loss": 0.6309,
+      "step": 4596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981836315181927e-06,
+      "loss": 0.6979,
+      "step": 4597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978203583491765e-06,
+      "loss": 0.6486,
+      "step": 4598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97457085467803e-06,
+      "loss": 0.6175,
+      "step": 4599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97093812922013e-06,
+      "loss": 0.6623,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967305407597462e-06,
+      "loss": 0.6691,
+      "step": 4601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963672690289432e-06,
+      "loss": 0.7212,
+      "step": 4602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960039977775442e-06,
+      "loss": 0.6031,
+      "step": 4603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956407270534894e-06,
+      "loss": 0.5896,
+      "step": 4604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.952774569047184e-06,
+      "loss": 0.6605,
+      "step": 4605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94914187379172e-06,
+      "loss": 0.669,
+      "step": 4606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945509185247894e-06,
+      "loss": 0.5803,
+      "step": 4607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941876503895108e-06,
+      "loss": 0.7282,
+      "step": 4608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93824383021276e-06,
+      "loss": 0.6904,
+      "step": 4609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934611164680247e-06,
+      "loss": 0.7042,
+      "step": 4610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930978507776962e-06,
+      "loss": 0.7024,
+      "step": 4611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9273458599823e-06,
+      "loss": 0.7714,
+      "step": 4612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923713221775653e-06,
+      "loss": 0.714,
+      "step": 4613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920080593636413e-06,
+      "loss": 0.6628,
+      "step": 4614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916447976043972e-06,
+      "loss": 0.6784,
+      "step": 4615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912815369477718e-06,
+      "loss": 0.6565,
+      "step": 4616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909182774417035e-06,
+      "loss": 0.6305,
+      "step": 4617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905550191341316e-06,
+      "loss": 0.6287,
+      "step": 4618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901917620729942e-06,
+      "loss": 0.7285,
+      "step": 4619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898285063062298e-06,
+      "loss": 0.7509,
+      "step": 4620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89465251881776e-06,
+      "loss": 0.6624,
+      "step": 4621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891019988475712e-06,
+      "loss": 0.7683,
+      "step": 4622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887387472515528e-06,
+      "loss": 0.5907,
+      "step": 4623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883754971416586e-06,
+      "loss": 0.5918,
+      "step": 4624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880122485658259e-06,
+      "loss": 0.6129,
+      "step": 4625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876490015719919e-06,
+      "loss": 0.7381,
+      "step": 4626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872857562080929e-06,
+      "loss": 0.6654,
+      "step": 4627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869225125220669e-06,
+      "loss": 0.7098,
+      "step": 4628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865592705618497e-06,
+      "loss": 0.6228,
+      "step": 4629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861960303753774e-06,
+      "loss": 0.6891,
+      "step": 4630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858327920105865e-06,
+      "loss": 0.6481,
+      "step": 4631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854695555154127e-06,
+      "loss": 0.6071,
+      "step": 4632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85106320937791e-06,
+      "loss": 0.6378,
+      "step": 4633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847430883256576e-06,
+      "loss": 0.5593,
+      "step": 4634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84379857726947e-06,
+      "loss": 0.6032,
+      "step": 4635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840166291895942e-06,
+      "loss": 0.6972,
+      "step": 4636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836534027615332e-06,
+      "loss": 0.7414,
+      "step": 4637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83290178490699e-06,
+      "loss": 0.6498,
+      "step": 4638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829269564250254e-06,
+      "loss": 0.6027,
+      "step": 4639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825637366124458e-06,
+      "loss": 0.6031,
+      "step": 4640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822005191008937e-06,
+      "loss": 0.6934,
+      "step": 4641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818373039383017e-06,
+      "loss": 0.6412,
+      "step": 4642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814740911726034e-06,
+      "loss": 0.6265,
+      "step": 4643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811108808517307e-06,
+      "loss": 0.6545,
+      "step": 4644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807476730236157e-06,
+      "loss": 0.6876,
+      "step": 4645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803844677361905e-06,
+      "loss": 0.6789,
+      "step": 4646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800212650373856e-06,
+      "loss": 0.6831,
+      "step": 4647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.796580649751336e-06,
+      "loss": 0.6476,
+      "step": 4648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792948675973644e-06,
+      "loss": 0.6476,
+      "step": 4649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789316729520084e-06,
+      "loss": 0.6978,
+      "step": 4650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785684810869956e-06,
+      "loss": 0.6263,
+      "step": 4651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78205292050256e-06,
+      "loss": 0.6272,
+      "step": 4652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778421058897188e-06,
+      "loss": 0.5816,
+      "step": 4653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774789226533127e-06,
+      "loss": 0.6334,
+      "step": 4654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771157423889666e-06,
+      "loss": 0.6515,
+      "step": 4655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76752565144608e-06,
+      "loss": 0.7106,
+      "step": 4656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763893909681653e-06,
+      "loss": 0.6838,
+      "step": 4657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760262199075656e-06,
+      "loss": 0.6411,
+      "step": 4658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75663052010736e-06,
+      "loss": 0.708,
+      "step": 4659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752998873256026e-06,
+      "loss": 0.61,
+      "step": 4660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74936725900092e-06,
+      "loss": 0.5829,
+      "step": 4661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.745735677821294e-06,
+      "loss": 0.6211,
+      "step": 4662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742104130196405e-06,
+      "loss": 0.6587,
+      "step": 4663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738472616605496e-06,
+      "loss": 0.5639,
+      "step": 4664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734841137527812e-06,
+      "loss": 0.6257,
+      "step": 4665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73120969344259e-06,
+      "loss": 0.6311,
+      "step": 4666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72757828482907e-06,
+      "loss": 0.6718,
+      "step": 4667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723946912166477e-06,
+      "loss": 0.6215,
+      "step": 4668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720315575934037e-06,
+      "loss": 0.6411,
+      "step": 4669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716684276610966e-06,
+      "loss": 0.7321,
+      "step": 4670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713053014676486e-06,
+      "loss": 0.732,
+      "step": 4671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.709421790609804e-06,
+      "loss": 0.6439,
+      "step": 4672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705790604890121e-06,
+      "loss": 0.7085,
+      "step": 4673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702159457996643e-06,
+      "loss": 0.6476,
+      "step": 4674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698528350408562e-06,
+      "loss": 0.7245,
+      "step": 4675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694897282605065e-06,
+      "loss": 0.6609,
+      "step": 4676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691266255065344e-06,
+      "loss": 0.6255,
+      "step": 4677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687635268268571e-06,
+      "loss": 0.6939,
+      "step": 4678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684004322693923e-06,
+      "loss": 0.6908,
+      "step": 4679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68037341882057e-06,
+      "loss": 0.6634,
+      "step": 4680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676742557127673e-06,
+      "loss": 0.7555,
+      "step": 4681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673111738094387e-06,
+      "loss": 0.6536,
+      "step": 4682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669480962199868e-06,
+      "loss": 0.6064,
+      "step": 4683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665850229923258e-06,
+      "loss": 0.7313,
+      "step": 4684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662219541743697e-06,
+      "loss": 0.6848,
+      "step": 4685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658588898140322e-06,
+      "loss": 0.7073,
+      "step": 4686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654958299592262e-06,
+      "loss": 0.6853,
+      "step": 4687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651327746578635e-06,
+      "loss": 0.6548,
+      "step": 4688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647697239578566e-06,
+      "loss": 0.7022,
+      "step": 4689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644066779071159e-06,
+      "loss": 0.7273,
+      "step": 4690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640436365535519e-06,
+      "loss": 0.6122,
+      "step": 4691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636805999450745e-06,
+      "loss": 0.6659,
+      "step": 4692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63317568129593e-06,
+      "loss": 0.6379,
+      "step": 4693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629545411550156e-06,
+      "loss": 0.6794,
+      "step": 4694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625915190692502e-06,
+      "loss": 0.7087,
+      "step": 4695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622285019202044e-06,
+      "loss": 0.6499,
+      "step": 4696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618654897557847e-06,
+      "loss": 0.6641,
+      "step": 4697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615024826238972e-06,
+      "loss": 0.7059,
+      "step": 4698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611394805724473e-06,
+      "loss": 0.6651,
+      "step": 4699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607764836493392e-06,
+      "loss": 0.679,
+      "step": 4700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604134919024771e-06,
+      "loss": 0.6525,
+      "step": 4701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600505053797644e-06,
+      "loss": 0.7059,
+      "step": 4702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596875241291033e-06,
+      "loss": 0.7776,
+      "step": 4703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593245481983956e-06,
+      "loss": 0.7421,
+      "step": 4704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58961577635543e-06,
+      "loss": 0.6615,
+      "step": 4705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585986124884457e-06,
+      "loss": 0.6301,
+      "step": 4706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582356528050034e-06,
+      "loss": 0.6851,
+      "step": 4707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578726986331154e-06,
+      "loss": 0.6851,
+      "step": 4708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575097500206799e-06,
+      "loss": 0.73,
+      "step": 4709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571468070155943e-06,
+      "loss": 0.7561,
+      "step": 4710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567838696657556e-06,
+      "loss": 0.6954,
+      "step": 4711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564209380190596e-06,
+      "loss": 0.7743,
+      "step": 4712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56058012123402e-06,
+      "loss": 0.5773,
+      "step": 4713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556950920266772e-06,
+      "loss": 0.7148,
+      "step": 4714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553321777767788e-06,
+      "loss": 0.7053,
+      "step": 4715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549692694216002e-06,
+      "loss": 0.609,
+      "step": 4716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546063670090338e-06,
+      "loss": 0.5807,
+      "step": 4717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54243470586971e-06,
+      "loss": 0.6248,
+      "step": 4718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53880580203302e-06,
+      "loss": 0.6685,
+      "step": 4719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535176959059171e-06,
+      "loss": 0.652,
+      "step": 4720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531548177427053e-06,
+      "loss": 0.714,
+      "step": 4721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527919457615546e-06,
+      "loss": 0.6044,
+      "step": 4722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524290800103531e-06,
+      "loss": 0.6604,
+      "step": 4723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520662205369868e-06,
+      "loss": 0.6055,
+      "step": 4724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517033673893415e-06,
+      "loss": 0.6622,
+      "step": 4725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51340520615303e-06,
+      "loss": 0.6879,
+      "step": 4726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509776802627548e-06,
+      "loss": 0.598,
+      "step": 4727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5061484637958e-06,
+      "loss": 0.5536,
+      "step": 4728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502520190136617e-06,
+      "loss": 0.7471,
+      "step": 4729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498891982128809e-06,
+      "loss": 0.6998,
+      "step": 4730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495263840251186e-06,
+      "loss": 0.5926,
+      "step": 4731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491635764982541e-06,
+      "loss": 0.7265,
+      "step": 4732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488007756801672e-06,
+      "loss": 0.7097,
+      "step": 4733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484379816187354e-06,
+      "loss": 0.5904,
+      "step": 4734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480751943618358e-06,
+      "loss": 0.741,
+      "step": 4735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47712413957345e-06,
+      "loss": 0.6252,
+      "step": 4736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473496404531386e-06,
+      "loss": 0.6665,
+      "step": 4737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469868738970906e-06,
+      "loss": 0.6846,
+      "step": 4738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466241143370748e-06,
+      "loss": 0.6163,
+      "step": 4739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462613618209636e-06,
+      "loss": 0.6318,
+      "step": 4740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458986163966285e-06,
+      "loss": 0.6738,
+      "step": 4741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455358781119411e-06,
+      "loss": 0.727,
+      "step": 4742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451731470147705e-06,
+      "loss": 0.6287,
+      "step": 4743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448104231529859e-06,
+      "loss": 0.6959,
+      "step": 4744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444477065744545e-06,
+      "loss": 0.6067,
+      "step": 4745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440849973270443e-06,
+      "loss": 0.6149,
+      "step": 4746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43722295458621e-06,
+      "loss": 0.6799,
+      "step": 4747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433596010170492e-06,
+      "loss": 0.623,
+      "step": 4748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429969140501934e-06,
+      "loss": 0.7736,
+      "step": 4749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42634234605916e-06,
+      "loss": 0.57,
+      "step": 4750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422715627320796e-06,
+      "loss": 0.6505,
+      "step": 4751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419088984765453e-06,
+      "loss": 0.6736,
+      "step": 4752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415462418871729e-06,
+      "loss": 0.6782,
+      "step": 4753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.411835930118215e-06,
+      "loss": 0.6565,
+      "step": 4754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408209518983487e-06,
+      "loss": 0.7356,
+      "step": 4755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404583185946124e-06,
+      "loss": 0.6541,
+      "step": 4756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40095693148468e-06,
+      "loss": 0.638,
+      "step": 4757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397330756077703e-06,
+      "loss": 0.6653,
+      "step": 4758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393704660203736e-06,
+      "loss": 0.6598,
+      "step": 4759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390078644341304e-06,
+      "loss": 0.7058,
+      "step": 4760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386452708968928e-06,
+      "loss": 0.7446,
+      "step": 4761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382826854565112e-06,
+      "loss": 0.7265,
+      "step": 4762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379201081608354e-06,
+      "loss": 0.6223,
+      "step": 4763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.375575390577135e-06,
+      "loss": 0.6025,
+      "step": 4764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371949781949937e-06,
+      "loss": 0.6951,
+      "step": 4765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368324256205224e-06,
+      "loss": 0.6252,
+      "step": 4766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364698813821444e-06,
+      "loss": 0.6382,
+      "step": 4767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361073455277043e-06,
+      "loss": 0.6311,
+      "step": 4768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357448181050447e-06,
+      "loss": 0.6362,
+      "step": 4769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353822991620083e-06,
+      "loss": 0.5913,
+      "step": 4770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350197887464357e-06,
+      "loss": 0.5922,
+      "step": 4771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346572869061663e-06,
+      "loss": 0.6409,
+      "step": 4772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34294793689039e-06,
+      "loss": 0.6257,
+      "step": 4773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339323091428911e-06,
+      "loss": 0.6709,
+      "step": 4774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335698333155592e-06,
+      "loss": 0.6533,
+      "step": 4775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332073662548785e-06,
+      "loss": 0.7641,
+      "step": 4776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32844908008683e-06,
+      "loss": 0.751,
+      "step": 4777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32482458624805e-06,
+      "loss": 0.7516,
+      "step": 4778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32120018151077e-06,
+      "loss": 0.6506,
+      "step": 4779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317575866353293e-06,
+      "loss": 0.6562,
+      "step": 4780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31395164125391e-06,
+      "loss": 0.6508,
+      "step": 4781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310327506690901e-06,
+      "loss": 0.613,
+      "step": 4782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30670346314254e-06,
+      "loss": 0.6871,
+      "step": 4783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303079511087078e-06,
+      "loss": 0.6942,
+      "step": 4784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299455651002769e-06,
+      "loss": 0.6385,
+      "step": 4785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29583188336784e-06,
+      "loss": 0.6432,
+      "step": 4786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292208208660516e-06,
+      "loss": 0.6015,
+      "step": 4787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288584627359004e-06,
+      "loss": 0.635,
+      "step": 4788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284961139941501e-06,
+      "loss": 0.6936,
+      "step": 4789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28133774688619e-06,
+      "loss": 0.6591,
+      "step": 4790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277714448671244e-06,
+      "loss": 0.6342,
+      "step": 4791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27409124577482e-06,
+      "loss": 0.697,
+      "step": 4792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270468138675065e-06,
+      "loss": 0.6518,
+      "step": 4793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266845127850113e-06,
+      "loss": 0.6442,
+      "step": 4794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263222213778088e-06,
+      "loss": 0.677,
+      "step": 4795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259599396937097e-06,
+      "loss": 0.6547,
+      "step": 4796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255976677805234e-06,
+      "loss": 0.6391,
+      "step": 4797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252354056860587e-06,
+      "loss": 0.6369,
+      "step": 4798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248731534581217e-06,
+      "loss": 0.6707,
+      "step": 4799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.245109111445189e-06,
+      "loss": 0.6993,
+      "step": 4800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241486787930542e-06,
+      "loss": 0.6899,
+      "step": 4801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.237864564515306e-06,
+      "loss": 0.5809,
+      "step": 4802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234242441677499e-06,
+      "loss": 0.6351,
+      "step": 4803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230620419895124e-06,
+      "loss": 0.6539,
+      "step": 4804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226998499646177e-06,
+      "loss": 0.6607,
+      "step": 4805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22337668140863e-06,
+      "loss": 0.6714,
+      "step": 4806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21975496566045e-06,
+      "loss": 0.6554,
+      "step": 4807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216133352879587e-06,
+      "loss": 0.6504,
+      "step": 4808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212511843543978e-06,
+      "loss": 0.6493,
+      "step": 4809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208890438131543e-06,
+      "loss": 0.6691,
+      "step": 4810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205269137120192e-06,
+      "loss": 0.5914,
+      "step": 4811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20164794098782e-06,
+      "loss": 0.5946,
+      "step": 4812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198026850212314e-06,
+      "loss": 0.6839,
+      "step": 4813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194405865271535e-06,
+      "loss": 0.5729,
+      "step": 4814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19078498664334e-06,
+      "loss": 0.595,
+      "step": 4815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18716421480557e-06,
+      "loss": 0.5851,
+      "step": 4816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183543550236052e-06,
+      "loss": 0.7333,
+      "step": 4817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179922993412594e-06,
+      "loss": 0.6779,
+      "step": 4818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176302544812994e-06,
+      "loss": 0.6525,
+      "step": 4819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172682204915036e-06,
+      "loss": 0.7253,
+      "step": 4820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169061974196487e-06,
+      "loss": 0.6708,
+      "step": 4821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165441853135104e-06,
+      "loss": 0.6706,
+      "step": 4822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161821842208624e-06,
+      "loss": 0.7231,
+      "step": 4823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158201941894773e-06,
+      "loss": 0.69,
+      "step": 4824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154582152671266e-06,
+      "loss": 0.6255,
+      "step": 4825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150962475015794e-06,
+      "loss": 0.7227,
+      "step": 4826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14734290940604e-06,
+      "loss": 0.5882,
+      "step": 4827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143723456319673e-06,
+      "loss": 0.7112,
+      "step": 4828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14010411623434e-06,
+      "loss": 0.6249,
+      "step": 4829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136484889627679e-06,
+      "loss": 0.6742,
+      "step": 4830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132865776977315e-06,
+      "loss": 0.6933,
+      "step": 4831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129246778760851e-06,
+      "loss": 0.5994,
+      "step": 4832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125627895455878e-06,
+      "loss": 0.7028,
+      "step": 4833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122009127539978e-06,
+      "loss": 0.6042,
+      "step": 4834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11839047549071e-06,
+      "loss": 0.7283,
+      "step": 4835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114771939785618e-06,
+      "loss": 0.6383,
+      "step": 4836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111153520902235e-06,
+      "loss": 0.6647,
+      "step": 4837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107535219318073e-06,
+      "loss": 0.7338,
+      "step": 4838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103917035510636e-06,
+      "loss": 0.6751,
+      "step": 4839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100298969957402e-06,
+      "loss": 0.6276,
+      "step": 4840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096681023135848e-06,
+      "loss": 0.6028,
+      "step": 4841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093063195523419e-06,
+      "loss": 0.5646,
+      "step": 4842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089445487597553e-06,
+      "loss": 0.6894,
+      "step": 4843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08582789983568e-06,
+      "loss": 0.6362,
+      "step": 4844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082210432715197e-06,
+      "loss": 0.6994,
+      "step": 4845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078593086713496e-06,
+      "loss": 0.5489,
+      "step": 4846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074975862307953e-06,
+      "loss": 0.7215,
+      "step": 4847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071358759975922e-06,
+      "loss": 0.6964,
+      "step": 4848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067741780194745e-06,
+      "loss": 0.6831,
+      "step": 4849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064124923441749e-06,
+      "loss": 0.6635,
+      "step": 4850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060508190194241e-06,
+      "loss": 0.6848,
+      "step": 4851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056891580929517e-06,
+      "loss": 0.6968,
+      "step": 4852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053275096124846e-06,
+      "loss": 0.6269,
+      "step": 4853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0496587362575e-06,
+      "loss": 0.6573,
+      "step": 4854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046042501804715e-06,
+      "loss": 0.7132,
+      "step": 4855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042426393243715e-06,
+      "loss": 0.6835,
+      "step": 4856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03881041105172e-06,
+      "loss": 0.6152,
+      "step": 4857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035194555705912e-06,
+      "loss": 0.699,
+      "step": 4858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031578827683477e-06,
+      "loss": 0.5574,
+      "step": 4859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027963227461571e-06,
+      "loss": 0.6499,
+      "step": 4860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02434775551734e-06,
+      "loss": 0.6302,
+      "step": 4861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020732412327906e-06,
+      "loss": 0.6548,
+      "step": 4862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017117198370378e-06,
+      "loss": 0.6548,
+      "step": 4863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013502114121857e-06,
+      "loss": 0.6799,
+      "step": 4864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009887160059409e-06,
+      "loss": 0.6334,
+      "step": 4865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006272336660096e-06,
+      "loss": 0.6064,
+      "step": 4866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002657644400957e-06,
+      "loss": 0.6759,
+      "step": 4867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.999043083759016e-06,
+      "loss": 0.6201,
+      "step": 4868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995428655211283e-06,
+      "loss": 0.6122,
+      "step": 4869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991814359234741e-06,
+      "loss": 0.64,
+      "step": 4870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988200196306363e-06,
+      "loss": 0.7187,
+      "step": 4871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984586166903103e-06,
+      "loss": 0.6939,
+      "step": 4872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980972271501892e-06,
+      "loss": 0.7041,
+      "step": 4873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977358510579658e-06,
+      "loss": 0.728,
+      "step": 4874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973744884613298e-06,
+      "loss": 0.7064,
+      "step": 4875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970131394079689e-06,
+      "loss": 0.7303,
+      "step": 4876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9665180394557e-06,
+      "loss": 0.6748,
+      "step": 4877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962904821218181e-06,
+      "loss": 0.6553,
+      "step": 4878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959291739843958e-06,
+      "loss": 0.6275,
+      "step": 4879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95567879580984e-06,
+      "loss": 0.7394,
+      "step": 4880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952065989592622e-06,
+      "loss": 0.7392,
+      "step": 4881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948453321669073e-06,
+      "loss": 0.628,
+      "step": 4882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94484079251596e-06,
+      "loss": 0.6425,
+      "step": 4883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941228402610013e-06,
+      "loss": 0.6099,
+      "step": 4884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937616152427956e-06,
+      "loss": 0.679,
+      "step": 4885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934004042446485e-06,
+      "loss": 0.6078,
+      "step": 4886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930392073142286e-06,
+      "loss": 0.5336,
+      "step": 4887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926780244992024e-06,
+      "loss": 0.6234,
+      "step": 4888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923168558472345e-06,
+      "loss": 0.6079,
+      "step": 4889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91955701405987e-06,
+      "loss": 0.6839,
+      "step": 4890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915945612231213e-06,
+      "loss": 0.7031,
+      "step": 4891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912334353462957e-06,
+      "loss": 0.6634,
+      "step": 4892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908723238231682e-06,
+      "loss": 0.5891,
+      "step": 4893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905112267013931e-06,
+      "loss": 0.5949,
+      "step": 4894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901501440286237e-06,
+      "loss": 0.6525,
+      "step": 4895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89789075852512e-06,
+      "loss": 0.6404,
+      "step": 4896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894280222207067e-06,
+      "loss": 0.5704,
+      "step": 4897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.890669831808558e-06,
+      "loss": 0.6398,
+      "step": 4898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887059587806043e-06,
+      "loss": 0.5768,
+      "step": 4899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883449490675964e-06,
+      "loss": 0.6065,
+      "step": 4900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879839540894736e-06,
+      "loss": 0.6707,
+      "step": 4901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87622973893875e-06,
+      "loss": 0.6132,
+      "step": 4902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872620085284397e-06,
+      "loss": 0.6642,
+      "step": 4903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869010580408026e-06,
+      "loss": 0.6715,
+      "step": 4904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865401224785979e-06,
+      "loss": 0.6641,
+      "step": 4905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861792018894576e-06,
+      "loss": 0.6057,
+      "step": 4906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858182963210116e-06,
+      "loss": 0.7148,
+      "step": 4907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854574058208877e-06,
+      "loss": 0.6068,
+      "step": 4908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85096530436712e-06,
+      "loss": 0.6071,
+      "step": 4909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847356702161085e-06,
+      "loss": 0.6074,
+      "step": 4910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843748252066986e-06,
+      "loss": 0.71,
+      "step": 4911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84013995456103e-06,
+      "loss": 0.6731,
+      "step": 4912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836531810119396e-06,
+      "loss": 0.6485,
+      "step": 4913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832923819218238e-06,
+      "loss": 0.652,
+      "step": 4914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829315982333702e-06,
+      "loss": 0.6868,
+      "step": 4915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825708299941905e-06,
+      "loss": 0.7017,
+      "step": 4916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82210077251894e-06,
+      "loss": 0.6325,
+      "step": 4917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81849340054089e-06,
+      "loss": 0.5962,
+      "step": 4918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814886184483812e-06,
+      "loss": 0.6476,
+      "step": 4919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811279124823738e-06,
+      "loss": 0.7218,
+      "step": 4920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807672222036692e-06,
+      "loss": 0.6787,
+      "step": 4921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804065476598662e-06,
+      "loss": 0.7146,
+      "step": 4922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800458888985625e-06,
+      "loss": 0.6596,
+      "step": 4923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796852459673541e-06,
+      "loss": 0.6531,
+      "step": 4924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793246189138335e-06,
+      "loss": 0.6203,
+      "step": 4925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789640077855924e-06,
+      "loss": 0.6825,
+      "step": 4926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786034126302196e-06,
+      "loss": 0.7076,
+      "step": 4927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782428334953022e-06,
+      "loss": 0.6346,
+      "step": 4928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778822704284248e-06,
+      "loss": 0.6641,
+      "step": 4929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775217234771709e-06,
+      "loss": 0.6583,
+      "step": 4930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771611926891204e-06,
+      "loss": 0.647,
+      "step": 4931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768006781118517e-06,
+      "loss": 0.6236,
+      "step": 4932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764401797929419e-06,
+      "loss": 0.6583,
+      "step": 4933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760796977799649e-06,
+      "loss": 0.6113,
+      "step": 4934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757192321204927e-06,
+      "loss": 0.6595,
+      "step": 4935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753587828620951e-06,
+      "loss": 0.7184,
+      "step": 4936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7499835005234e-06,
+      "loss": 0.738,
+      "step": 4937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74637933738793e-06,
+      "loss": 0.6475,
+      "step": 4938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74277533969017e-06,
+      "loss": 0.7459,
+      "step": 4939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739171507905741e-06,
+      "loss": 0.56,
+      "step": 4940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735567842510221e-06,
+      "loss": 0.6743,
+      "step": 4941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73196434397919e-06,
+      "loss": 0.7223,
+      "step": 4942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728361012788193e-06,
+      "loss": 0.6383,
+      "step": 4943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724757849412748e-06,
+      "loss": 0.6032,
+      "step": 4944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72115485432836e-06,
+      "loss": 0.65,
+      "step": 4945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71755202801051e-06,
+      "loss": 0.6045,
+      "step": 4946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713949370934654e-06,
+      "loss": 0.7969,
+      "step": 4947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710346883576223e-06,
+      "loss": 0.6373,
+      "step": 4948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706744566410638e-06,
+      "loss": 0.6763,
+      "step": 4949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703142419913287e-06,
+      "loss": 0.6908,
+      "step": 4950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699540444559528e-06,
+      "loss": 0.6547,
+      "step": 4951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695938640824721e-06,
+      "loss": 0.6226,
+      "step": 4952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692337009184181e-06,
+      "loss": 0.6582,
+      "step": 4953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68873555011321e-06,
+      "loss": 0.6619,
+      "step": 4954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685134264087082e-06,
+      "loss": 0.6766,
+      "step": 4955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681533151581056e-06,
+      "loss": 0.6421,
+      "step": 4956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.677932213070356e-06,
+      "loss": 0.6581,
+      "step": 4957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674331449030198e-06,
+      "loss": 0.6505,
+      "step": 4958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670730859935763e-06,
+      "loss": 0.7449,
+      "step": 4959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667130446262214e-06,
+      "loss": 0.6606,
+      "step": 4960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663530208484688e-06,
+      "loss": 0.7601,
+      "step": 4961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659930147078307e-06,
+      "loss": 0.6163,
+      "step": 4962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65633026251816e-06,
+      "loss": 0.5307,
+      "step": 4963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652730555279314e-06,
+      "loss": 0.6865,
+      "step": 4964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64913102583682e-06,
+      "loss": 0.5638,
+      "step": 4965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645531674665695e-06,
+      "loss": 0.5784,
+      "step": 4966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641932502240943e-06,
+      "loss": 0.5949,
+      "step": 4967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638333509037537e-06,
+      "loss": 0.5407,
+      "step": 4968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634734695530427e-06,
+      "loss": 0.6577,
+      "step": 4969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631136062194542e-06,
+      "loss": 0.6547,
+      "step": 4970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627537609504783e-06,
+      "loss": 0.7307,
+      "step": 4971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62393933793604e-06,
+      "loss": 0.5919,
+      "step": 4972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620341247963163e-06,
+      "loss": 0.6193,
+      "step": 4973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616743340060983e-06,
+      "loss": 0.7206,
+      "step": 4974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613145614704307e-06,
+      "loss": 0.6605,
+      "step": 4975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.609548072367928e-06,
+      "loss": 0.6291,
+      "step": 4976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6059507135266e-06,
+      "loss": 0.6223,
+      "step": 4977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602353538655057e-06,
+      "loss": 0.5824,
+      "step": 4978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598756548228015e-06,
+      "loss": 0.6489,
+      "step": 4979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59515974272016e-06,
+      "loss": 0.6925,
+      "step": 4980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591563122606152e-06,
+      "loss": 0.6598,
+      "step": 4981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587966688360634e-06,
+      "loss": 0.6085,
+      "step": 4982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58437044045822e-06,
+      "loss": 0.625,
+      "step": 4983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580774379373495e-06,
+      "loss": 0.5417,
+      "step": 4984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577178505581027e-06,
+      "loss": 0.6098,
+      "step": 4985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573582819555355e-06,
+      "loss": 0.6223,
+      "step": 4986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569987321770997e-06,
+      "loss": 0.528,
+      "step": 4987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566392012702438e-06,
+      "loss": 0.6398,
+      "step": 4988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562796892824148e-06,
+      "loss": 0.6402,
+      "step": 4989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559201962610564e-06,
+      "loss": 0.7388,
+      "step": 4990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555607222536102e-06,
+      "loss": 0.7169,
+      "step": 4991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552012673075155e-06,
+      "loss": 0.642,
+      "step": 4992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548418314702086e-06,
+      "loss": 0.656,
+      "step": 4993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544824147891238e-06,
+      "loss": 0.667,
+      "step": 4994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541230173116922e-06,
+      "loss": 0.6386,
+      "step": 4995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53763639085343e-06,
+      "loss": 0.708,
+      "step": 4996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534042801575025e-06,
+      "loss": 0.6354,
+      "step": 4997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530449405755946e-06,
+      "loss": 0.6459,
+      "step": 4998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526856203870404e-06,
+      "loss": 0.5911,
+      "step": 4999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523263196392585e-06,
+      "loss": 0.7405,
+      "step": 5000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519670383796658e-06,
+      "loss": 0.6283,
+      "step": 5001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516077766556754e-06,
+      "loss": 0.5294,
+      "step": 5002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512485345146982e-06,
+      "loss": 0.6445,
+      "step": 5003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508893120041428e-06,
+      "loss": 0.6501,
+      "step": 5004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505301091714153e-06,
+      "loss": 0.6376,
+      "step": 5005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501709260639187e-06,
+      "loss": 0.6835,
+      "step": 5006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498117627290536e-06,
+      "loss": 0.6252,
+      "step": 5007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494526192142181e-06,
+      "loss": 0.645,
+      "step": 5008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490934955668077e-06,
+      "loss": 0.6546,
+      "step": 5009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487343918342146e-06,
+      "loss": 0.5983,
+      "step": 5010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483753080638299e-06,
+      "loss": 0.703,
+      "step": 5011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480162443030407e-06,
+      "loss": 0.6329,
+      "step": 5012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47657200599232e-06,
+      "loss": 0.6469,
+      "step": 5013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472981769997862e-06,
+      "loss": 0.6184,
+      "step": 5014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469391735520824e-06,
+      "loss": 0.6935,
+      "step": 5015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46580190303498e-06,
+      "loss": 0.5985,
+      "step": 5016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462212273014072e-06,
+      "loss": 0.5996,
+      "step": 5017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458622845931812e-06,
+      "loss": 0.5885,
+      "step": 5018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455033622261892e-06,
+      "loss": 0.6023,
+      "step": 5019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451444602477975e-06,
+      "loss": 0.6755,
+      "step": 5020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447855787053698e-06,
+      "loss": 0.6371,
+      "step": 5021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444267176462665e-06,
+      "loss": 0.595,
+      "step": 5022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440678771178462e-06,
+      "loss": 0.6382,
+      "step": 5023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437090571674642e-06,
+      "loss": 0.5841,
+      "step": 5024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433502578424732e-06,
+      "loss": 0.6605,
+      "step": 5025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429914791902231e-06,
+      "loss": 0.5979,
+      "step": 5026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426327212580614e-06,
+      "loss": 0.7118,
+      "step": 5027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422739840933322e-06,
+      "loss": 0.5705,
+      "step": 5028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419152677433779e-06,
+      "loss": 0.627,
+      "step": 5029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415565722555368e-06,
+      "loss": 0.7779,
+      "step": 5030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411978976771457e-06,
+      "loss": 0.7897,
+      "step": 5031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408392440555384e-06,
+      "loss": 0.6629,
+      "step": 5032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404806114380452e-06,
+      "loss": 0.6105,
+      "step": 5033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401219998719945e-06,
+      "loss": 0.6251,
+      "step": 5034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397634094047112e-06,
+      "loss": 0.6714,
+      "step": 5035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394048400835177e-06,
+      "loss": 0.7263,
+      "step": 5036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390462919557334e-06,
+      "loss": 0.6523,
+      "step": 5037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38687765068676e-06,
+      "loss": 0.723,
+      "step": 5038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383292594696589e-06,
+      "loss": 0.6548,
+      "step": 5039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379707752059932e-06,
+      "loss": 0.7518,
+      "step": 5040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37612312324988e-06,
+      "loss": 0.6033,
+      "step": 5041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372538708739484e-06,
+      "loss": 0.6053,
+      "step": 5042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368954509001774e-06,
+      "loss": 0.6766,
+      "step": 5043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36537052450975e-06,
+      "loss": 0.6989,
+      "step": 5044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361786755736379e-06,
+      "loss": 0.6522,
+      "step": 5045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358203203154607e-06,
+      "loss": 0.6934,
+      "step": 5046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.354619867237344e-06,
+      "loss": 0.7674,
+      "step": 5047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351036748457481e-06,
+      "loss": 0.575,
+      "step": 5048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347453847287868e-06,
+      "loss": 0.647,
+      "step": 5049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34387116420134e-06,
+      "loss": 0.645,
+      "step": 5050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340288699670694e-06,
+      "loss": 0.6915,
+      "step": 5051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.336706454168701e-06,
+      "loss": 0.612,
+      "step": 5052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3331244281681e-06,
+      "loss": 0.6712,
+      "step": 5053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329542622141604e-06,
+      "loss": 0.582,
+      "step": 5054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325961036561899e-06,
+      "loss": 0.6068,
+      "step": 5055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322379671901632e-06,
+      "loss": 0.7208,
+      "step": 5056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318798528633439e-06,
+      "loss": 0.7066,
+      "step": 5057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315217607229911e-06,
+      "loss": 0.685,
+      "step": 5058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31163690816361e-06,
+      "loss": 0.7087,
+      "step": 5059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308056431907081e-06,
+      "loss": 0.6401,
+      "step": 5060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30447617893283e-06,
+      "loss": 0.6381,
+      "step": 5061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300896149713334e-06,
+      "loss": 0.5567,
+      "step": 5062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297316344721044e-06,
+      "loss": 0.6394,
+      "step": 5063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293736764428377e-06,
+      "loss": 0.7182,
+      "step": 5064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290157409307723e-06,
+      "loss": 0.5746,
+      "step": 5065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286578279831443e-06,
+      "loss": 0.6745,
+      "step": 5066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282999376471868e-06,
+      "loss": 0.586,
+      "step": 5067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279420699701297e-06,
+      "loss": 0.6492,
+      "step": 5068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275842249991999e-06,
+      "loss": 0.652,
+      "step": 5069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272264027816218e-06,
+      "loss": 0.7466,
+      "step": 5070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268686033646166e-06,
+      "loss": 0.6493,
+      "step": 5071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26510826795402e-06,
+      "loss": 0.7124,
+      "step": 5072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26153073121193e-06,
+      "loss": 0.614,
+      "step": 5073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257953423892015e-06,
+      "loss": 0.6631,
+      "step": 5074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25437634646637e-06,
+      "loss": 0.7148,
+      "step": 5075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250799499407053e-06,
+      "loss": 0.6633,
+      "step": 5076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247222883186089e-06,
+      "loss": 0.6396,
+      "step": 5077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243646498275479e-06,
+      "loss": 0.6316,
+      "step": 5078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240070345147187e-06,
+      "loss": 0.6867,
+      "step": 5079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236494424273159e-06,
+      "loss": 0.5465,
+      "step": 5080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232918736125299e-06,
+      "loss": 0.5752,
+      "step": 5081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22934328117548e-06,
+      "loss": 0.6448,
+      "step": 5082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22576805989555e-06,
+      "loss": 0.6882,
+      "step": 5083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222193072757321e-06,
+      "loss": 0.6806,
+      "step": 5084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218618320232578e-06,
+      "loss": 0.5579,
+      "step": 5085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215043802793077e-06,
+      "loss": 0.6271,
+      "step": 5086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211469520910534e-06,
+      "loss": 0.6365,
+      "step": 5087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207895475056642e-06,
+      "loss": 0.6258,
+      "step": 5088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204321665703057e-06,
+      "loss": 0.7318,
+      "step": 5089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200748093321415e-06,
+      "loss": 0.6783,
+      "step": 5090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197174758383308e-06,
+      "loss": 0.669,
+      "step": 5091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193601661360301e-06,
+      "loss": 0.7116,
+      "step": 5092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19002880272393e-06,
+      "loss": 0.6101,
+      "step": 5093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186456182945698e-06,
+      "loss": 0.6374,
+      "step": 5094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182883802497074e-06,
+      "loss": 0.6946,
+      "step": 5095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1793116618495e-06,
+      "loss": 0.654,
+      "step": 5096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175739761474383e-06,
+      "loss": 0.6666,
+      "step": 5097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1721681018431e-06,
+      "loss": 0.6029,
+      "step": 5098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16859668342699e-06,
+      "loss": 0.6136,
+      "step": 5099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165025506697373e-06,
+      "loss": 0.6731,
+      "step": 5100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161454572125528e-06,
+      "loss": 0.7315,
+      "step": 5101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1578838801827e-06,
+      "loss": 0.6856,
+      "step": 5102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154313431340112e-06,
+      "loss": 0.586,
+      "step": 5103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150743226068944e-06,
+      "loss": 0.6806,
+      "step": 5104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147173264840352e-06,
+      "loss": 0.6499,
+      "step": 5105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143603548125453e-06,
+      "loss": 0.6644,
+      "step": 5106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140034076395336e-06,
+      "loss": 0.5958,
+      "step": 5107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136464850121054e-06,
+      "loss": 0.6277,
+      "step": 5108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.132895869773638e-06,
+      "loss": 0.6791,
+      "step": 5109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129327135824075e-06,
+      "loss": 0.662,
+      "step": 5110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125758648743318e-06,
+      "loss": 0.6862,
+      "step": 5111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122190409002302e-06,
+      "loss": 0.6326,
+      "step": 5112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118622417071916e-06,
+      "loss": 0.6085,
+      "step": 5113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11505467342302e-06,
+      "loss": 0.5638,
+      "step": 5114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111487178526442e-06,
+      "loss": 0.6979,
+      "step": 5115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107919932852976e-06,
+      "loss": 0.6015,
+      "step": 5116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104352936873387e-06,
+      "loss": 0.6413,
+      "step": 5117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100786191058397e-06,
+      "loss": 0.6594,
+      "step": 5118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09721969587871e-06,
+      "loss": 0.6561,
+      "step": 5119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.093653451804987e-06,
+      "loss": 0.6057,
+      "step": 5120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090087459307854e-06,
+      "loss": 0.6001,
+      "step": 5121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086521718857915e-06,
+      "loss": 0.675,
+      "step": 5122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082956230925727e-06,
+      "loss": 0.6536,
+      "step": 5123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079390995981821e-06,
+      "loss": 0.6303,
+      "step": 5124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075826014496697e-06,
+      "loss": 0.6976,
+      "step": 5125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072261286940814e-06,
+      "loss": 0.8261,
+      "step": 5126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068696813784602e-06,
+      "loss": 0.6904,
+      "step": 5127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065132595498458e-06,
+      "loss": 0.713,
+      "step": 5128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.061568632552748e-06,
+      "loss": 0.5572,
+      "step": 5129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058004925417794e-06,
+      "loss": 0.692,
+      "step": 5130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054441474563898e-06,
+      "loss": 0.7395,
+      "step": 5131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050878280461319e-06,
+      "loss": 0.5822,
+      "step": 5132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047315343580281e-06,
+      "loss": 0.5668,
+      "step": 5133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043752664390981e-06,
+      "loss": 0.6647,
+      "step": 5134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040190243363576e-06,
+      "loss": 0.6729,
+      "step": 5135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036628080968191e-06,
+      "loss": 0.6996,
+      "step": 5136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033066177674922e-06,
+      "loss": 0.7044,
+      "step": 5137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02950453395382e-06,
+      "loss": 0.7252,
+      "step": 5138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02594315027491e-06,
+      "loss": 0.7066,
+      "step": 5139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022382027108182e-06,
+      "loss": 0.5779,
+      "step": 5140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018821164923589e-06,
+      "loss": 0.6546,
+      "step": 5141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01526056419105e-06,
+      "loss": 0.6497,
+      "step": 5142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.011700225380451e-06,
+      "loss": 0.585,
+      "step": 5143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008140148961642e-06,
+      "loss": 0.7465,
+      "step": 5144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004580335404435e-06,
+      "loss": 0.7112,
+      "step": 5145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001020785178619e-06,
+      "loss": 0.671,
+      "step": 5146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997461498753936e-06,
+      "loss": 0.7051,
+      "step": 5147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993902476600093e-06,
+      "loss": 0.6455,
+      "step": 5148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990343719186778e-06,
+      "loss": 0.6873,
+      "step": 5149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986785226983624e-06,
+      "loss": 0.703,
+      "step": 5150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983227000460241e-06,
+      "loss": 0.599,
+      "step": 5151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979669040086201e-06,
+      "loss": 0.6567,
+      "step": 5152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976111346331038e-06,
+      "loss": 0.6169,
+      "step": 5153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97255391966426e-06,
+      "loss": 0.6641,
+      "step": 5154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96899676055532e-06,
+      "loss": 0.6566,
+      "step": 5155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965439869473664e-06,
+      "loss": 0.6426,
+      "step": 5156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96188324688868e-06,
+      "loss": 0.5874,
+      "step": 5157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958326893269724e-06,
+      "loss": 0.6578,
+      "step": 5158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95477080908613e-06,
+      "loss": 0.6198,
+      "step": 5159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951214994807184e-06,
+      "loss": 0.6312,
+      "step": 5160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947659450902136e-06,
+      "loss": 0.5328,
+      "step": 5161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944104177840207e-06,
+      "loss": 0.6733,
+      "step": 5162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940549176090578e-06,
+      "loss": 0.6602,
+      "step": 5163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936994446122391e-06,
+      "loss": 0.6136,
+      "step": 5164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933439988404765e-06,
+      "loss": 0.5407,
+      "step": 5165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929885803406768e-06,
+      "loss": 0.6319,
+      "step": 5166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926331891597436e-06,
+      "loss": 0.6975,
+      "step": 5167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92277825344578e-06,
+      "loss": 0.5418,
+      "step": 5168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919224889420762e-06,
+      "loss": 0.463,
+      "step": 5169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91567179999131e-06,
+      "loss": 0.6473,
+      "step": 5170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912118985626324e-06,
+      "loss": 0.5883,
+      "step": 5171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908566446794655e-06,
+      "loss": 0.6715,
+      "step": 5172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905014183965122e-06,
+      "loss": 0.6361,
+      "step": 5173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901462197606517e-06,
+      "loss": 0.6712,
+      "step": 5174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897910488187588e-06,
+      "loss": 0.6638,
+      "step": 5175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89435905617704e-06,
+      "loss": 0.5583,
+      "step": 5176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890807902043551e-06,
+      "loss": 0.6132,
+      "step": 5177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887257026255765e-06,
+      "loss": 0.7307,
+      "step": 5178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883706429282278e-06,
+      "loss": 0.6819,
+      "step": 5179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880156111591655e-06,
+      "loss": 0.775,
+      "step": 5180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876606073652427e-06,
+      "loss": 0.6904,
+      "step": 5181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873056315933081e-06,
+      "loss": 0.6758,
+      "step": 5182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869506838902076e-06,
+      "loss": 0.6387,
+      "step": 5183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865957643027826e-06,
+      "loss": 0.6453,
+      "step": 5184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862408728778713e-06,
+      "loss": 0.6726,
+      "step": 5185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858860096623075e-06,
+      "loss": 0.6078,
+      "step": 5186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85531174702922e-06,
+      "loss": 0.7388,
+      "step": 5187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.851763680465421e-06,
+      "loss": 0.5861,
+      "step": 5188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848215897399905e-06,
+      "loss": 0.5955,
+      "step": 5189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844668398300866e-06,
+      "loss": 0.6844,
+      "step": 5190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84112118363646e-06,
+      "loss": 0.671,
+      "step": 5191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837574253874802e-06,
+      "loss": 0.6199,
+      "step": 5192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83402760948398e-06,
+      "loss": 0.6065,
+      "step": 5193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830481250932035e-06,
+      "loss": 0.7323,
+      "step": 5194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826935178686969e-06,
+      "loss": 0.5682,
+      "step": 5195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823389393216753e-06,
+      "loss": 0.665,
+      "step": 5196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819843894989314e-06,
+      "loss": 0.6485,
+      "step": 5197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816298684472547e-06,
+      "loss": 0.6658,
+      "step": 5198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812753762134307e-06,
+      "loss": 0.6274,
+      "step": 5199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809209128442408e-06,
+      "loss": 0.6391,
+      "step": 5200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805664783864627e-06,
+      "loss": 0.7223,
+      "step": 5201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802120728868706e-06,
+      "loss": 0.6342,
+      "step": 5202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798576963922347e-06,
+      "loss": 0.6861,
+      "step": 5203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79503348949321e-06,
+      "loss": 0.5954,
+      "step": 5204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79149030604892e-06,
+      "loss": 0.5681,
+      "step": 5205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78794741405707e-06,
+      "loss": 0.6712,
+      "step": 5206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784404813985195e-06,
+      "loss": 0.7303,
+      "step": 5207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78086250630082e-06,
+      "loss": 0.6518,
+      "step": 5208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777320491471409e-06,
+      "loss": 0.6697,
+      "step": 5209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77377876996439e-06,
+      "loss": 0.6838,
+      "step": 5210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770237342247164e-06,
+      "loss": 0.5792,
+      "step": 5211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766696208787085e-06,
+      "loss": 0.7061,
+      "step": 5212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763155370051465e-06,
+      "loss": 0.6804,
+      "step": 5213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759614826507582e-06,
+      "loss": 0.6528,
+      "step": 5214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756074578622676e-06,
+      "loss": 0.637,
+      "step": 5215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.752534626863946e-06,
+      "loss": 0.5857,
+      "step": 5216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748994971698547e-06,
+      "loss": 0.6435,
+      "step": 5217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745455613593609e-06,
+      "loss": 0.579,
+      "step": 5218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741916553016205e-06,
+      "loss": 0.6384,
+      "step": 5219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738377790433385e-06,
+      "loss": 0.6239,
+      "step": 5220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73483932631215e-06,
+      "loss": 0.6273,
+      "step": 5221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73130116111946e-06,
+      "loss": 0.6874,
+      "step": 5222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727763295322243e-06,
+      "loss": 0.5841,
+      "step": 5223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724225729387382e-06,
+      "loss": 0.6186,
+      "step": 5224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720688463781722e-06,
+      "loss": 0.751,
+      "step": 5225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717151498972066e-06,
+      "loss": 0.6241,
+      "step": 5226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713614835425188e-06,
+      "loss": 0.7171,
+      "step": 5227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710078473607809e-06,
+      "loss": 0.713,
+      "step": 5228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706542413986614e-06,
+      "loss": 0.6089,
+      "step": 5229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703006657028253e-06,
+      "loss": 0.6818,
+      "step": 5230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69947120319933e-06,
+      "loss": 0.677,
+      "step": 5231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695936052966412e-06,
+      "loss": 0.6327,
+      "step": 5232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692401206796027e-06,
+      "loss": 0.7011,
+      "step": 5233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688866665154659e-06,
+      "loss": 0.6112,
+      "step": 5234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685332428508754e-06,
+      "loss": 0.6727,
+      "step": 5235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681798497324717e-06,
+      "loss": 0.6001,
+      "step": 5236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678264872068919e-06,
+      "loss": 0.6886,
+      "step": 5237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67473155320768e-06,
+      "loss": 0.6579,
+      "step": 5238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671198541207287e-06,
+      "loss": 0.6338,
+      "step": 5239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667665836533984e-06,
+      "loss": 0.597,
+      "step": 5240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664133439653973e-06,
+      "loss": 0.6689,
+      "step": 5241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660601351033422e-06,
+      "loss": 0.6983,
+      "step": 5242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657069571138446e-06,
+      "loss": 0.6606,
+      "step": 5243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653538100435129e-06,
+      "loss": 0.6196,
+      "step": 5244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650006939389518e-06,
+      "loss": 0.7078,
+      "step": 5245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646476088467604e-06,
+      "loss": 0.5787,
+      "step": 5246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64294554813535e-06,
+      "loss": 0.6548,
+      "step": 5247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639415318858678e-06,
+      "loss": 0.5722,
+      "step": 5248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63588540110346e-06,
+      "loss": 0.6297,
+      "step": 5249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632355795335533e-06,
+      "loss": 0.5548,
+      "step": 5250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628826502020694e-06,
+      "loss": 0.5295,
+      "step": 5251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625297521624694e-06,
+      "loss": 0.6126,
+      "step": 5252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621768854613245e-06,
+      "loss": 0.7302,
+      "step": 5253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618240501452017e-06,
+      "loss": 0.6049,
+      "step": 5254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614712462606642e-06,
+      "loss": 0.6248,
+      "step": 5255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6111847385427054e-06,
+      "loss": 0.6206,
+      "step": 5256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607657329725757e-06,
+      "loss": 0.5872,
+      "step": 5257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604130236621301e-06,
+      "loss": 0.6385,
+      "step": 5258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6006034596948e-06,
+      "loss": 0.6027,
+      "step": 5259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597076999411673e-06,
+      "loss": 0.6878,
+      "step": 5260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593550856237303e-06,
+      "loss": 0.7598,
+      "step": 5261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590025030637026e-06,
+      "loss": 0.6408,
+      "step": 5262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586499523076135e-06,
+      "loss": 0.6927,
+      "step": 5263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58297433401989e-06,
+      "loss": 0.6057,
+      "step": 5264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5794494639335e-06,
+      "loss": 0.5659,
+      "step": 5265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575924913282131e-06,
+      "loss": 0.6414,
+      "step": 5266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572400682530917e-06,
+      "loss": 0.6775,
+      "step": 5267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568876772144942e-06,
+      "loss": 0.6694,
+      "step": 5268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565353182589248e-06,
+      "loss": 0.786,
+      "step": 5269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561829914328835e-06,
+      "loss": 0.6239,
+      "step": 5270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558306967828661e-06,
+      "loss": 0.643,
+      "step": 5271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5547843435536425e-06,
+      "loss": 0.6021,
+      "step": 5272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551262041968656e-06,
+      "loss": 0.5601,
+      "step": 5273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547740063538528e-06,
+      "loss": 0.6682,
+      "step": 5274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5442184087280456e-06,
+      "loss": 0.6461,
+      "step": 5275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5406970780019595e-06,
+      "loss": 0.5907,
+      "step": 5276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537176071824972e-06,
+      "loss": 0.6921,
+      "step": 5277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533655390661739e-06,
+      "loss": 0.629,
+      "step": 5278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5301350349768805e-06,
+      "loss": 0.6642,
+      "step": 5279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52661500523497e-06,
+      "loss": 0.6101,
+      "step": 5280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523095301900534e-06,
+      "loss": 0.7183,
+      "step": 5281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5195759254380674e-06,
+      "loss": 0.7236,
+      "step": 5282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516056876312012e-06,
+      "loss": 0.62,
+      "step": 5283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51253815498677e-06,
+      "loss": 0.7341,
+      "step": 5284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509019761926694e-06,
+      "loss": 0.7443,
+      "step": 5285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505501697596109e-06,
+      "loss": 0.7246,
+      "step": 5286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501983962459282e-06,
+      "loss": 0.6552,
+      "step": 5287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4984665569804395e-06,
+      "loss": 0.5951,
+      "step": 5288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494949481623769e-06,
+      "loss": 0.6064,
+      "step": 5289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49143273685341e-06,
+      "loss": 0.6405,
+      "step": 5290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487916323133458e-06,
+      "loss": 0.6388,
+      "step": 5291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484400240927971e-06,
+      "loss": 0.5753,
+      "step": 5292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480884490700957e-06,
+      "loss": 0.6668,
+      "step": 5293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.477369072916383e-06,
+      "loss": 0.6026,
+      "step": 5294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4738539880381665e-06,
+      "loss": 0.7198,
+      "step": 5295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470339236530194e-06,
+      "loss": 0.6405,
+      "step": 5296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466824818856296e-06,
+      "loss": 0.6474,
+      "step": 5297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463310735480265e-06,
+      "loss": 0.69,
+      "step": 5298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459796986865844e-06,
+      "loss": 0.6339,
+      "step": 5299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456283573476733e-06,
+      "loss": 0.6283,
+      "step": 5300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4527704957765965e-06,
+      "loss": 0.6491,
+      "step": 5301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449257754229046e-06,
+      "loss": 0.65,
+      "step": 5302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445745349297648e-06,
+      "loss": 0.6754,
+      "step": 5303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442233281445929e-06,
+      "loss": 0.5354,
+      "step": 5304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438721551137367e-06,
+      "loss": 0.6707,
+      "step": 5305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435210158835402e-06,
+      "loss": 0.6371,
+      "step": 5306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4316991050034245e-06,
+      "loss": 0.684,
+      "step": 5307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428188390104778e-06,
+      "loss": 0.5896,
+      "step": 5308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424678014602765e-06,
+      "loss": 0.6483,
+      "step": 5309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421167978960645e-06,
+      "loss": 0.6764,
+      "step": 5310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417658283641629e-06,
+      "loss": 0.6322,
+      "step": 5311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414148929108882e-06,
+      "loss": 0.6625,
+      "step": 5312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410639915825529e-06,
+      "loss": 0.6895,
+      "step": 5313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407131244254646e-06,
+      "loss": 0.6834,
+      "step": 5314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403622914859261e-06,
+      "loss": 0.5786,
+      "step": 5315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.400114928102369e-06,
+      "loss": 0.6995,
+      "step": 5316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396607284446906e-06,
+      "loss": 0.6455,
+      "step": 5317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39309998435577e-06,
+      "loss": 0.6501,
+      "step": 5318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389593028291815e-06,
+      "loss": 0.638,
+      "step": 5319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386086416717844e-06,
+      "loss": 0.6944,
+      "step": 5320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382580150096616e-06,
+      "loss": 0.5824,
+      "step": 5321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379074228890847e-06,
+      "loss": 0.5437,
+      "step": 5322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375568653563206e-06,
+      "loss": 0.5882,
+      "step": 5323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3720634245763166e-06,
+      "loss": 0.6099,
+      "step": 5324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368558542392754e-06,
+      "loss": 0.5328,
+      "step": 5325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.365054007475056e-06,
+      "loss": 0.6669,
+      "step": 5326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361549820285703e-06,
+      "loss": 0.5985,
+      "step": 5327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358045981287141e-06,
+      "loss": 0.668,
+      "step": 5328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3545424909417594e-06,
+      "loss": 0.6813,
+      "step": 5329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351039349711911e-06,
+      "loss": 0.5832,
+      "step": 5330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347536558059893e-06,
+      "loss": 0.6587,
+      "step": 5331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344034116447964e-06,
+      "loss": 0.6611,
+      "step": 5332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340532025338336e-06,
+      "loss": 0.5859,
+      "step": 5333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337030285193165e-06,
+      "loss": 0.778,
+      "step": 5334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33352889647458e-06,
+      "loss": 0.6978,
+      "step": 5335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330027859644644e-06,
+      "loss": 0.6221,
+      "step": 5336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326527175165382e-06,
+      "loss": 0.6592,
+      "step": 5337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323026843498776e-06,
+      "loss": 0.7034,
+      "step": 5338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319526865106755e-06,
+      "loss": 0.6848,
+      "step": 5339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316027240451205e-06,
+      "loss": 0.6189,
+      "step": 5340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312527969993961e-06,
+      "loss": 0.6024,
+      "step": 5341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309029054196818e-06,
+      "loss": 0.6198,
+      "step": 5342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305530493521517e-06,
+      "loss": 0.6449,
+      "step": 5343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3020322884297565e-06,
+      "loss": 0.6942,
+      "step": 5344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298534439383193e-06,
+      "loss": 0.6885,
+      "step": 5345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295036946843423e-06,
+      "loss": 0.6122,
+      "step": 5346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29153981127201e-06,
+      "loss": 0.6959,
+      "step": 5347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288043033130461e-06,
+      "loss": 0.6567,
+      "step": 5348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284546612880235e-06,
+      "loss": 0.6043,
+      "step": 5349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281050550982752e-06,
+      "loss": 0.673,
+      "step": 5350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2775548478993785e-06,
+      "loss": 0.6325,
+      "step": 5351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274059504091432e-06,
+      "loss": 0.6426,
+      "step": 5352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270564520020191e-06,
+      "loss": 0.6654,
+      "step": 5353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267069896146877e-06,
+      "loss": 0.6332,
+      "step": 5354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263575632932669e-06,
+      "loss": 0.6031,
+      "step": 5355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260081730838703e-06,
+      "loss": 0.7402,
+      "step": 5356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256588190326057e-06,
+      "loss": 0.6135,
+      "step": 5357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2530950118557666e-06,
+      "loss": 0.6959,
+      "step": 5358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.249602195888819e-06,
+      "loss": 0.7173,
+      "step": 5359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246109742886156e-06,
+      "loss": 0.5864,
+      "step": 5360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242617653308668e-06,
+      "loss": 0.5915,
+      "step": 5361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239125927617197e-06,
+      "loss": 0.6152,
+      "step": 5362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235634566272542e-06,
+      "loss": 0.7042,
+      "step": 5363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232143569735446e-06,
+      "loss": 0.6531,
+      "step": 5364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2286529384666145e-06,
+      "loss": 0.6745,
+      "step": 5365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225162672926698e-06,
+      "loss": 0.6549,
+      "step": 5366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221672773576297e-06,
+      "loss": 0.6783,
+      "step": 5367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218183240875967e-06,
+      "loss": 0.6349,
+      "step": 5368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2146940752862145e-06,
+      "loss": 0.6208,
+      "step": 5369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211205277267498e-06,
+      "loss": 0.6118,
+      "step": 5370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207716847280224e-06,
+      "loss": 0.6657,
+      "step": 5371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204228785784757e-06,
+      "loss": 0.5598,
+      "step": 5372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20074109324141e-06,
+      "loss": 0.5651,
+      "step": 5373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197253770110438e-06,
+      "loss": 0.6751,
+      "step": 5374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193766816852068e-06,
+      "loss": 0.6218,
+      "step": 5375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1902802339264586e-06,
+      "loss": 0.5529,
+      "step": 5376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186794021793731e-06,
+      "loss": 0.6288,
+      "step": 5377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1833081809139485e-06,
+      "loss": 0.6572,
+      "step": 5378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179822711747134e-06,
+      "loss": 0.7659,
+      "step": 5379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176337614753253e-06,
+      "loss": 0.6097,
+      "step": 5380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172852890392233e-06,
+      "loss": 0.6718,
+      "step": 5381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169368539123943e-06,
+      "loss": 0.6305,
+      "step": 5382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165884561408205e-06,
+      "loss": 0.5691,
+      "step": 5383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162400957704787e-06,
+      "loss": 0.6978,
+      "step": 5384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158917728473425e-06,
+      "loss": 0.6655,
+      "step": 5385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155434874173786e-06,
+      "loss": 0.6287,
+      "step": 5386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151952395265495e-06,
+      "loss": 0.5973,
+      "step": 5387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148470292208129e-06,
+      "loss": 0.694,
+      "step": 5388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144988565461213e-06,
+      "loss": 0.5937,
+      "step": 5389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.141507215484225e-06,
+      "loss": 0.7027,
+      "step": 5390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1380262427365885e-06,
+      "loss": 0.5874,
+      "step": 5391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134545647677684e-06,
+      "loss": 0.5567,
+      "step": 5392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131065430766831e-06,
+      "loss": 0.6437,
+      "step": 5393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127585592463317e-06,
+      "loss": 0.669,
+      "step": 5394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124106133226363e-06,
+      "loss": 0.5969,
+      "step": 5395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1206270535151475e-06,
+      "loss": 0.6849,
+      "step": 5396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1171483537887964e-06,
+      "loss": 0.6902,
+      "step": 5397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113670034506388e-06,
+      "loss": 0.6544,
+      "step": 5398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110192096126944e-06,
+      "loss": 0.5911,
+      "step": 5399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106714539109448e-06,
+      "loss": 0.6293,
+      "step": 5400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.103237363912823e-06,
+      "loss": 0.6547,
+      "step": 5401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099760570995943e-06,
+      "loss": 0.596,
+      "step": 5402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096284160817629e-06,
+      "loss": 0.6845,
+      "step": 5403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092808133836666e-06,
+      "loss": 0.6401,
+      "step": 5404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089332490511773e-06,
+      "loss": 0.6549,
+      "step": 5405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085857231301623e-06,
+      "loss": 0.6284,
+      "step": 5406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08238235666484e-06,
+      "loss": 0.6197,
+      "step": 5407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078907867059992e-06,
+      "loss": 0.5842,
+      "step": 5408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075433762945605e-06,
+      "loss": 0.746,
+      "step": 5409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071960044780148e-06,
+      "loss": 0.5675,
+      "step": 5410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0684867130220395e-06,
+      "loss": 0.6379,
+      "step": 5411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065013768129649e-06,
+      "loss": 0.6899,
+      "step": 5412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061541210561287e-06,
+      "loss": 0.6479,
+      "step": 5413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058069040775231e-06,
+      "loss": 0.5862,
+      "step": 5414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054597259229692e-06,
+      "loss": 0.6107,
+      "step": 5415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051125866382831e-06,
+      "loss": 0.5871,
+      "step": 5416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047654862692761e-06,
+      "loss": 0.6286,
+      "step": 5417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044184248617545e-06,
+      "loss": 0.6625,
+      "step": 5418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040714024615194e-06,
+      "loss": 0.6221,
+      "step": 5419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037244191143662e-06,
+      "loss": 0.6965,
+      "step": 5420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03377474866086e-06,
+      "loss": 0.6041,
+      "step": 5421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0303056976246394e-06,
+      "loss": 0.6216,
+      "step": 5422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026837038492802e-06,
+      "loss": 0.6301,
+      "step": 5423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023368771723107e-06,
+      "loss": 0.5847,
+      "step": 5424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019900897773248e-06,
+      "loss": 0.7069,
+      "step": 5425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016433417100876e-06,
+      "loss": 0.7313,
+      "step": 5426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012966330163588e-06,
+      "loss": 0.6367,
+      "step": 5427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009499637418926e-06,
+      "loss": 0.5942,
+      "step": 5428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006033339324384e-06,
+      "loss": 0.5932,
+      "step": 5429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002567436337401e-06,
+      "loss": 0.6763,
+      "step": 5430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999101928915365e-06,
+      "loss": 0.6762,
+      "step": 5431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.995636817515612e-06,
+      "loss": 0.6744,
+      "step": 5432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992172102595421e-06,
+      "loss": 0.6813,
+      "step": 5433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988707784612032e-06,
+      "loss": 0.7759,
+      "step": 5434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98524386402262e-06,
+      "loss": 0.7237,
+      "step": 5435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9817803412843095e-06,
+      "loss": 0.6338,
+      "step": 5436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978317216854178e-06,
+      "loss": 0.6758,
+      "step": 5437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974854491189243e-06,
+      "loss": 0.609,
+      "step": 5438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971392164746476e-06,
+      "loss": 0.5338,
+      "step": 5439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967930237982793e-06,
+      "loss": 0.6609,
+      "step": 5440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9644687113550555e-06,
+      "loss": 0.6944,
+      "step": 5441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961007585320072e-06,
+      "loss": 0.7182,
+      "step": 5442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957546860334604e-06,
+      "loss": 0.5628,
+      "step": 5443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954086536855358e-06,
+      "loss": 0.6139,
+      "step": 5444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950626615338978e-06,
+      "loss": 0.528,
+      "step": 5445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947167096242071e-06,
+      "loss": 0.587,
+      "step": 5446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943707980021179e-06,
+      "loss": 0.6256,
+      "step": 5447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940249267132795e-06,
+      "loss": 0.5898,
+      "step": 5448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936790958033357e-06,
+      "loss": 0.6752,
+      "step": 5449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93333305317925e-06,
+      "loss": 0.5819,
+      "step": 5450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929875553026808e-06,
+      "loss": 0.6161,
+      "step": 5451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926418458032309e-06,
+      "loss": 0.6212,
+      "step": 5452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9229617686519814e-06,
+      "loss": 0.6662,
+      "step": 5453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919505485341993e-06,
+      "loss": 0.6576,
+      "step": 5454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916049608558468e-06,
+      "loss": 0.7279,
+      "step": 5455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912594138757468e-06,
+      "loss": 0.6738,
+      "step": 5456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909139076395004e-06,
+      "loss": 0.5827,
+      "step": 5457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9056844219270334e-06,
+      "loss": 0.653,
+      "step": 5458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9022301758094616e-06,
+      "loss": 0.7131,
+      "step": 5459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898776338498135e-06,
+      "loss": 0.7272,
+      "step": 5460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89532291044885e-06,
+      "loss": 0.6899,
+      "step": 5461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8918698921173485e-06,
+      "loss": 0.608,
+      "step": 5462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888417283959321e-06,
+      "loss": 0.6309,
+      "step": 5463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884965086430399e-06,
+      "loss": 0.6019,
+      "step": 5464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881513299986164e-06,
+      "loss": 0.7358,
+      "step": 5465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878061925082138e-06,
+      "loss": 0.6317,
+      "step": 5466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874610962173793e-06,
+      "loss": 0.5771,
+      "step": 5467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871160411716544e-06,
+      "loss": 0.6044,
+      "step": 5468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867710274165755e-06,
+      "loss": 0.6617,
+      "step": 5469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864260549976731e-06,
+      "loss": 0.7045,
+      "step": 5470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86081123960473e-06,
+      "loss": 0.6335,
+      "step": 5471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857362343504943e-06,
+      "loss": 0.6802,
+      "step": 5472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853913862132518e-06,
+      "loss": 0.5547,
+      "step": 5473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850465795942546e-06,
+      "loss": 0.5688,
+      "step": 5474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847018145390059e-06,
+      "loss": 0.7198,
+      "step": 5475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843570910930034e-06,
+      "loss": 0.5514,
+      "step": 5476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840124093017398e-06,
+      "loss": 0.6886,
+      "step": 5477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836677692107021e-06,
+      "loss": 0.6265,
+      "step": 5478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833231708653712e-06,
+      "loss": 0.5867,
+      "step": 5479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829786143112238e-06,
+      "loss": 0.5217,
+      "step": 5480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8263409959372995e-06,
+      "loss": 0.593,
+      "step": 5481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822896267583541e-06,
+      "loss": 0.6525,
+      "step": 5482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8194519585055644e-06,
+      "loss": 0.6964,
+      "step": 5483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816008069157903e-06,
+      "loss": 0.6531,
+      "step": 5484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812564599995042e-06,
+      "loss": 0.6923,
+      "step": 5485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809121551471408e-06,
+      "loss": 0.6914,
+      "step": 5486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805678924041373e-06,
+      "loss": 0.6617,
+      "step": 5487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80223671815925e-06,
+      "loss": 0.6584,
+      "step": 5488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7987949342793045e-06,
+      "loss": 0.5963,
+      "step": 5489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.79535357285574e-06,
+      "loss": 0.5682,
+      "step": 5490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791912634342708e-06,
+      "loss": 0.5947,
+      "step": 5491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788472119194294e-06,
+      "loss": 0.6152,
+      "step": 5492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785032027864544e-06,
+      "loss": 0.7249,
+      "step": 5493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781592360807438e-06,
+      "loss": 0.6116,
+      "step": 5494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7781531184769026e-06,
+      "loss": 0.6438,
+      "step": 5495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774714301326804e-06,
+      "loss": 0.645,
+      "step": 5496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771275909810955e-06,
+      "loss": 0.622,
+      "step": 5497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7678379443831175e-06,
+      "loss": 0.6692,
+      "step": 5498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764400405496991e-06,
+      "loss": 0.72,
+      "step": 5499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76096329360622e-06,
+      "loss": 0.6157,
+      "step": 5500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757526609164393e-06,
+      "loss": 0.6128,
+      "step": 5501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7540903526250375e-06,
+      "loss": 0.6858,
+      "step": 5502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.750654524441638e-06,
+      "loss": 0.5888,
+      "step": 5503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747219125067609e-06,
+      "loss": 0.604,
+      "step": 5504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743784154956313e-06,
+      "loss": 0.6886,
+      "step": 5505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740349614561057e-06,
+      "loss": 0.5651,
+      "step": 5506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736915504335086e-06,
+      "loss": 0.6206,
+      "step": 5507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733481824731598e-06,
+      "loss": 0.6965,
+      "step": 5508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7300485762037275e-06,
+      "loss": 0.604,
+      "step": 5509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726615759204551e-06,
+      "loss": 0.7266,
+      "step": 5510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723183374187088e-06,
+      "loss": 0.6986,
+      "step": 5511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719751421604309e-06,
+      "loss": 0.6911,
+      "step": 5512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716319901909119e-06,
+      "loss": 0.7409,
+      "step": 5513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712888815554371e-06,
+      "loss": 0.6896,
+      "step": 5514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709458162992853e-06,
+      "loss": 0.5731,
+      "step": 5515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706027944677302e-06,
+      "loss": 0.6413,
+      "step": 5516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702598161060403e-06,
+      "loss": 0.6408,
+      "step": 5517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699168812594771e-06,
+      "loss": 0.7062,
+      "step": 5518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695739899732973e-06,
+      "loss": 0.6757,
+      "step": 5519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692311422927515e-06,
+      "loss": 0.6738,
+      "step": 5520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688883382630842e-06,
+      "loss": 0.7347,
+      "step": 5521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6854557792953515e-06,
+      "loss": 0.7144,
+      "step": 5522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682028613373376e-06,
+      "loss": 0.6341,
+      "step": 5523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67860188531719e-06,
+      "loss": 0.6551,
+      "step": 5524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675175595579009e-06,
+      "loss": 0.6335,
+      "step": 5525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671749744610998e-06,
+      "loss": 0.5872,
+      "step": 5526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.668324332865259e-06,
+      "loss": 0.6605,
+      "step": 5527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664899360793834e-06,
+      "loss": 0.6528,
+      "step": 5528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661474828848711e-06,
+      "loss": 0.6234,
+      "step": 5529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658050737481818e-06,
+      "loss": 0.5307,
+      "step": 5530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65462708714502e-06,
+      "loss": 0.6176,
+      "step": 5531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.651203878290139e-06,
+      "loss": 0.7055,
+      "step": 5532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647781111368924e-06,
+      "loss": 0.609,
+      "step": 5533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6443587868330676e-06,
+      "loss": 0.5022,
+      "step": 5534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640936905134212e-06,
+      "loss": 0.6675,
+      "step": 5535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637515466723934e-06,
+      "loss": 0.6519,
+      "step": 5536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634094472053752e-06,
+      "loss": 0.6027,
+      "step": 5537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630673921575129e-06,
+      "loss": 0.6212,
+      "step": 5538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627253815739467e-06,
+      "loss": 0.6388,
+      "step": 5539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623834154998112e-06,
+      "loss": 0.6337,
+      "step": 5540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620414939802345e-06,
+      "loss": 0.7397,
+      "step": 5541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616996170603398e-06,
+      "loss": 0.6751,
+      "step": 5542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613577847852439e-06,
+      "loss": 0.6549,
+      "step": 5543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610159972000572e-06,
+      "loss": 0.6673,
+      "step": 5544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606742543498852e-06,
+      "loss": 0.7275,
+      "step": 5545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603325562798268e-06,
+      "loss": 0.6734,
+      "step": 5546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599909030349752e-06,
+      "loss": 0.6534,
+      "step": 5547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596492946604174e-06,
+      "loss": 0.6397,
+      "step": 5548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593077312012353e-06,
+      "loss": 0.6413,
+      "step": 5549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589662127025036e-06,
+      "loss": 0.7335,
+      "step": 5550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586247392092923e-06,
+      "loss": 0.664,
+      "step": 5551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5828331076666495e-06,
+      "loss": 0.6136,
+      "step": 5552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579419274196789e-06,
+      "loss": 0.6628,
+      "step": 5553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576005892133861e-06,
+      "loss": 0.5859,
+      "step": 5554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572592961928322e-06,
+      "loss": 0.6406,
+      "step": 5555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5691804840305665e-06,
+      "loss": 0.6341,
+      "step": 5556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565768458890935e-06,
+      "loss": 0.6045,
+      "step": 5557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562356886959704e-06,
+      "loss": 0.6578,
+      "step": 5558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558945768687089e-06,
+      "loss": 0.5363,
+      "step": 5559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555535104523252e-06,
+      "loss": 0.6004,
+      "step": 5560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5521248949182915e-06,
+      "loss": 0.7359,
+      "step": 5561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548715140322242e-06,
+      "loss": 0.627,
+      "step": 5562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545305841185087e-06,
+      "loss": 0.6523,
+      "step": 5563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5418969979567415e-06,
+      "loss": 0.6162,
+      "step": 5564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538488611087064e-06,
+      "loss": 0.6466,
+      "step": 5565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5350806810258515e-06,
+      "loss": 0.5798,
+      "step": 5566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5316732082228415e-06,
+      "loss": 0.5908,
+      "step": 5567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5282661931277126e-06,
+      "loss": 0.6559,
+      "step": 5568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524859636190077e-06,
+      "loss": 0.705,
+      "step": 5569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521453537859497e-06,
+      "loss": 0.6961,
+      "step": 5570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518047898585462e-06,
+      "loss": 0.5588,
+      "step": 5571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514642718817417e-06,
+      "loss": 0.5384,
+      "step": 5572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511237999004727e-06,
+      "loss": 0.5461,
+      "step": 5573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5078337395967104e-06,
+      "loss": 0.6133,
+      "step": 5574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504429941042619e-06,
+      "loss": 0.7367,
+      "step": 5575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501026603791645e-06,
+      "loss": 0.6478,
+      "step": 5576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49762372829292e-06,
+      "loss": 0.6477,
+      "step": 5577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4942213149955115e-06,
+      "loss": 0.6002,
+      "step": 5578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490819364348434e-06,
+      "loss": 0.6621,
+      "step": 5579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487417876800632e-06,
+      "loss": 0.5488,
+      "step": 5580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484016852800993e-06,
+      "loss": 0.6449,
+      "step": 5581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480616292798346e-06,
+      "loss": 0.6783,
+      "step": 5582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477216197241455e-06,
+      "loss": 0.5931,
+      "step": 5583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473816566579023e-06,
+      "loss": 0.696,
+      "step": 5584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470417401259691e-06,
+      "loss": 0.6872,
+      "step": 5585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467018701732041e-06,
+      "loss": 0.5722,
+      "step": 5586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463620468444589e-06,
+      "loss": 0.781,
+      "step": 5587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4602227018457985e-06,
+      "loss": 0.6412,
+      "step": 5588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4568254023840636e-06,
+      "loss": 0.6844,
+      "step": 5589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453428570507712e-06,
+      "loss": 0.6305,
+      "step": 5590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450032206665029e-06,
+      "loss": 0.6306,
+      "step": 5591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446636311304218e-06,
+      "loss": 0.5503,
+      "step": 5592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.443240884873431e-06,
+      "loss": 0.6649,
+      "step": 5593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439845927820753e-06,
+      "loss": 0.6741,
+      "step": 5594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43645144059421e-06,
+      "loss": 0.6247,
+      "step": 5595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433057423641764e-06,
+      "loss": 0.626,
+      "step": 5596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429663877411321e-06,
+      "loss": 0.6571,
+      "step": 5597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426270802350717e-06,
+      "loss": 0.6824,
+      "step": 5598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4228781989077295e-06,
+      "loss": 0.6379,
+      "step": 5599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4194860675300695e-06,
+      "loss": 0.7118,
+      "step": 5600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416094408665396e-06,
+      "loss": 0.69,
+      "step": 5601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4127032227612985e-06,
+      "loss": 0.6619,
+      "step": 5602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409312510265302e-06,
+      "loss": 0.6713,
+      "step": 5603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405922271624874e-06,
+      "loss": 0.6281,
+      "step": 5604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402532507287413e-06,
+      "loss": 0.5582,
+      "step": 5605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.399143217700263e-06,
+      "loss": 0.599,
+      "step": 5606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395754403310701e-06,
+      "loss": 0.6128,
+      "step": 5607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392366064565941e-06,
+      "loss": 0.621,
+      "step": 5608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388978201913136e-06,
+      "loss": 0.6676,
+      "step": 5609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385590815799369e-06,
+      "loss": 0.5948,
+      "step": 5610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382203906671678e-06,
+      "loss": 0.761,
+      "step": 5611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378817474977019e-06,
+      "loss": 0.5503,
+      "step": 5612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375431521162294e-06,
+      "loss": 0.6771,
+      "step": 5613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37204604567434e-06,
+      "loss": 0.678,
+      "step": 5614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368661048959929e-06,
+      "loss": 0.6238,
+      "step": 5615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365276531465775e-06,
+      "loss": 0.6206,
+      "step": 5616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361892493638526e-06,
+      "loss": 0.6604,
+      "step": 5617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358508935924764e-06,
+      "loss": 0.6956,
+      "step": 5618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3551258587710074e-06,
+      "loss": 0.6417,
+      "step": 5619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351743262623722e-06,
+      "loss": 0.6099,
+      "step": 5620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348361147929298e-06,
+      "loss": 0.7252,
+      "step": 5621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3449795151340656e-06,
+      "loss": 0.6743,
+      "step": 5622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341598364684292e-06,
+      "loss": 0.6613,
+      "step": 5623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338217697026179e-06,
+      "loss": 0.6794,
+      "step": 5624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3348375126058696e-06,
+      "loss": 0.5687,
+      "step": 5625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331457811869437e-06,
+      "loss": 0.6419,
+      "step": 5626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3280785952628945e-06,
+      "loss": 0.716,
+      "step": 5627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324699863232191e-06,
+      "loss": 0.6481,
+      "step": 5628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321321616223205e-06,
+      "loss": 0.5877,
+      "step": 5629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317943854681765e-06,
+      "loss": 0.7369,
+      "step": 5630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3145665790536245e-06,
+      "loss": 0.6098,
+      "step": 5631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311189789784475e-06,
+      "loss": 0.7663,
+      "step": 5632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3078134873199424e-06,
+      "loss": 0.6476,
+      "step": 5633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304437672105594e-06,
+      "loss": 0.6851,
+      "step": 5634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301062344586926e-06,
+      "loss": 0.6958,
+      "step": 5635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297687505209375e-06,
+      "loss": 0.6321,
+      "step": 5636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294313154418312e-06,
+      "loss": 0.6015,
+      "step": 5637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290939292659042e-06,
+      "loss": 0.5276,
+      "step": 5638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287565920376802e-06,
+      "loss": 0.6294,
+      "step": 5639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284193038016778e-06,
+      "loss": 0.5885,
+      "step": 5640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280820646024078e-06,
+      "loss": 0.6407,
+      "step": 5641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277448744843749e-06,
+      "loss": 0.7108,
+      "step": 5642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274077334920773e-06,
+      "loss": 0.5212,
+      "step": 5643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27070641670007e-06,
+      "loss": 0.6318,
+      "step": 5644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267335990626493e-06,
+      "loss": 0.6311,
+      "step": 5645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263966057144829e-06,
+      "loss": 0.5732,
+      "step": 5646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.260596616699802e-06,
+      "loss": 0.5519,
+      "step": 5647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25722766973607e-06,
+      "loss": 0.6469,
+      "step": 5648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.253859216698219e-06,
+      "loss": 0.6438,
+      "step": 5649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2504912580307905e-06,
+      "loss": 0.618,
+      "step": 5650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247123794178238e-06,
+      "loss": 0.7459,
+      "step": 5651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243756825584959e-06,
+      "loss": 0.6635,
+      "step": 5652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240390352695288e-06,
+      "loss": 0.6242,
+      "step": 5653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237024375953491e-06,
+      "loss": 0.5678,
+      "step": 5654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233658895803767e-06,
+      "loss": 0.6774,
+      "step": 5655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230293912690255e-06,
+      "loss": 0.719,
+      "step": 5656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22692942705702e-06,
+      "loss": 0.7003,
+      "step": 5657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223565439348067e-06,
+      "loss": 0.6484,
+      "step": 5658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220201950007338e-06,
+      "loss": 0.6378,
+      "step": 5659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216838959478704e-06,
+      "loss": 0.5912,
+      "step": 5660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21347646820597e-06,
+      "loss": 0.7212,
+      "step": 5661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210114476632881e-06,
+      "loss": 0.6221,
+      "step": 5662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20675298520311e-06,
+      "loss": 0.6774,
+      "step": 5663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.203391994360265e-06,
+      "loss": 0.6344,
+      "step": 5664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2000315045478896e-06,
+      "loss": 0.7264,
+      "step": 5665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196671516209462e-06,
+      "loss": 0.6166,
+      "step": 5666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193312029788388e-06,
+      "loss": 0.6811,
+      "step": 5667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189953045728019e-06,
+      "loss": 0.7312,
+      "step": 5668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186594564471627e-06,
+      "loss": 0.6459,
+      "step": 5669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1832365864624265e-06,
+      "loss": 0.5942,
+      "step": 5670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179879112143566e-06,
+      "loss": 0.6757,
+      "step": 5671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176522141958121e-06,
+      "loss": 0.601,
+      "step": 5672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.7286,
+      "step": 5673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16980971575946e-06,
+      "loss": 0.6778,
+      "step": 5674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166454260632071e-06,
+      "loss": 0.6824,
+      "step": 5675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163099311409748e-06,
+      "loss": 0.569,
+      "step": 5676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1597448685352336e-06,
+      "loss": 0.5271,
+      "step": 5677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156390932451208e-06,
+      "loss": 0.6344,
+      "step": 5678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153037503600288e-06,
+      "loss": 0.6254,
+      "step": 5679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149684582425013e-06,
+      "loss": 0.6848,
+      "step": 5680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146332169367866e-06,
+      "loss": 0.5893,
+      "step": 5681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1429802648712546e-06,
+      "loss": 0.7077,
+      "step": 5682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139628869377523e-06,
+      "loss": 0.6967,
+      "step": 5683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136277983328949e-06,
+      "loss": 0.6365,
+      "step": 5684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1329276071677405e-06,
+      "loss": 0.6156,
+      "step": 5685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129577741336039e-06,
+      "loss": 0.5822,
+      "step": 5686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126228386275923e-06,
+      "loss": 0.6517,
+      "step": 5687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122879542429395e-06,
+      "loss": 0.6476,
+      "step": 5688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119531210238398e-06,
+      "loss": 0.6482,
+      "step": 5689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1161833901448055e-06,
+      "loss": 0.7192,
+      "step": 5690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.112836082590423e-06,
+      "loss": 0.6111,
+      "step": 5691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109489288016983e-06,
+      "loss": 0.6456,
+      "step": 5692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.10614300686616e-06,
+      "loss": 0.7576,
+      "step": 5693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102797239579552e-06,
+      "loss": 0.6442,
+      "step": 5694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099451986598695e-06,
+      "loss": 0.6577,
+      "step": 5695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0961072483650526e-06,
+      "loss": 0.6046,
+      "step": 5696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092763025320027e-06,
+      "loss": 0.6757,
+      "step": 5697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089419317904943e-06,
+      "loss": 0.7151,
+      "step": 5698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086076126561069e-06,
+      "loss": 0.6279,
+      "step": 5699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082733451729597e-06,
+      "loss": 0.562,
+      "step": 5700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079391293851651e-06,
+      "loss": 0.7043,
+      "step": 5701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07604965336829e-06,
+      "loss": 0.5641,
+      "step": 5702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072708530720503e-06,
+      "loss": 0.6028,
+      "step": 5703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069367926349208e-06,
+      "loss": 0.6163,
+      "step": 5704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0660278406952635e-06,
+      "loss": 0.5924,
+      "step": 5705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062688274199448e-06,
+      "loss": 0.5838,
+      "step": 5706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059349227302482e-06,
+      "loss": 0.6779,
+      "step": 5707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056010700445008e-06,
+      "loss": 0.675,
+      "step": 5708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052672694067608e-06,
+      "loss": 0.6552,
+      "step": 5709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049335208610792e-06,
+      "loss": 0.6447,
+      "step": 5710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045998244514999e-06,
+      "loss": 0.7189,
+      "step": 5711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042661802220602e-06,
+      "loss": 0.6091,
+      "step": 5712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039325882167904e-06,
+      "loss": 0.5804,
+      "step": 5713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0359904847971386e-06,
+      "loss": 0.6882,
+      "step": 5714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032655610548474e-06,
+      "loss": 0.5843,
+      "step": 5715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029321259862004e-06,
+      "loss": 0.6957,
+      "step": 5716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0259874331777575e-06,
+      "loss": 0.6812,
+      "step": 5717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022654130935688e-06,
+      "loss": 0.666,
+      "step": 5718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019321353575692e-06,
+      "loss": 0.601,
+      "step": 5719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015989101537586e-06,
+      "loss": 0.6671,
+      "step": 5720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012657375261121e-06,
+      "loss": 0.6221,
+      "step": 5721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009326175185975e-06,
+      "loss": 0.6907,
+      "step": 5722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0059955017517615e-06,
+      "loss": 0.6026,
+      "step": 5723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002665355398024e-06,
+      "loss": 0.5018,
+      "step": 5724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9993357365642336e-06,
+      "loss": 0.662,
+      "step": 5725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996006645689793e-06,
+      "loss": 0.6536,
+      "step": 5726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992678083214036e-06,
+      "loss": 0.5978,
+      "step": 5727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9893500495762214e-06,
+      "loss": 0.7091,
+      "step": 5728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986022545215553e-06,
+      "loss": 0.5835,
+      "step": 5729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982695570571147e-06,
+      "loss": 0.6053,
+      "step": 5730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97936912608206e-06,
+      "loss": 0.6489,
+      "step": 5731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976043212187273e-06,
+      "loss": 0.6091,
+      "step": 5732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972717829325703e-06,
+      "loss": 0.5703,
+      "step": 5733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9693929779361936e-06,
+      "loss": 0.6769,
+      "step": 5734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966068658457518e-06,
+      "loss": 0.6138,
+      "step": 5735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96274487132838e-06,
+      "loss": 0.6598,
+      "step": 5736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959421616987409e-06,
+      "loss": 0.6249,
+      "step": 5737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956098895873174e-06,
+      "loss": 0.526,
+      "step": 5738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952776708424166e-06,
+      "loss": 0.6327,
+      "step": 5739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949455055078806e-06,
+      "loss": 0.5638,
+      "step": 5740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946133936275443e-06,
+      "loss": 0.6572,
+      "step": 5741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942813352452365e-06,
+      "loss": 0.5703,
+      "step": 5742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939493304047777e-06,
+      "loss": 0.7219,
+      "step": 5743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936173791499822e-06,
+      "loss": 0.6347,
+      "step": 5744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932854815246566e-06,
+      "loss": 0.6146,
+      "step": 5745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929536375726011e-06,
+      "loss": 0.6395,
+      "step": 5746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926218473376077e-06,
+      "loss": 0.6865,
+      "step": 5747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922901108634631e-06,
+      "loss": 0.664,
+      "step": 5748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9195842819394545e-06,
+      "loss": 0.6265,
+      "step": 5749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916267993728261e-06,
+      "loss": 0.5439,
+      "step": 5750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9129522444386936e-06,
+      "loss": 0.7121,
+      "step": 5751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909637034508328e-06,
+      "loss": 0.6223,
+      "step": 5752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906322364374664e-06,
+      "loss": 0.6717,
+      "step": 5753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9030082344751305e-06,
+      "loss": 0.6587,
+      "step": 5754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899694645247088e-06,
+      "loss": 0.6311,
+      "step": 5755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896381597127824e-06,
+      "loss": 0.6642,
+      "step": 5756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893069090554548e-06,
+      "loss": 0.7314,
+      "step": 5757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8897571259644175e-06,
+      "loss": 0.6215,
+      "step": 5758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886445703794497e-06,
+      "loss": 0.6263,
+      "step": 5759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883134824481786e-06,
+      "loss": 0.6564,
+      "step": 5760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.879824488463222e-06,
+      "loss": 0.6563,
+      "step": 5761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876514696175659e-06,
+      "loss": 0.6153,
+      "step": 5762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873205448055883e-06,
+      "loss": 0.6921,
+      "step": 5763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86989674454061e-06,
+      "loss": 0.6498,
+      "step": 5764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866588586066481e-06,
+      "loss": 0.5588,
+      "step": 5765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8632809730700655e-06,
+      "loss": 0.5851,
+      "step": 5766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859973905987866e-06,
+      "loss": 0.5818,
+      "step": 5767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856667385256307e-06,
+      "loss": 0.6344,
+      "step": 5768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853361411311744e-06,
+      "loss": 0.6849,
+      "step": 5769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850055984590461e-06,
+      "loss": 0.6521,
+      "step": 5770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846751105528668e-06,
+      "loss": 0.6162,
+      "step": 5771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8434467745625005e-06,
+      "loss": 0.6636,
+      "step": 5772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840142992128024e-06,
+      "loss": 0.619,
+      "step": 5773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836839758661234e-06,
+      "loss": 0.5982,
+      "step": 5774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83353707459805e-06,
+      "loss": 0.6267,
+      "step": 5775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830234940374323e-06,
+      "loss": 0.6317,
+      "step": 5776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826933356425822e-06,
+      "loss": 0.6989,
+      "step": 5777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82363232318826e-06,
+      "loss": 0.6306,
+      "step": 5778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8203318410972595e-06,
+      "loss": 0.6828,
+      "step": 5779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817031910588384e-06,
+      "loss": 0.6943,
+      "step": 5780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813732532097115e-06,
+      "loss": 0.6323,
+      "step": 5781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810433706058864e-06,
+      "loss": 0.6751,
+      "step": 5782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8071354329089736e-06,
+      "loss": 0.6163,
+      "step": 5783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803837713082707e-06,
+      "loss": 0.6937,
+      "step": 5784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800540547015259e-06,
+      "loss": 0.7191,
+      "step": 5785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7972439351417476e-06,
+      "loss": 0.6998,
+      "step": 5786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7939478778972184e-06,
+      "loss": 0.7398,
+      "step": 5787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790652375716653e-06,
+      "loss": 0.632,
+      "step": 5788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7873574290349455e-06,
+      "loss": 0.6355,
+      "step": 5789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784063038286924e-06,
+      "loss": 0.6376,
+      "step": 5790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780769203907342e-06,
+      "loss": 0.6333,
+      "step": 5791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777475926330884e-06,
+      "loss": 0.6518,
+      "step": 5792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774183205992154e-06,
+      "loss": 0.6173,
+      "step": 5793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770891043325686e-06,
+      "loss": 0.6197,
+      "step": 5794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767599438765939e-06,
+      "loss": 0.6844,
+      "step": 5795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7643083927472965e-06,
+      "loss": 0.7187,
+      "step": 5796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761017905704078e-06,
+      "loss": 0.6019,
+      "step": 5797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757727978070518e-06,
+      "loss": 0.6803,
+      "step": 5798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7544386102807825e-06,
+      "loss": 0.6157,
+      "step": 5799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751149802768963e-06,
+      "loss": 0.6761,
+      "step": 5800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7478615559690735e-06,
+      "loss": 0.698,
+      "step": 5801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74457387031506e-06,
+      "loss": 0.7095,
+      "step": 5802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741286746240791e-06,
+      "loss": 0.736,
+      "step": 5803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73800018418006e-06,
+      "loss": 0.5835,
+      "step": 5804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73471418456659e-06,
+      "loss": 0.6336,
+      "step": 5805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731428747834023e-06,
+      "loss": 0.6427,
+      "step": 5806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728143874415937e-06,
+      "loss": 0.703,
+      "step": 5807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724859564745828e-06,
+      "loss": 0.6475,
+      "step": 5808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721575819257114e-06,
+      "loss": 0.6008,
+      "step": 5809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718292638383156e-06,
+      "loss": 0.6533,
+      "step": 5810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71501002255722e-06,
+      "loss": 0.7029,
+      "step": 5811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711727972212509e-06,
+      "loss": 0.7146,
+      "step": 5812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708446487782147e-06,
+      "loss": 0.5563,
+      "step": 5813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7051655696991825e-06,
+      "loss": 0.6016,
+      "step": 5814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701885218396595e-06,
+      "loss": 0.5978,
+      "step": 5815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698605434307281e-06,
+      "loss": 0.6948,
+      "step": 5816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6953262178640744e-06,
+      "loss": 0.6548,
+      "step": 5817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6920475694997214e-06,
+      "loss": 0.7052,
+      "step": 5818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6887694896469e-06,
+      "loss": 0.7903,
+      "step": 5819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685491978738212e-06,
+      "loss": 0.6118,
+      "step": 5820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68221503720618e-06,
+      "loss": 0.6664,
+      "step": 5821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.678938665483259e-06,
+      "loss": 0.5875,
+      "step": 5822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675662864001823e-06,
+      "loss": 0.6806,
+      "step": 5823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672387633194175e-06,
+      "loss": 0.616,
+      "step": 5824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669112973492534e-06,
+      "loss": 0.645,
+      "step": 5825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66583888532906e-06,
+      "loss": 0.6591,
+      "step": 5826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662565369135822e-06,
+      "loss": 0.7144,
+      "step": 5827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659292425344815e-06,
+      "loss": 0.6682,
+      "step": 5828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656020054387971e-06,
+      "loss": 0.5855,
+      "step": 5829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652748256697136e-06,
+      "loss": 0.6828,
+      "step": 5830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64947703270408e-06,
+      "loss": 0.6965,
+      "step": 5831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6462063828405e-06,
+      "loss": 0.6064,
+      "step": 5832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642936307538016e-06,
+      "loss": 0.6702,
+      "step": 5833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6396668072281745e-06,
+      "loss": 0.5205,
+      "step": 5834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636397882342446e-06,
+      "loss": 0.6445,
+      "step": 5835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.633129533312216e-06,
+      "loss": 0.6197,
+      "step": 5836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629861760568813e-06,
+      "loss": 0.6006,
+      "step": 5837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626594564543473e-06,
+      "loss": 0.6313,
+      "step": 5838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62332794566736e-06,
+      "loss": 0.6845,
+      "step": 5839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.620061904371565e-06,
+      "loss": 0.6856,
+      "step": 5840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6167964410871e-06,
+      "loss": 0.6187,
+      "step": 5841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6135315562449e-06,
+      "loss": 0.6875,
+      "step": 5842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610267250275827e-06,
+      "loss": 0.6355,
+      "step": 5843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607003523610659e-06,
+      "loss": 0.6448,
+      "step": 5844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603740376680111e-06,
+      "loss": 0.6524,
+      "step": 5845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6004778099148086e-06,
+      "loss": 0.7196,
+      "step": 5846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597215823745311e-06,
+      "loss": 0.6162,
+      "step": 5847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593954418602094e-06,
+      "loss": 0.645,
+      "step": 5848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590693594915555e-06,
+      "loss": 0.5816,
+      "step": 5849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587433353116023e-06,
+      "loss": 0.7427,
+      "step": 5850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584173693633741e-06,
+      "loss": 0.6315,
+      "step": 5851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580914616898881e-06,
+      "loss": 0.641,
+      "step": 5852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577656123341537e-06,
+      "loss": 0.5791,
+      "step": 5853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574398213391727e-06,
+      "loss": 0.4788,
+      "step": 5854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571140887479383e-06,
+      "loss": 0.6999,
+      "step": 5855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567884146034379e-06,
+      "loss": 0.5924,
+      "step": 5856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564627989486494e-06,
+      "loss": 0.7084,
+      "step": 5857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5613724182654384e-06,
+      "loss": 0.7049,
+      "step": 5858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55811743280084e-06,
+      "loss": 0.6159,
+      "step": 5859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554863033522255e-06,
+      "loss": 0.6442,
+      "step": 5860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55160922085916e-06,
+      "loss": 0.5145,
+      "step": 5861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548355995240949e-06,
+      "loss": 0.5929,
+      "step": 5862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545103357096951e-06,
+      "loss": 0.6132,
+      "step": 5863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5418513068564075e-06,
+      "loss": 0.6697,
+      "step": 5864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5385998449484775e-06,
+      "loss": 0.6183,
+      "step": 5865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535348971802261e-06,
+      "loss": 0.6434,
+      "step": 5866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.532098687846764e-06,
+      "loss": 0.6608,
+      "step": 5867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528848993510919e-06,
+      "loss": 0.6285,
+      "step": 5868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525599889223583e-06,
+      "loss": 0.6914,
+      "step": 5869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52235137541353e-06,
+      "loss": 0.6335,
+      "step": 5870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5191034525094645e-06,
+      "loss": 0.7208,
+      "step": 5871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515856120940005e-06,
+      "loss": 0.6634,
+      "step": 5872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512609381133695e-06,
+      "loss": 0.665,
+      "step": 5873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5093632335190026e-06,
+      "loss": 0.6174,
+      "step": 5874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506117678524309e-06,
+      "loss": 0.6195,
+      "step": 5875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5028727165779306e-06,
+      "loss": 0.6961,
+      "step": 5876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4996283481080955e-06,
+      "loss": 0.6342,
+      "step": 5877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496384573542957e-06,
+      "loss": 0.6982,
+      "step": 5878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493141393310588e-06,
+      "loss": 0.6472,
+      "step": 5879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48989880783898e-06,
+      "loss": 0.5658,
+      "step": 5880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486656817556061e-06,
+      "loss": 0.6114,
+      "step": 5881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483415422889662e-06,
+      "loss": 0.6393,
+      "step": 5882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4801746242675454e-06,
+      "loss": 0.6678,
+      "step": 5883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476934422117393e-06,
+      "loss": 0.6875,
+      "step": 5884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4736948168668015e-06,
+      "loss": 0.6428,
+      "step": 5885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470455808943304e-06,
+      "loss": 0.6557,
+      "step": 5886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.467217398774343e-06,
+      "loss": 0.656,
+      "step": 5887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4639795867872815e-06,
+      "loss": 0.5999,
+      "step": 5888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460742373409409e-06,
+      "loss": 0.5998,
+      "step": 5889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457505759067934e-06,
+      "loss": 0.575,
+      "step": 5890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454269744189986e-06,
+      "loss": 0.6154,
+      "step": 5891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451034329202613e-06,
+      "loss": 0.6213,
+      "step": 5892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44779951453279e-06,
+      "loss": 0.6942,
+      "step": 5893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444565300607404e-06,
+      "loss": 0.6675,
+      "step": 5894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4413316878532664e-06,
+      "loss": 0.636,
+      "step": 5895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438098676697118e-06,
+      "loss": 0.5544,
+      "step": 5896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434866267565608e-06,
+      "loss": 0.6423,
+      "step": 5897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431634460885311e-06,
+      "loss": 0.6353,
+      "step": 5898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428403257082717e-06,
+      "loss": 0.6682,
+      "step": 5899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425172656584252e-06,
+      "loss": 0.5889,
+      "step": 5900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421942659816244e-06,
+      "loss": 0.6128,
+      "step": 5901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418713267204952e-06,
+      "loss": 0.6416,
+      "step": 5902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415484479176549e-06,
+      "loss": 0.6875,
+      "step": 5903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412256296157131e-06,
+      "loss": 0.6924,
+      "step": 5904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409028718572721e-06,
+      "loss": 0.603,
+      "step": 5905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405801746849251e-06,
+      "loss": 0.7621,
+      "step": 5906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402575381412578e-06,
+      "loss": 0.6144,
+      "step": 5907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399349622688479e-06,
+      "loss": 0.6216,
+      "step": 5908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396124471102653e-06,
+      "loss": 0.6479,
+      "step": 5909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392899927080713e-06,
+      "loss": 0.638,
+      "step": 5910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389675991048195e-06,
+      "loss": 0.6655,
+      "step": 5911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3864526634305595e-06,
+      "loss": 0.669,
+      "step": 5912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383229944653177e-06,
+      "loss": 0.7554,
+      "step": 5913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380007835141344e-06,
+      "loss": 0.595,
+      "step": 5914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376786335320281e-06,
+      "loss": 0.553,
+      "step": 5915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373565445615119e-06,
+      "loss": 0.6225,
+      "step": 5916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370345166450908e-06,
+      "loss": 0.6552,
+      "step": 5917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367125498252629e-06,
+      "loss": 0.6758,
+      "step": 5918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363906441445171e-06,
+      "loss": 0.551,
+      "step": 5919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360687996453348e-06,
+      "loss": 0.6991,
+      "step": 5920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35747016370189e-06,
+      "loss": 0.6394,
+      "step": 5921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354252943615448e-06,
+      "loss": 0.573,
+      "step": 5922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351036336618591e-06,
+      "loss": 0.5822,
+      "step": 5923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347820343135807e-06,
+      "loss": 0.6617,
+      "step": 5924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3446049635915085e-06,
+      "loss": 0.7283,
+      "step": 5925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3413901984100195e-06,
+      "loss": 0.6591,
+      "step": 5926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338176048015587e-06,
+      "loss": 0.6399,
+      "step": 5927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3349625128323735e-06,
+      "loss": 0.5942,
+      "step": 5928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331749593284464e-06,
+      "loss": 0.7082,
+      "step": 5929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328537289795861e-06,
+      "loss": 0.6483,
+      "step": 5930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325325602790485e-06,
+      "loss": 0.629,
+      "step": 5931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3221145326921776e-06,
+      "loss": 0.6081,
+      "step": 5932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318904079924689e-06,
+      "loss": 0.6635,
+      "step": 5933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315694244911709e-06,
+      "loss": 0.638,
+      "step": 5934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312485028076824e-06,
+      "loss": 0.6454,
+      "step": 5935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309276429843546e-06,
+      "loss": 0.5907,
+      "step": 5936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306068450635316e-06,
+      "loss": 0.6539,
+      "step": 5937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3028610908754795e-06,
+      "loss": 0.597,
+      "step": 5938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299654350987304e-06,
+      "loss": 0.6346,
+      "step": 5939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296448231393977e-06,
+      "loss": 0.5524,
+      "step": 5940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293242732518604e-06,
+      "loss": 0.5944,
+      "step": 5941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2900378547842065e-06,
+      "loss": 0.6143,
+      "step": 5942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286833598613729e-06,
+      "loss": 0.676,
+      "step": 5943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283629964430023e-06,
+      "loss": 0.5932,
+      "step": 5944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280426952655875e-06,
+      "loss": 0.6743,
+      "step": 5945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277224563713974e-06,
+      "loss": 0.7322,
+      "step": 5946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274022798026935e-06,
+      "loss": 0.6028,
+      "step": 5947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270821656017289e-06,
+      "loss": 0.6631,
+      "step": 5948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2676211381074804e-06,
+      "loss": 0.627,
+      "step": 5949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2644212447198776e-06,
+      "loss": 0.6952,
+      "step": 5950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261221976276764e-06,
+      "loss": 0.7609,
+      "step": 5951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258023333200335e-06,
+      "loss": 0.5936,
+      "step": 5952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2548253159127175e-06,
+      "loss": 0.6524,
+      "step": 5953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25162792483594e-06,
+      "loss": 0.6287,
+      "step": 5954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248431160391963e-06,
+      "loss": 0.5986,
+      "step": 5955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24523502300265e-06,
+      "loss": 0.6175,
+      "step": 5956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2420395130897936e-06,
+      "loss": 0.6144,
+      "step": 5957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2388446310750975e-06,
+      "loss": 0.734,
+      "step": 5958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23565037738018e-06,
+      "loss": 0.6996,
+      "step": 5959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2324567524265845e-06,
+      "loss": 0.5989,
+      "step": 5960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.229263756635763e-06,
+      "loss": 0.7742,
+      "step": 5961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226071390429093e-06,
+      "loss": 0.6858,
+      "step": 5962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222879654227857e-06,
+      "loss": 0.5946,
+      "step": 5963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21968854845327e-06,
+      "loss": 0.6562,
+      "step": 5964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216498073526454e-06,
+      "loss": 0.6254,
+      "step": 5965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213308229868447e-06,
+      "loss": 0.5867,
+      "step": 5966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210119017900206e-06,
+      "loss": 0.6563,
+      "step": 5967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206930438042608e-06,
+      "loss": 0.6827,
+      "step": 5968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203742490716439e-06,
+      "loss": 0.6509,
+      "step": 5969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200555176342404e-06,
+      "loss": 0.7533,
+      "step": 5970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197368495341133e-06,
+      "loss": 0.6343,
+      "step": 5971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194182448133163e-06,
+      "loss": 0.674,
+      "step": 5972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190997035138946e-06,
+      "loss": 0.6972,
+      "step": 5973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187812256778861e-06,
+      "loss": 0.7131,
+      "step": 5974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184628113473196e-06,
+      "loss": 0.6717,
+      "step": 5975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18144460564215e-06,
+      "loss": 0.6337,
+      "step": 5976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1782617337058495e-06,
+      "loss": 0.7452,
+      "step": 5977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175079498084328e-06,
+      "loss": 0.5724,
+      "step": 5978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171897899197539e-06,
+      "loss": 0.6669,
+      "step": 5979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168716937465352e-06,
+      "loss": 0.6721,
+      "step": 5980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165536613307554e-06,
+      "loss": 0.7052,
+      "step": 5981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162356927143842e-06,
+      "loss": 0.7229,
+      "step": 5982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159177879393832e-06,
+      "loss": 0.7912,
+      "step": 5983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155999470477062e-06,
+      "loss": 0.68,
+      "step": 5984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152821700812977e-06,
+      "loss": 0.6299,
+      "step": 5985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1496445708209396e-06,
+      "loss": 0.6571,
+      "step": 5986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14646808092023e-06,
+      "loss": 0.6889,
+      "step": 5987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143292231530043e-06,
+      "loss": 0.6737,
+      "step": 5988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140117023069485e-06,
+      "loss": 0.6196,
+      "step": 5989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13694245595759e-06,
+      "loss": 0.7009,
+      "step": 5990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133768530613293e-06,
+      "loss": 0.6426,
+      "step": 5991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130595247455452e-06,
+      "loss": 0.7361,
+      "step": 5992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127422606902834e-06,
+      "loss": 0.6358,
+      "step": 5993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124250609374135e-06,
+      "loss": 0.567,
+      "step": 5994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121079255287953e-06,
+      "loss": 0.6384,
+      "step": 5995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117908545062802e-06,
+      "loss": 0.5696,
+      "step": 5996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1147384791171166e-06,
+      "loss": 0.6508,
+      "step": 5997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111569057869243e-06,
+      "loss": 0.6459,
+      "step": 5998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108400281737444e-06,
+      "loss": 0.6863,
+      "step": 5999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105232151139895e-06,
+      "loss": 0.6178,
+      "step": 6000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1020646664946894e-06,
+      "loss": 0.6251,
+      "step": 6001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098897828219831e-06,
+      "loss": 0.712,
+      "step": 6002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095731636733241e-06,
+      "loss": 0.636,
+      "step": 6003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092566092452757e-06,
+      "loss": 0.7176,
+      "step": 6004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089401195796132e-06,
+      "loss": 0.6387,
+      "step": 6005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086236947181027e-06,
+      "loss": 0.6188,
+      "step": 6006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0830733470250185e-06,
+      "loss": 0.5899,
+      "step": 6007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079910395745607e-06,
+      "loss": 0.5391,
+      "step": 6008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076748093760195e-06,
+      "loss": 0.6561,
+      "step": 6009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07358644148611e-06,
+      "loss": 0.655,
+      "step": 6010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0704254393405846e-06,
+      "loss": 0.6284,
+      "step": 6011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0672650877407725e-06,
+      "loss": 0.6611,
+      "step": 6012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064105387103732e-06,
+      "loss": 0.6335,
+      "step": 6013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060946337846453e-06,
+      "loss": 0.5959,
+      "step": 6014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057787940385822e-06,
+      "loss": 0.6209,
+      "step": 6015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054630195138648e-06,
+      "loss": 0.5422,
+      "step": 6016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051473102521651e-06,
+      "loss": 0.6329,
+      "step": 6017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0483166629514654e-06,
+      "loss": 0.5829,
+      "step": 6018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045160876844641e-06,
+      "loss": 0.5591,
+      "step": 6019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042005744617642e-06,
+      "loss": 0.6291,
+      "step": 6020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038851266686841e-06,
+      "loss": 0.5855,
+      "step": 6021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035697443468528e-06,
+      "loss": 0.6751,
+      "step": 6022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032544275378911e-06,
+      "loss": 0.6968,
+      "step": 6023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029391762834104e-06,
+      "loss": 0.7181,
+      "step": 6024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026239906250139e-06,
+      "loss": 0.5975,
+      "step": 6025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023088706042954e-06,
+      "loss": 0.5637,
+      "step": 6026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019938162628415e-06,
+      "loss": 0.5924,
+      "step": 6027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0167882764222885e-06,
+      "loss": 0.5816,
+      "step": 6028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01363904784026e-06,
+      "loss": 0.5987,
+      "step": 6029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010490477297925e-06,
+      "loss": 0.6903,
+      "step": 6030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007342565210793e-06,
+      "loss": 0.6837,
+      "step": 6031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004195311994286e-06,
+      "loss": 0.5264,
+      "step": 6032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001048718063747e-06,
+      "loss": 0.6593,
+      "step": 6033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997902783834421e-06,
+      "loss": 0.6143,
+      "step": 6034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994757509721472e-06,
+      "loss": 0.5457,
+      "step": 6035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991612896139974e-06,
+      "loss": 0.6767,
+      "step": 6036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988468943504914e-06,
+      "loss": 0.6596,
+      "step": 6037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985325652231196e-06,
+      "loss": 0.5797,
+      "step": 6038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982183022733631e-06,
+      "loss": 0.6589,
+      "step": 6039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979041055426947e-06,
+      "loss": 0.5615,
+      "step": 6040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975899750725777e-06,
+      "loss": 0.6553,
+      "step": 6041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972759109044682e-06,
+      "loss": 0.5585,
+      "step": 6042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969619130798121e-06,
+      "loss": 0.6797,
+      "step": 6043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966479816400468e-06,
+      "loss": 0.65,
+      "step": 6044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963341166266018e-06,
+      "loss": 0.6442,
+      "step": 6045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.960203180808969e-06,
+      "loss": 0.6442,
+      "step": 6046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957065860443434e-06,
+      "loss": 0.5839,
+      "step": 6047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95392920558344e-06,
+      "loss": 0.5872,
+      "step": 6048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950793216642923e-06,
+      "loss": 0.6087,
+      "step": 6049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947657894035734e-06,
+      "loss": 0.639,
+      "step": 6050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944523238175634e-06,
+      "loss": 0.6842,
+      "step": 6051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9413892494762935e-06,
+      "loss": 0.6128,
+      "step": 6052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938255928351308e-06,
+      "loss": 0.5458,
+      "step": 6053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935123275214168e-06,
+      "loss": 0.5568,
+      "step": 6054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931991290478287e-06,
+      "loss": 0.5912,
+      "step": 6055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928859974556985e-06,
+      "loss": 0.6333,
+      "step": 6056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9257293278634945e-06,
+      "loss": 0.6481,
+      "step": 6057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922599350810962e-06,
+      "loss": 0.6082,
+      "step": 6058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919470043812441e-06,
+      "loss": 0.6905,
+      "step": 6059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916341407280899e-06,
+      "loss": 0.642,
+      "step": 6060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9132134416292224e-06,
+      "loss": 0.5107,
+      "step": 6061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910086147270195e-06,
+      "loss": 0.5544,
+      "step": 6062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906959524616525e-06,
+      "loss": 0.6251,
+      "step": 6063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903833574080825e-06,
+      "loss": 0.6528,
+      "step": 6064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900708296075619e-06,
+      "loss": 0.7724,
+      "step": 6065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897583691013343e-06,
+      "loss": 0.6574,
+      "step": 6066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894459759306346e-06,
+      "loss": 0.6544,
+      "step": 6067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891336501366886e-06,
+      "loss": 0.5896,
+      "step": 6068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888213917607132e-06,
+      "loss": 0.6544,
+      "step": 6069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885092008439166e-06,
+      "loss": 0.662,
+      "step": 6070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88197077427498e-06,
+      "loss": 0.6767,
+      "step": 6071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878850215526473e-06,
+      "loss": 0.5962,
+      "step": 6072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875730332605465e-06,
+      "loss": 0.6957,
+      "step": 6073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872611125923678e-06,
+      "loss": 0.6197,
+      "step": 6074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8694925958927465e-06,
+      "loss": 0.5895,
+      "step": 6075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866374742924217e-06,
+      "loss": 0.5887,
+      "step": 6076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863257567429544e-06,
+      "loss": 0.5656,
+      "step": 6077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860141069820096e-06,
+      "loss": 0.6465,
+      "step": 6078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857025250507153e-06,
+      "loss": 0.7217,
+      "step": 6079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853910109901901e-06,
+      "loss": 0.7011,
+      "step": 6080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8507956484154355e-06,
+      "loss": 0.6981,
+      "step": 6081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847681866458774e-06,
+      "loss": 0.5598,
+      "step": 6082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844568764442829e-06,
+      "loss": 0.717,
+      "step": 6083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841456342778432e-06,
+      "loss": 0.6564,
+      "step": 6084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838344601876324e-06,
+      "loss": 0.7128,
+      "step": 6085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835233542147153e-06,
+      "loss": 0.6108,
+      "step": 6086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83212316400148e-06,
+      "loss": 0.5802,
+      "step": 6087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8290134678497745e-06,
+      "loss": 0.6996,
+      "step": 6088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8259044541024165e-06,
+      "loss": 0.6668,
+      "step": 6089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822796123169698e-06,
+      "loss": 0.6375,
+      "step": 6090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819688475461814e-06,
+      "loss": 0.6684,
+      "step": 6091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8165815113888814e-06,
+      "loss": 0.6913,
+      "step": 6092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813475231360916e-06,
+      "loss": 0.6103,
+      "step": 6093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810369635787848e-06,
+      "loss": 0.5598,
+      "step": 6094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807264725079516e-06,
+      "loss": 0.6473,
+      "step": 6095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804160499645667e-06,
+      "loss": 0.6797,
+      "step": 6096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8010569598959575e-06,
+      "loss": 0.6806,
+      "step": 6097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797954106239963e-06,
+      "loss": 0.543,
+      "step": 6098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794851939087156e-06,
+      "loss": 0.6783,
+      "step": 6099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791750458846922e-06,
+      "loss": 0.6774,
+      "step": 6100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788649665928555e-06,
+      "loss": 0.6311,
+      "step": 6101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785549560741267e-06,
+      "loss": 0.6573,
+      "step": 6102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7824501436941685e-06,
+      "loss": 0.6324,
+      "step": 6103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779351415196283e-06,
+      "loss": 0.6228,
+      "step": 6104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776253375656543e-06,
+      "loss": 0.6715,
+      "step": 6105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773156025483792e-06,
+      "loss": 0.5766,
+      "step": 6106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770059365086779e-06,
+      "loss": 0.6128,
+      "step": 6107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7669633948741645e-06,
+      "loss": 0.5517,
+      "step": 6108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763868115254517e-06,
+      "loss": 0.6857,
+      "step": 6109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760773526636315e-06,
+      "loss": 0.6146,
+      "step": 6110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757679629427941e-06,
+      "loss": 0.5876,
+      "step": 6111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754586424037697e-06,
+      "loss": 0.67,
+      "step": 6112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751493910873783e-06,
+      "loss": 0.6347,
+      "step": 6113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748402090344312e-06,
+      "loss": 0.6268,
+      "step": 6114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745310962857301e-06,
+      "loss": 0.5618,
+      "step": 6115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742220528820689e-06,
+      "loss": 0.6083,
+      "step": 6116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739130788642308e-06,
+      "loss": 0.6305,
+      "step": 6117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7360417427299045e-06,
+      "loss": 0.705,
+      "step": 6118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732953391491135e-06,
+      "loss": 0.6859,
+      "step": 6119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729865735333563e-06,
+      "loss": 0.56,
+      "step": 6120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726778774664655e-06,
+      "loss": 0.5529,
+      "step": 6121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7236925098917995e-06,
+      "loss": 0.5737,
+      "step": 6122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.72060694142228e-06,
+      "loss": 0.6299,
+      "step": 6123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717522069663292e-06,
+      "loss": 0.6399,
+      "step": 6124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714437895021941e-06,
+      "loss": 0.5447,
+      "step": 6125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711354417905237e-06,
+      "loss": 0.6188,
+      "step": 6126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7082716387201035e-06,
+      "loss": 0.6874,
+      "step": 6127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705189557873364e-06,
+      "loss": 0.6446,
+      "step": 6128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702108175771759e-06,
+      "loss": 0.6647,
+      "step": 6129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699027492821925e-06,
+      "loss": 0.6362,
+      "step": 6130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695947509430422e-06,
+      "loss": 0.6564,
+      "step": 6131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692868226003705e-06,
+      "loss": 0.7247,
+      "step": 6132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689789642948141e-06,
+      "loss": 0.6466,
+      "step": 6133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6867117606699995e-06,
+      "loss": 0.6894,
+      "step": 6134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683634579575471e-06,
+      "loss": 0.6675,
+      "step": 6135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68055810007064e-06,
+      "loss": 0.5422,
+      "step": 6136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677482322561504e-06,
+      "loss": 0.6466,
+      "step": 6137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674407247453966e-06,
+      "loss": 0.6426,
+      "step": 6138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6713328751538385e-06,
+      "loss": 0.6588,
+      "step": 6139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6682592060668354e-06,
+      "loss": 0.5716,
+      "step": 6140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66518624059859e-06,
+      "loss": 0.6596,
+      "step": 6141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662113979154631e-06,
+      "loss": 0.7048,
+      "step": 6142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659042422140399e-06,
+      "loss": 0.6407,
+      "step": 6143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655971569961241e-06,
+      "loss": 0.6908,
+      "step": 6144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6529014230224105e-06,
+      "loss": 0.6406,
+      "step": 6145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649831981729069e-06,
+      "loss": 0.657,
+      "step": 6146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646763246486283e-06,
+      "loss": 0.5782,
+      "step": 6147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64369521769903e-06,
+      "loss": 0.5882,
+      "step": 6148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640627895772185e-06,
+      "loss": 0.6355,
+      "step": 6149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6375612811105445e-06,
+      "loss": 0.5977,
+      "step": 6150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634495374118798e-06,
+      "loss": 0.7573,
+      "step": 6151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631430175201546e-06,
+      "loss": 0.586,
+      "step": 6152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628365684763302e-06,
+      "loss": 0.6618,
+      "step": 6153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625301903208477e-06,
+      "loss": 0.639,
+      "step": 6154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6222388309413924e-06,
+      "loss": 0.7214,
+      "step": 6155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619176468366274e-06,
+      "loss": 0.7012,
+      "step": 6156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6161148158872575e-06,
+      "loss": 0.6661,
+      "step": 6157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613053873908382e-06,
+      "loss": 0.6605,
+      "step": 6158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609993642833593e-06,
+      "loss": 0.6117,
+      "step": 6159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606934123066739e-06,
+      "loss": 0.6092,
+      "step": 6160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603875315011587e-06,
+      "loss": 0.5934,
+      "step": 6161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600817219071798e-06,
+      "loss": 0.6635,
+      "step": 6162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597759835650941e-06,
+      "loss": 0.584,
+      "step": 6163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594703165152493e-06,
+      "loss": 0.6963,
+      "step": 6164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591647207979838e-06,
+      "loss": 0.6643,
+      "step": 6165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588591964536263e-06,
+      "loss": 0.6202,
+      "step": 6166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5855374352249626e-06,
+      "loss": 0.6496,
+      "step": 6167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582483620449032e-06,
+      "loss": 0.6222,
+      "step": 6168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579430520611485e-06,
+      "loss": 0.6466,
+      "step": 6169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576378136115229e-06,
+      "loss": 0.6777,
+      "step": 6170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573326467363078e-06,
+      "loss": 0.6701,
+      "step": 6171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570275514757759e-06,
+      "loss": 0.7141,
+      "step": 6172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5672252787019e-06,
+      "loss": 0.6916,
+      "step": 6173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564175759598032e-06,
+      "loss": 0.5963,
+      "step": 6174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561126957848594e-06,
+      "loss": 0.5124,
+      "step": 6175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558078873855931e-06,
+      "loss": 0.6414,
+      "step": 6176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5550315080222905e-06,
+      "loss": 0.6258,
+      "step": 6177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551984860749829e-06,
+      "loss": 0.6408,
+      "step": 6178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5489389324406045e-06,
+      "loss": 0.7216,
+      "step": 6179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545893723496578e-06,
+      "loss": 0.6111,
+      "step": 6180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542849234319628e-06,
+      "loss": 0.6472,
+      "step": 6181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539805465311527e-06,
+      "loss": 0.701,
+      "step": 6182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536762416873952e-06,
+      "loss": 0.5775,
+      "step": 6183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53372008940849e-06,
+      "loss": 0.5579,
+      "step": 6184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530678483316628e-06,
+      "loss": 0.5503,
+      "step": 6185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5276375989997605e-06,
+      "loss": 0.6474,
+      "step": 6186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524597436859191e-06,
+      "loss": 0.5824,
+      "step": 6187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52155799729612e-06,
+      "loss": 0.5837,
+      "step": 6188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518519280711653e-06,
+      "loss": 0.5391,
+      "step": 6189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515481287506811e-06,
+      "loss": 0.646,
+      "step": 6190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512444018082508e-06,
+      "loss": 0.7079,
+      "step": 6191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509407472839565e-06,
+      "loss": 0.6653,
+      "step": 6192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506371652178709e-06,
+      "loss": 0.6948,
+      "step": 6193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503336556500572e-06,
+      "loss": 0.6981,
+      "step": 6194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5003021862056876e-06,
+      "loss": 0.6636,
+      "step": 6195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497268541694496e-06,
+      "loss": 0.6291,
+      "step": 6196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4942356233673415e-06,
+      "loss": 0.6159,
+      "step": 6197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491203431624471e-06,
+      "loss": 0.6312,
+      "step": 6198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488171966866034e-06,
+      "loss": 0.6466,
+      "step": 6199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485141229492094e-06,
+      "loss": 0.6166,
+      "step": 6200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482111219902606e-06,
+      "loss": 0.7335,
+      "step": 6201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479081938497435e-06,
+      "loss": 0.6102,
+      "step": 6202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476053385676349e-06,
+      "loss": 0.5158,
+      "step": 6203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473025561839019e-06,
+      "loss": 0.6532,
+      "step": 6204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46999846738502e-06,
+      "loss": 0.6414,
+      "step": 6205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466972102713835e-06,
+      "loss": 0.6761,
+      "step": 6206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463946468224844e-06,
+      "loss": 0.5785,
+      "step": 6207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460921564317336e-06,
+      "loss": 0.7005,
+      "step": 6208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4578973913904946e-06,
+      "loss": 0.6555,
+      "step": 6209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4548739498434235e-06,
+      "loss": 0.6598,
+      "step": 6210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451851240075115e-06,
+      "loss": 0.6445,
+      "step": 6211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44882926248447e-06,
+      "loss": 0.6704,
+      "step": 6212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445808017470293e-06,
+      "loss": 0.6907,
+      "step": 6213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44278750543129e-06,
+      "loss": 0.6478,
+      "step": 6214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439767726766073e-06,
+      "loss": 0.6093,
+      "step": 6215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436748681873157e-06,
+      "loss": 0.7159,
+      "step": 6216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433730371150957e-06,
+      "loss": 0.565,
+      "step": 6217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430712794997795e-06,
+      "loss": 0.58,
+      "step": 6218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427695953811889e-06,
+      "loss": 0.5534,
+      "step": 6219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424679847991373e-06,
+      "loss": 0.5698,
+      "step": 6220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421664477934272e-06,
+      "loss": 0.6349,
+      "step": 6221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418649844038521e-06,
+      "loss": 0.5985,
+      "step": 6222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4156359467019475e-06,
+      "loss": 0.6142,
+      "step": 6223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412622786322298e-06,
+      "loss": 0.6918,
+      "step": 6224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409610363297211e-06,
+      "loss": 0.5613,
+      "step": 6225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406598678024228e-06,
+      "loss": 0.6038,
+      "step": 6226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403587730900795e-06,
+      "loss": 0.6126,
+      "step": 6227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40057752232426e-06,
+      "loss": 0.778,
+      "step": 6228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3975680526918715e-06,
+      "loss": 0.6202,
+      "step": 6229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39455932240079e-06,
+      "loss": 0.6102,
+      "step": 6230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3915513318480665e-06,
+      "loss": 0.6668,
+      "step": 6231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388544081430661e-06,
+      "loss": 0.6998,
+      "step": 6232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385537571545433e-06,
+      "loss": 0.6777,
+      "step": 6233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382531802589146e-06,
+      "loss": 0.6212,
+      "step": 6234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3795267749584645e-06,
+      "loss": 0.5517,
+      "step": 6235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3765224890499556e-06,
+      "loss": 0.6371,
+      "step": 6236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37351894526009e-06,
+      "loss": 0.6894,
+      "step": 6237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370516143985237e-06,
+      "loss": 0.6183,
+      "step": 6238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367514085621668e-06,
+      "loss": 0.5122,
+      "step": 6239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3645127705655654e-06,
+      "loss": 0.5957,
+      "step": 6240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361512199213004e-06,
+      "loss": 0.6453,
+      "step": 6241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358512371959957e-06,
+      "loss": 0.6121,
+      "step": 6242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355513289202313e-06,
+      "loss": 0.6482,
+      "step": 6243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352514951335852e-06,
+      "loss": 0.6442,
+      "step": 6244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349517358756259e-06,
+      "loss": 0.6598,
+      "step": 6245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3465205118591195e-06,
+      "loss": 0.6036,
+      "step": 6246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3435244110399195e-06,
+      "loss": 0.6641,
+      "step": 6247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3405290566940475e-06,
+      "loss": 0.5881,
+      "step": 6248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337534449216799e-06,
+      "loss": 0.622,
+      "step": 6249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334540589003364e-06,
+      "loss": 0.5976,
+      "step": 6250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331547476448835e-06,
+      "loss": 0.5991,
+      "step": 6251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328555111948208e-06,
+      "loss": 0.535,
+      "step": 6252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325563495896378e-06,
+      "loss": 0.6779,
+      "step": 6253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322572628688143e-06,
+      "loss": 0.6192,
+      "step": 6254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319582510718201e-06,
+      "loss": 0.6374,
+      "step": 6255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316593142381152e-06,
+      "loss": 0.623,
+      "step": 6256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313604524071494e-06,
+      "loss": 0.673,
+      "step": 6257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3106166561836335e-06,
+      "loss": 0.7117,
+      "step": 6258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3076295391118725e-06,
+      "loss": 0.5959,
+      "step": 6259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304643173250409e-06,
+      "loss": 0.6413,
+      "step": 6260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3016575589933564e-06,
+      "loss": 0.6844,
+      "step": 6261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298672696734714e-06,
+      "loss": 0.662,
+      "step": 6262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295688586868389e-06,
+      "loss": 0.6051,
+      "step": 6263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29270522978819e-06,
+      "loss": 0.5924,
+      "step": 6264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2897226258878235e-06,
+      "loss": 0.5906,
+      "step": 6265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286740775560894e-06,
+      "loss": 0.5727,
+      "step": 6266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2837596792009166e-06,
+      "loss": 0.6214,
+      "step": 6267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280779337201291e-06,
+      "loss": 0.6329,
+      "step": 6268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277799749955336e-06,
+      "loss": 0.7255,
+      "step": 6269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274820917856259e-06,
+      "loss": 0.6549,
+      "step": 6270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271842841297169e-06,
+      "loss": 0.6589,
+      "step": 6271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268865520671077e-06,
+      "loss": 0.6221,
+      "step": 6272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265888956370894e-06,
+      "loss": 0.6694,
+      "step": 6273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262913148789431e-06,
+      "loss": 0.6097,
+      "step": 6274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259938098319398e-06,
+      "loss": 0.7081,
+      "step": 6275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256963805353403e-06,
+      "loss": 0.6396,
+      "step": 6276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253990270283968e-06,
+      "loss": 0.6826,
+      "step": 6277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2510174935034945e-06,
+      "loss": 0.679,
+      "step": 6278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2480454754042945e-06,
+      "loss": 0.6598,
+      "step": 6279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2450742163785855e-06,
+      "loss": 0.643,
+      "step": 6280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2421037168184735e-06,
+      "loss": 0.7465,
+      "step": 6281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23913397711597e-06,
+      "loss": 0.6946,
+      "step": 6282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2361649976629856e-06,
+      "loss": 0.7644,
+      "step": 6283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2331967788513295e-06,
+      "loss": 0.6408,
+      "step": 6284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230229321072713e-06,
+      "loss": 0.7001,
+      "step": 6285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227262624718743e-06,
+      "loss": 0.5906,
+      "step": 6286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2242966901809305e-06,
+      "loss": 0.5687,
+      "step": 6287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221331517850679e-06,
+      "loss": 0.6824,
+      "step": 6288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2183671081193036e-06,
+      "loss": 0.6231,
+      "step": 6289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215403461378007e-06,
+      "loss": 0.5854,
+      "step": 6290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212440578017895e-06,
+      "loss": 0.6519,
+      "step": 6291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2094784584299765e-06,
+      "loss": 0.6426,
+      "step": 6292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206517103005152e-06,
+      "loss": 0.5942,
+      "step": 6293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203556512134225e-06,
+      "loss": 0.6976,
+      "step": 6294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200596686207905e-06,
+      "loss": 0.6452,
+      "step": 6295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197637625616788e-06,
+      "loss": 0.6767,
+      "step": 6296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194679330751379e-06,
+      "loss": 0.6799,
+      "step": 6297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191721802002072e-06,
+      "loss": 0.5665,
+      "step": 6298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188765039759174e-06,
+      "loss": 0.5886,
+      "step": 6299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185809044412878e-06,
+      "loss": 0.6181,
+      "step": 6300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182853816353282e-06,
+      "loss": 0.6369,
+      "step": 6301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17989935597038e-06,
+      "loss": 0.7225,
+      "step": 6302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176945663654067e-06,
+      "loss": 0.6394,
+      "step": 6303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173992739794135e-06,
+      "loss": 0.6258,
+      "step": 6304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171040584780276e-06,
+      "loss": 0.7389,
+      "step": 6305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.168089199002079e-06,
+      "loss": 0.6494,
+      "step": 6306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16513858284903e-06,
+      "loss": 0.6815,
+      "step": 6307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16218873671052e-06,
+      "loss": 0.5612,
+      "step": 6308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159239660975834e-06,
+      "loss": 0.539,
+      "step": 6309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156291356034152e-06,
+      "loss": 0.5671,
+      "step": 6310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1533438222745594e-06,
+      "loss": 0.62,
+      "step": 6311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150397060086032e-06,
+      "loss": 0.6682,
+      "step": 6312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147451069857448e-06,
+      "loss": 0.6144,
+      "step": 6313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144505851977588e-06,
+      "loss": 0.5703,
+      "step": 6314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141561406835125e-06,
+      "loss": 0.6433,
+      "step": 6315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138617734818629e-06,
+      "loss": 0.6558,
+      "step": 6316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135674836316569e-06,
+      "loss": 0.7277,
+      "step": 6317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132732711717319e-06,
+      "loss": 0.5443,
+      "step": 6318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129791361409141e-06,
+      "loss": 0.5787,
+      "step": 6319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126850785780199e-06,
+      "loss": 0.6503,
+      "step": 6320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1239109852185565e-06,
+      "loss": 0.6159,
+      "step": 6321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1209719601121725e-06,
+      "loss": 0.6639,
+      "step": 6322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118033710848901e-06,
+      "loss": 0.5994,
+      "step": 6323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1150962378165e-06,
+      "loss": 0.6247,
+      "step": 6324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112159541402621e-06,
+      "loss": 0.6322,
+      "step": 6325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109223621994813e-06,
+      "loss": 0.6212,
+      "step": 6326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106288479980518e-06,
+      "loss": 0.5843,
+      "step": 6327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103354115747092e-06,
+      "loss": 0.6055,
+      "step": 6328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100420529681769e-06,
+      "loss": 0.6426,
+      "step": 6329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097487722171691e-06,
+      "loss": 0.5424,
+      "step": 6330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094555693603891e-06,
+      "loss": 0.6952,
+      "step": 6331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0916244443653084e-06,
+      "loss": 0.6235,
+      "step": 6332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088693974842771e-06,
+      "loss": 0.7208,
+      "step": 6333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085764285423007e-06,
+      "loss": 0.5893,
+      "step": 6334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082835376492642e-06,
+      "loss": 0.6176,
+      "step": 6335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079907248438196e-06,
+      "loss": 0.6619,
+      "step": 6336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076979901646088e-06,
+      "loss": 0.5823,
+      "step": 6337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074053336502637e-06,
+      "loss": 0.5922,
+      "step": 6338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071127553394054e-06,
+      "loss": 0.6678,
+      "step": 6339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0682025527064486e-06,
+      "loss": 0.5866,
+      "step": 6340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065278334825828e-06,
+      "loss": 0.6523,
+      "step": 6341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062354900138092e-06,
+      "loss": 0.639,
+      "step": 6342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059432249029044e-06,
+      "loss": 0.5348,
+      "step": 6343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056510381884377e-06,
+      "loss": 0.5738,
+      "step": 6344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053589299089686e-06,
+      "loss": 0.6608,
+      "step": 6345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05066900103046e-06,
+      "loss": 0.6823,
+      "step": 6346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0477494880920794e-06,
+      "loss": 0.681,
+      "step": 6347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044830760659836e-06,
+      "loss": 0.6388,
+      "step": 6348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041912819118902e-06,
+      "loss": 0.5876,
+      "step": 6349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0389956638543494e-06,
+      "loss": 0.5957,
+      "step": 6350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036079295251157e-06,
+      "loss": 0.6284,
+      "step": 6351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033163713694188e-06,
+      "loss": 0.6041,
+      "step": 6352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030248919568206e-06,
+      "loss": 0.6376,
+      "step": 6353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0273349132578685e-06,
+      "loss": 0.5909,
+      "step": 6354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024421695147733e-06,
+      "loss": 0.5861,
+      "step": 6355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021509265622247e-06,
+      "loss": 0.6623,
+      "step": 6356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018597625065763e-06,
+      "loss": 0.6012,
+      "step": 6357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015686773862522e-06,
+      "loss": 0.5385,
+      "step": 6358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012776712396664e-06,
+      "loss": 0.6385,
+      "step": 6359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009867441052222e-06,
+      "loss": 0.5946,
+      "step": 6360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006958960213127e-06,
+      "loss": 0.663,
+      "step": 6361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.004051270263206e-06,
+      "loss": 0.6066,
+      "step": 6362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0011443715861805e-06,
+      "loss": 0.6683,
+      "step": 6363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998238264565667e-06,
+      "loss": 0.6824,
+      "step": 6364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995332949585179e-06,
+      "loss": 0.6366,
+      "step": 6365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9924284270281215e-06,
+      "loss": 0.7222,
+      "step": 6366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9895246972778046e-06,
+      "loss": 0.5769,
+      "step": 6367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986621760717421e-06,
+      "loss": 0.6661,
+      "step": 6368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983719617730072e-06,
+      "loss": 0.6143,
+      "step": 6369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980818268698745e-06,
+      "loss": 0.6146,
+      "step": 6370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977917714006324e-06,
+      "loss": 0.6327,
+      "step": 6371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975017954035588e-06,
+      "loss": 0.6562,
+      "step": 6372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972118989169215e-06,
+      "loss": 0.6123,
+      "step": 6373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969220819789773e-06,
+      "loss": 0.6902,
+      "step": 6374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966323446279729e-06,
+      "loss": 0.6115,
+      "step": 6375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963426869021439e-06,
+      "loss": 0.6666,
+      "step": 6376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960531088397166e-06,
+      "loss": 0.6392,
+      "step": 6377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957636104789056e-06,
+      "loss": 0.5743,
+      "step": 6378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954741918579154e-06,
+      "loss": 0.6693,
+      "step": 6379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9518485301493985e-06,
+      "loss": 0.6777,
+      "step": 6380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.948955939881627e-06,
+      "loss": 0.6272,
+      "step": 6381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946064148157565e-06,
+      "loss": 0.609,
+      "step": 6382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94317315535884e-06,
+      "loss": 0.6007,
+      "step": 6383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940282961866965e-06,
+      "loss": 0.6584,
+      "step": 6384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937393568063359e-06,
+      "loss": 0.729,
+      "step": 6385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934504974329326e-06,
+      "loss": 0.6723,
+      "step": 6386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931617181046067e-06,
+      "loss": 0.5796,
+      "step": 6387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928730188594681e-06,
+      "loss": 0.6177,
+      "step": 6388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925843997356157e-06,
+      "loss": 0.621,
+      "step": 6389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9229586077113805e-06,
+      "loss": 0.6043,
+      "step": 6390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920074020041128e-06,
+      "loss": 0.6248,
+      "step": 6391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917190234726075e-06,
+      "loss": 0.5882,
+      "step": 6392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914307252146787e-06,
+      "loss": 0.5631,
+      "step": 6393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911425072683726e-06,
+      "loss": 0.579,
+      "step": 6394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.908543696717246e-06,
+      "loss": 0.6938,
+      "step": 6395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905663124627596e-06,
+      "loss": 0.6378,
+      "step": 6396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902783356794921e-06,
+      "loss": 0.5948,
+      "step": 6397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89990439359926e-06,
+      "loss": 0.5092,
+      "step": 6398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897026235420539e-06,
+      "loss": 0.6028,
+      "step": 6399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8941488826385855e-06,
+      "loss": 0.6197,
+      "step": 6400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891272335633116e-06,
+      "loss": 0.6087,
+      "step": 6401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888396594783741e-06,
+      "loss": 0.6198,
+      "step": 6402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885521660469972e-06,
+      "loss": 0.5611,
+      "step": 6403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8826475330712035e-06,
+      "loss": 0.6231,
+      "step": 6404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8797742129667294e-06,
+      "loss": 0.6665,
+      "step": 6405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.876901700535732e-06,
+      "loss": 0.6567,
+      "step": 6406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874029996157298e-06,
+      "loss": 0.645,
+      "step": 6407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871159100210397e-06,
+      "loss": 0.6471,
+      "step": 6408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868289013073892e-06,
+      "loss": 0.6351,
+      "step": 6409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865419735126549e-06,
+      "loss": 0.5772,
+      "step": 6410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8625512667470146e-06,
+      "loss": 0.5886,
+      "step": 6411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859683608313837e-06,
+      "loss": 0.5865,
+      "step": 6412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856816760205455e-06,
+      "loss": 0.6037,
+      "step": 6413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8539507228002e-06,
+      "loss": 0.5842,
+      "step": 6414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851085496476295e-06,
+      "loss": 0.5983,
+      "step": 6415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848221081611864e-06,
+      "loss": 0.702,
+      "step": 6416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845357478584915e-06,
+      "loss": 0.6156,
+      "step": 6417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8424946877733495e-06,
+      "loss": 0.5782,
+      "step": 6418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839632709554965e-06,
+      "loss": 0.6395,
+      "step": 6419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836771544307454e-06,
+      "loss": 0.566,
+      "step": 6420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833911192408392e-06,
+      "loss": 0.7266,
+      "step": 6421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831051654235259e-06,
+      "loss": 0.6463,
+      "step": 6422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8281929301654234e-06,
+      "loss": 0.6745,
+      "step": 6423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825335020576142e-06,
+      "loss": 0.5921,
+      "step": 6424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822477925844564e-06,
+      "loss": 0.6297,
+      "step": 6425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819621646347742e-06,
+      "loss": 0.7067,
+      "step": 6426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816766182462609e-06,
+      "loss": 0.63,
+      "step": 6427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8139115345659948e-06,
+      "loss": 0.7333,
+      "step": 6428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811057703034622e-06,
+      "loss": 0.7191,
+      "step": 6429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808204688245105e-06,
+      "loss": 0.5696,
+      "step": 6430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8053524905739494e-06,
+      "loss": 0.6258,
+      "step": 6431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802501110397553e-06,
+      "loss": 0.5751,
+      "step": 6432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7996505480922084e-06,
+      "loss": 0.6504,
+      "step": 6433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7968008040340975e-06,
+      "loss": 0.6824,
+      "step": 6434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7939518785992925e-06,
+      "loss": 0.6922,
+      "step": 6435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7911037721637646e-06,
+      "loss": 0.6702,
+      "step": 6436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788256485103372e-06,
+      "loss": 0.6373,
+      "step": 6437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7854100177938634e-06,
+      "loss": 0.6638,
+      "step": 6438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782564370610878e-06,
+      "loss": 0.6541,
+      "step": 6439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779719543929958e-06,
+      "loss": 0.6306,
+      "step": 6440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7768755381265245e-06,
+      "loss": 0.6525,
+      "step": 6441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774032353575896e-06,
+      "loss": 0.5282,
+      "step": 6442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.77118999065328e-06,
+      "loss": 0.6423,
+      "step": 6443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76834844973378e-06,
+      "loss": 0.6933,
+      "step": 6444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7655077311923815e-06,
+      "loss": 0.5824,
+      "step": 6445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762667835403977e-06,
+      "loss": 0.6862,
+      "step": 6446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7598287627433395e-06,
+      "loss": 0.5476,
+      "step": 6447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7569905135851325e-06,
+      "loss": 0.7252,
+      "step": 6448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7541530883039155e-06,
+      "loss": 0.6064,
+      "step": 6449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751316487274138e-06,
+      "loss": 0.6278,
+      "step": 6450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7484807108701403e-06,
+      "loss": 0.5995,
+      "step": 6451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745645759466152e-06,
+      "loss": 0.6672,
+      "step": 6452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7428116334362983e-06,
+      "loss": 0.624,
+      "step": 6453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7399783331545914e-06,
+      "loss": 0.6804,
+      "step": 6454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7371458589949337e-06,
+      "loss": 0.6935,
+      "step": 6455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.734314211331126e-06,
+      "loss": 0.6358,
+      "step": 6456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731483390536854e-06,
+      "loss": 0.6654,
+      "step": 6457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7286533969856896e-06,
+      "loss": 0.6032,
+      "step": 6458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725824231051108e-06,
+      "loss": 0.5672,
+      "step": 6459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7229958931064682e-06,
+      "loss": 0.7025,
+      "step": 6460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7201683835250157e-06,
+      "loss": 0.6028,
+      "step": 6461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.717341702679894e-06,
+      "loss": 0.6079,
+      "step": 6462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7145158509441324e-06,
+      "loss": 0.5793,
+      "step": 6463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711690828690655e-06,
+      "loss": 0.665,
+      "step": 6464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7088666362922677e-06,
+      "loss": 0.7133,
+      "step": 6465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7060432741216833e-06,
+      "loss": 0.601,
+      "step": 6466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7032207425514888e-06,
+      "loss": 0.6767,
+      "step": 6467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7003990419541703e-06,
+      "loss": 0.5649,
+      "step": 6468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6975781727020997e-06,
+      "loss": 0.6257,
+      "step": 6469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6947581351675423e-06,
+      "loss": 0.6582,
+      "step": 6470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6919389297226528e-06,
+      "loss": 0.6768,
+      "step": 6471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.689120556739475e-06,
+      "loss": 0.6485,
+      "step": 6472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.686303016589945e-06,
+      "loss": 0.6165,
+      "step": 6473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6834863096458827e-06,
+      "loss": 0.6276,
+      "step": 6474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6806704362790115e-06,
+      "loss": 0.6401,
+      "step": 6475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677855396860929e-06,
+      "loss": 0.6724,
+      "step": 6476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6750411917631356e-06,
+      "loss": 0.659,
+      "step": 6477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.672227821357014e-06,
+      "loss": 0.5953,
+      "step": 6478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669415286013839e-06,
+      "loss": 0.6358,
+      "step": 6479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6666035861047744e-06,
+      "loss": 0.6131,
+      "step": 6480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663792722000875e-06,
+      "loss": 0.6669,
+      "step": 6481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660982694073084e-06,
+      "loss": 0.6236,
+      "step": 6482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658173502692236e-06,
+      "loss": 0.6254,
+      "step": 6483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65536514822905e-06,
+      "loss": 0.4606,
+      "step": 6484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6525576310541446e-06,
+      "loss": 0.5673,
+      "step": 6485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6497509515380203e-06,
+      "loss": 0.6452,
+      "step": 6486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6469451100510677e-06,
+      "loss": 0.672,
+      "step": 6487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6441401069635686e-06,
+      "loss": 0.5851,
+      "step": 6488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6413359426456927e-06,
+      "loss": 0.6656,
+      "step": 6489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638532617467501e-06,
+      "loss": 0.6558,
+      "step": 6490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63573013179894e-06,
+      "loss": 0.6488,
+      "step": 6491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.632928486009847e-06,
+      "loss": 0.5937,
+      "step": 6492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630127680469955e-06,
+      "loss": 0.6505,
+      "step": 6493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627327715548877e-06,
+      "loss": 0.5878,
+      "step": 6494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6245285916161168e-06,
+      "loss": 0.6,
+      "step": 6495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6217303090410728e-06,
+      "loss": 0.5754,
+      "step": 6496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6189328681930282e-06,
+      "loss": 0.5958,
+      "step": 6497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6161362694411527e-06,
+      "loss": 0.7262,
+      "step": 6498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6133405131545108e-06,
+      "loss": 0.6209,
+      "step": 6499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6105455997020488e-06,
+      "loss": 0.7029,
+      "step": 6500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607751529452609e-06,
+      "loss": 0.5984,
+      "step": 6501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.604958302774918e-06,
+      "loss": 0.6576,
+      "step": 6502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60216592003759e-06,
+      "loss": 0.6531,
+      "step": 6503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5993743816091297e-06,
+      "loss": 0.6248,
+      "step": 6504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.596583687857936e-06,
+      "loss": 0.5985,
+      "step": 6505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5937938391522876e-06,
+      "loss": 0.548,
+      "step": 6506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5910048358603554e-06,
+      "loss": 0.6361,
+      "step": 6507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588216678350198e-06,
+      "loss": 0.61,
+      "step": 6508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5854293669897623e-06,
+      "loss": 0.6619,
+      "step": 6509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5826429021468846e-06,
+      "loss": 0.6279,
+      "step": 6510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5798572841892855e-06,
+      "loss": 0.6005,
+      "step": 6511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.577072513484584e-06,
+      "loss": 0.5958,
+      "step": 6512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5742885904002766e-06,
+      "loss": 0.6222,
+      "step": 6513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.571505515303749e-06,
+      "loss": 0.651,
+      "step": 6514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5687232885622846e-06,
+      "loss": 0.7064,
+      "step": 6515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565941910543045e-06,
+      "loss": 0.6193,
+      "step": 6516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5631613816130816e-06,
+      "loss": 0.7026,
+      "step": 6517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560381702139337e-06,
+      "loss": 0.7181,
+      "step": 6518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.557602872488638e-06,
+      "loss": 0.659,
+      "step": 6519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5548248930277008e-06,
+      "loss": 0.6485,
+      "step": 6520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.552047764123131e-06,
+      "loss": 0.6325,
+      "step": 6521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54927148614142e-06,
+      "loss": 0.5611,
+      "step": 6522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5464960594489474e-06,
+      "loss": 0.6397,
+      "step": 6523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543721484411976e-06,
+      "loss": 0.601,
+      "step": 6524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540947761396669e-06,
+      "loss": 0.6487,
+      "step": 6525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538174890769066e-06,
+      "loss": 0.573,
+      "step": 6526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535402872895095e-06,
+      "loss": 0.6389,
+      "step": 6527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532631708140575e-06,
+      "loss": 0.6427,
+      "step": 6528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529861396871206e-06,
+      "loss": 0.7047,
+      "step": 6529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527091939452588e-06,
+      "loss": 0.6356,
+      "step": 6530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5243233362501973e-06,
+      "loss": 0.6744,
+      "step": 6531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5215555876293995e-06,
+      "loss": 0.6703,
+      "step": 6532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5187886939554473e-06,
+      "loss": 0.5034,
+      "step": 6533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.516022655593487e-06,
+      "loss": 0.6584,
+      "step": 6534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5132574729085443e-06,
+      "loss": 0.6814,
+      "step": 6535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5104931462655347e-06,
+      "loss": 0.6701,
+      "step": 6536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50772967602926e-06,
+      "loss": 0.6574,
+      "step": 6537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5049670625644095e-06,
+      "loss": 0.7012,
+      "step": 6538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502205306235561e-06,
+      "loss": 0.7337,
+      "step": 6539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4994444074071752e-06,
+      "loss": 0.588,
+      "step": 6540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4966843664436057e-06,
+      "loss": 0.6007,
+      "step": 6541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493925183709086e-06,
+      "loss": 0.59,
+      "step": 6542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4911668595677385e-06,
+      "loss": 0.5667,
+      "step": 6543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4884093943835796e-06,
+      "loss": 0.7039,
+      "step": 6544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4856527885205015e-06,
+      "loss": 0.6659,
+      "step": 6545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4828970423422905e-06,
+      "loss": 0.6486,
+      "step": 6546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4801421562126146e-06,
+      "loss": 0.6476,
+      "step": 6547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4773881304950284e-06,
+      "loss": 0.6878,
+      "step": 6548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4746349655529798e-06,
+      "loss": 0.5398,
+      "step": 6549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471882661749797e-06,
+      "loss": 0.7267,
+      "step": 6550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4691312194486948e-06,
+      "loss": 0.6935,
+      "step": 6551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4663806390127742e-06,
+      "loss": 0.5295,
+      "step": 6552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4636309208050223e-06,
+      "loss": 0.6087,
+      "step": 6553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4608820651883202e-06,
+      "loss": 0.6488,
+      "step": 6554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458134072525423e-06,
+      "loss": 0.5654,
+      "step": 6555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45538694317898e-06,
+      "loss": 0.5926,
+      "step": 6556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452640677511524e-06,
+      "loss": 0.6509,
+      "step": 6557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449895275885473e-06,
+      "loss": 0.585,
+      "step": 6558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447150738663132e-06,
+      "loss": 0.6254,
+      "step": 6559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444407066206692e-06,
+      "loss": 0.6404,
+      "step": 6560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.44166425887823e-06,
+      "loss": 0.6052,
+      "step": 6561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.438922317039709e-06,
+      "loss": 0.6056,
+      "step": 6562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.436181241052973e-06,
+      "loss": 0.6134,
+      "step": 6563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4334410312797626e-06,
+      "loss": 0.6537,
+      "step": 6564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430701688081697e-06,
+      "loss": 0.6896,
+      "step": 6565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4279632118202744e-06,
+      "loss": 0.695,
+      "step": 6566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.425225602856895e-06,
+      "loss": 0.7276,
+      "step": 6567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4224888615528317e-06,
+      "loss": 0.622,
+      "step": 6568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.419752988269246e-06,
+      "loss": 0.6145,
+      "step": 6569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4170179833671847e-06,
+      "loss": 0.6661,
+      "step": 6570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4142838472075832e-06,
+      "loss": 0.5731,
+      "step": 6571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411550580151257e-06,
+      "loss": 0.7034,
+      "step": 6572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4088181825589083e-06,
+      "loss": 0.5699,
+      "step": 6573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.406086654791132e-06,
+      "loss": 0.5839,
+      "step": 6574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4033559972083983e-06,
+      "loss": 0.5463,
+      "step": 6575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4006262101710665e-06,
+      "loss": 0.6064,
+      "step": 6576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3978972940393818e-06,
+      "loss": 0.5995,
+      "step": 6577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395169249173472e-06,
+      "loss": 0.6582,
+      "step": 6578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.392442075933353e-06,
+      "loss": 0.5844,
+      "step": 6579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3897157746789246e-06,
+      "loss": 0.6038,
+      "step": 6580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3869903457699694e-06,
+      "loss": 0.6764,
+      "step": 6581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3842657895661545e-06,
+      "loss": 0.6644,
+      "step": 6582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3815421064270403e-06,
+      "loss": 0.7308,
+      "step": 6583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.378819296712058e-06,
+      "loss": 0.5243,
+      "step": 6584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376097360780539e-06,
+      "loss": 0.6901,
+      "step": 6585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3733762989916875e-06,
+      "loss": 0.6087,
+      "step": 6586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3706561117045966e-06,
+      "loss": 0.6842,
+      "step": 6587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3679367992782454e-06,
+      "loss": 0.6273,
+      "step": 6588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.365218362071493e-06,
+      "loss": 0.6054,
+      "step": 6589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362500800443087e-06,
+      "loss": 0.5889,
+      "step": 6590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3597841147516585e-06,
+      "loss": 0.7594,
+      "step": 6591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3570683053557205e-06,
+      "loss": 0.5856,
+      "step": 6592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3543533726136778e-06,
+      "loss": 0.6371,
+      "step": 6593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3516393168838125e-06,
+      "loss": 0.6834,
+      "step": 6594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348926138524292e-06,
+      "loss": 0.6957,
+      "step": 6595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3462138378931687e-06,
+      "loss": 0.6609,
+      "step": 6596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.34350241534838e-06,
+      "loss": 0.6757,
+      "step": 6597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3407918712477462e-06,
+      "loss": 0.6844,
+      "step": 6598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3380822059489725e-06,
+      "loss": 0.5894,
+      "step": 6599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335373419809644e-06,
+      "loss": 0.7211,
+      "step": 6600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33266551318724e-06,
+      "loss": 0.6406,
+      "step": 6601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3299584864391153e-06,
+      "loss": 0.5868,
+      "step": 6602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3272523399225064e-06,
+      "loss": 0.6162,
+      "step": 6603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324547073994544e-06,
+      "loss": 0.6372,
+      "step": 6604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321842689012235e-06,
+      "loss": 0.6131,
+      "step": 6605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3191391853324683e-06,
+      "loss": 0.6919,
+      "step": 6606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3164365633120223e-06,
+      "loss": 0.6494,
+      "step": 6607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.313734823307555e-06,
+      "loss": 0.6991,
+      "step": 6608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311033965675611e-06,
+      "loss": 0.628,
+      "step": 6609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308333990772615e-06,
+      "loss": 0.6627,
+      "step": 6610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305634898954878e-06,
+      "loss": 0.4597,
+      "step": 6611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302936690578591e-06,
+      "loss": 0.6724,
+      "step": 6612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3002393659998357e-06,
+      "loss": 0.6437,
+      "step": 6613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2975429255745708e-06,
+      "loss": 0.6119,
+      "step": 6614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2948473696586392e-06,
+      "loss": 0.6723,
+      "step": 6615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292152698607768e-06,
+      "loss": 0.6627,
+      "step": 6616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2894589127775666e-06,
+      "loss": 0.6037,
+      "step": 6617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286766012523529e-06,
+      "loss": 0.6858,
+      "step": 6618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2840739982010284e-06,
+      "loss": 0.618,
+      "step": 6619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2813828701653315e-06,
+      "loss": 0.642,
+      "step": 6620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278692628771577e-06,
+      "loss": 0.6131,
+      "step": 6621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276003274374786e-06,
+      "loss": 0.66,
+      "step": 6622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273314807329876e-06,
+      "loss": 0.6576,
+      "step": 6623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2706272279916328e-06,
+      "loss": 0.5738,
+      "step": 6624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267940536714732e-06,
+      "loss": 0.6589,
+      "step": 6625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26525473385373e-06,
+      "loss": 0.6974,
+      "step": 6626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262569819763067e-06,
+      "loss": 0.6379,
+      "step": 6627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2598857947970663e-06,
+      "loss": 0.6167,
+      "step": 6628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2572026593099314e-06,
+      "loss": 0.6272,
+      "step": 6629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254520413655752e-06,
+      "loss": 0.6012,
+      "step": 6630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.251839058188496e-06,
+      "loss": 0.6018,
+      "step": 6631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2491585932620163e-06,
+      "loss": 0.6213,
+      "step": 6632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246479019230052e-06,
+      "loss": 0.6041,
+      "step": 6633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2438003364462188e-06,
+      "loss": 0.6811,
+      "step": 6634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2411225452640173e-06,
+      "loss": 0.575,
+      "step": 6635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2384456460368285e-06,
+      "loss": 0.6243,
+      "step": 6636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235769639117916e-06,
+      "loss": 0.6475,
+      "step": 6637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233094524860432e-06,
+      "loss": 0.5583,
+      "step": 6638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230420303617402e-06,
+      "loss": 0.6647,
+      "step": 6639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2277469757417403e-06,
+      "loss": 0.7247,
+      "step": 6640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2250745415862337e-06,
+      "loss": 0.6555,
+      "step": 6641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2224030015035658e-06,
+      "loss": 0.6174,
+      "step": 6642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2197323558462922e-06,
+      "loss": 0.6893,
+      "step": 6643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2170626049668495e-06,
+      "loss": 0.6885,
+      "step": 6644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2143937492175627e-06,
+      "loss": 0.6662,
+      "step": 6645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2117257889506315e-06,
+      "loss": 0.6274,
+      "step": 6646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2090587245181437e-06,
+      "loss": 0.6712,
+      "step": 6647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2063925562720653e-06,
+      "loss": 0.5921,
+      "step": 6648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2037272845642466e-06,
+      "loss": 0.6052,
+      "step": 6649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2010629097464153e-06,
+      "loss": 0.609,
+      "step": 6650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1983994321701806e-06,
+      "loss": 0.6518,
+      "step": 6651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1957368521870435e-06,
+      "loss": 0.7102,
+      "step": 6652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1930751701483764e-06,
+      "loss": 0.6698,
+      "step": 6653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190414386405436e-06,
+      "loss": 0.6808,
+      "step": 6654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187754501309358e-06,
+      "loss": 0.6494,
+      "step": 6655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1850955152111617e-06,
+      "loss": 0.6069,
+      "step": 6656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1824374284617533e-06,
+      "loss": 0.5922,
+      "step": 6657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1797802414119107e-06,
+      "loss": 0.662,
+      "step": 6658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177123954412298e-06,
+      "loss": 0.7049,
+      "step": 6659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174468567813461e-06,
+      "loss": 0.711,
+      "step": 6660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1718140819658204e-06,
+      "loss": 0.5607,
+      "step": 6661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1691604972196922e-06,
+      "loss": 0.5868,
+      "step": 6662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166507813925258e-06,
+      "loss": 0.6067,
+      "step": 6663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1638560324325895e-06,
+      "loss": 0.7365,
+      "step": 6664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1612051530916353e-06,
+      "loss": 0.681,
+      "step": 6665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1585551762522284e-06,
+      "loss": 0.6424,
+      "step": 6666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155906102264078e-06,
+      "loss": 0.6544,
+      "step": 6667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1532579314767787e-06,
+      "loss": 0.5962,
+      "step": 6668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1506106642398036e-06,
+      "loss": 0.6412,
+      "step": 6669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1479643009025075e-06,
+      "loss": 0.5715,
+      "step": 6670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1453188418141213e-06,
+      "loss": 0.7207,
+      "step": 6671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142674287323768e-06,
+      "loss": 0.6428,
+      "step": 6672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.14003063778044e-06,
+      "loss": 0.5767,
+      "step": 6673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1373878935330127e-06,
+      "loss": 0.6579,
+      "step": 6674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134746054930248e-06,
+      "loss": 0.6063,
+      "step": 6675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132105122320781e-06,
+      "loss": 0.6133,
+      "step": 6676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1294650960531313e-06,
+      "loss": 0.5534,
+      "step": 6677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.126825976475696e-06,
+      "loss": 0.6935,
+      "step": 6678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1241877639367556e-06,
+      "loss": 0.6289,
+      "step": 6679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.12155045878447e-06,
+      "loss": 0.6189,
+      "step": 6680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1189140613668732e-06,
+      "loss": 0.508,
+      "step": 6681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1162785720318945e-06,
+      "loss": 0.6281,
+      "step": 6682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1136439911273287e-06,
+      "loss": 0.707,
+      "step": 6683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1110103190008557e-06,
+      "loss": 0.6381,
+      "step": 6684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1083775560000373e-06,
+      "loss": 0.7046,
+      "step": 6685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1057457024723138e-06,
+      "loss": 0.6124,
+      "step": 6686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1031147587650033e-06,
+      "loss": 0.6206,
+      "step": 6687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.100484725225308e-06,
+      "loss": 0.6765,
+      "step": 6688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097855602200308e-06,
+      "loss": 0.7061,
+      "step": 6689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095227390036958e-06,
+      "loss": 0.6449,
+      "step": 6690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0926000890821063e-06,
+      "loss": 0.6661,
+      "step": 6691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089973699682468e-06,
+      "loss": 0.5703,
+      "step": 6692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0873482221846395e-06,
+      "loss": 0.5932,
+      "step": 6693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084723656935106e-06,
+      "loss": 0.5895,
+      "step": 6694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.082100004280222e-06,
+      "loss": 0.6111,
+      "step": 6695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0794772645662263e-06,
+      "loss": 0.6269,
+      "step": 6696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0768554381392366e-06,
+      "loss": 0.6287,
+      "step": 6697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074234525345249e-06,
+      "loss": 0.5622,
+      "step": 6698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.07161452653014e-06,
+      "loss": 0.7495,
+      "step": 6699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0689954420396627e-06,
+      "loss": 0.6128,
+      "step": 6700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0663772722194596e-06,
+      "loss": 0.7247,
+      "step": 6701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0637600174150393e-06,
+      "loss": 0.6671,
+      "step": 6702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0611436779717985e-06,
+      "loss": 0.4903,
+      "step": 6703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058528254235007e-06,
+      "loss": 0.6127,
+      "step": 6704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.05591374654982e-06,
+      "loss": 0.5652,
+      "step": 6705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0533001552612663e-06,
+      "loss": 0.6204,
+      "step": 6706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050687480714256e-06,
+      "loss": 0.7031,
+      "step": 6707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0480757232535773e-06,
+      "loss": 0.6533,
+      "step": 6708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045464883223903e-06,
+      "loss": 0.6568,
+      "step": 6709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0428549609697776e-06,
+      "loss": 0.6426,
+      "step": 6710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0402459568356247e-06,
+      "loss": 0.6395,
+      "step": 6711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037637871165754e-06,
+      "loss": 0.5987,
+      "step": 6712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035030704304346e-06,
+      "loss": 0.6712,
+      "step": 6713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032424456595465e-06,
+      "loss": 0.5722,
+      "step": 6714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0298191283830504e-06,
+      "loss": 0.6025,
+      "step": 6715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0272147200109213e-06,
+      "loss": 0.6136,
+      "step": 6716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024611231822778e-06,
+      "loss": 0.6109,
+      "step": 6717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022008664162197e-06,
+      "loss": 0.6538,
+      "step": 6718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0194070173726333e-06,
+      "loss": 0.6669,
+      "step": 6719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0168062917974173e-06,
+      "loss": 0.622,
+      "step": 6720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014206487779768e-06,
+      "loss": 0.6507,
+      "step": 6721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011607605662774e-06,
+      "loss": 0.6369,
+      "step": 6722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0090096457894026e-06,
+      "loss": 0.621,
+      "step": 6723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0064126085025024e-06,
+      "loss": 0.5532,
+      "step": 6724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003816494144798e-06,
+      "loss": 0.5798,
+      "step": 6725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0012213030588942e-06,
+      "loss": 0.6041,
+      "step": 6726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9986270355872694e-06,
+      "loss": 0.6641,
+      "step": 6727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99603369207229e-06,
+      "loss": 0.6339,
+      "step": 6728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9934412728561913e-06,
+      "loss": 0.6748,
+      "step": 6729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9908497782810864e-06,
+      "loss": 0.6768,
+      "step": 6730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9882592086889737e-06,
+      "loss": 0.6149,
+      "step": 6731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985669564421726e-06,
+      "loss": 0.6453,
+      "step": 6732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98308084582109e-06,
+      "loss": 0.699,
+      "step": 6733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9804930532286944e-06,
+      "loss": 0.6554,
+      "step": 6734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977906186986046e-06,
+      "loss": 0.7041,
+      "step": 6735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9753202474345266e-06,
+      "loss": 0.6439,
+      "step": 6736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9727352349153983e-06,
+      "loss": 0.6565,
+      "step": 6737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970151149769799e-06,
+      "loss": 0.6964,
+      "step": 6738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9675679923387455e-06,
+      "loss": 0.5221,
+      "step": 6739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9649857629631284e-06,
+      "loss": 0.5849,
+      "step": 6740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9624044619837256e-06,
+      "loss": 0.6673,
+      "step": 6741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9598240897411835e-06,
+      "loss": 0.6617,
+      "step": 6742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9572446465760275e-06,
+      "loss": 0.644,
+      "step": 6743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9546661328286606e-06,
+      "loss": 0.5695,
+      "step": 6744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9520885488393635e-06,
+      "loss": 0.6118,
+      "step": 6745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9495118949482983e-06,
+      "loss": 0.6561,
+      "step": 6746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.946936171495498e-06,
+      "loss": 0.5779,
+      "step": 6747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944361378820877e-06,
+      "loss": 0.6314,
+      "step": 6748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941787517264224e-06,
+      "loss": 0.6541,
+      "step": 6749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939214587165202e-06,
+      "loss": 0.595,
+      "step": 6750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9366425888633645e-06,
+      "loss": 0.7312,
+      "step": 6751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9340715226981278e-06,
+      "loss": 0.5763,
+      "step": 6752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9315013890087897e-06,
+      "loss": 0.4788,
+      "step": 6753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.6752,
+      "step": 6754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926363920414389e-06,
+      "loss": 0.6465,
+      "step": 6755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9237965861873065e-06,
+      "loss": 0.6823,
+      "step": 6756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9212301857920866e-06,
+      "loss": 0.6536,
+      "step": 6757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918664719567409e-06,
+      "loss": 0.5371,
+      "step": 6758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9161001878518325e-06,
+      "loss": 0.565,
+      "step": 6759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9135365909837975e-06,
+      "loss": 0.6226,
+      "step": 6760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9109739293016135e-06,
+      "loss": 0.6588,
+      "step": 6761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9084122031434716e-06,
+      "loss": 0.721,
+      "step": 6762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905851412847435e-06,
+      "loss": 0.6718,
+      "step": 6763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903291558751443e-06,
+      "loss": 0.6289,
+      "step": 6764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9007326411933223e-06,
+      "loss": 0.5899,
+      "step": 6765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.898174660510763e-06,
+      "loss": 0.6454,
+      "step": 6766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8956176170413375e-06,
+      "loss": 0.6769,
+      "step": 6767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8930615111224934e-06,
+      "loss": 0.5695,
+      "step": 6768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8905063430915515e-06,
+      "loss": 0.6124,
+      "step": 6769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887952113285718e-06,
+      "loss": 0.6257,
+      "step": 6770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885398822042066e-06,
+      "loss": 0.56,
+      "step": 6771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882846469697548e-06,
+      "loss": 0.6598,
+      "step": 6772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.880295056588994e-06,
+      "loss": 0.5491,
+      "step": 6773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877744583053107e-06,
+      "loss": 0.5689,
+      "step": 6774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.875195049426468e-06,
+      "loss": 0.6619,
+      "step": 6775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8726464560455346e-06,
+      "loss": 0.6558,
+      "step": 6776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8700988032466393e-06,
+      "loss": 0.6947,
+      "step": 6777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8675520913659892e-06,
+      "loss": 0.6633,
+      "step": 6778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.865006320739667e-06,
+      "loss": 0.5533,
+      "step": 6779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8624614917036373e-06,
+      "loss": 0.6226,
+      "step": 6780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8599176045937348e-06,
+      "loss": 0.6663,
+      "step": 6781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8573746597456664e-06,
+      "loss": 0.7157,
+      "step": 6782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.854832657495026e-06,
+      "loss": 0.6305,
+      "step": 6783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8522915981772726e-06,
+      "loss": 0.5656,
+      "step": 6784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849751482127746e-06,
+      "loss": 0.6786,
+      "step": 6785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8472123096816583e-06,
+      "loss": 0.611,
+      "step": 6786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8446740811741004e-06,
+      "loss": 0.6958,
+      "step": 6787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842136796940036e-06,
+      "loss": 0.6223,
+      "step": 6788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8396004573143023e-06,
+      "loss": 0.6205,
+      "step": 6789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.837065062631622e-06,
+      "loss": 0.683,
+      "step": 6790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8345306132265806e-06,
+      "loss": 0.5863,
+      "step": 6791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8319971094336463e-06,
+      "loss": 0.5732,
+      "step": 6792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829464551587159e-06,
+      "loss": 0.6647,
+      "step": 6793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826932940021335e-06,
+      "loss": 0.622,
+      "step": 6794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8244022750702672e-06,
+      "loss": 0.6078,
+      "step": 6795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8218725570679205e-06,
+      "loss": 0.5683,
+      "step": 6796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819343786348138e-06,
+      "loss": 0.5743,
+      "step": 6797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8168159632446323e-06,
+      "loss": 0.5967,
+      "step": 6798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.814289088091001e-06,
+      "loss": 0.6841,
+      "step": 6799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8117631612207084e-06,
+      "loss": 0.6461,
+      "step": 6800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809238182967092e-06,
+      "loss": 0.6513,
+      "step": 6801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8067141536633748e-06,
+      "loss": 0.601,
+      "step": 6802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8041910736426438e-06,
+      "loss": 0.6516,
+      "step": 6803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.801668943237864e-06,
+      "loss": 0.5785,
+      "step": 6804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7991477627818786e-06,
+      "loss": 0.5861,
+      "step": 6805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796627532607399e-06,
+      "loss": 0.6258,
+      "step": 6806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7941082530470178e-06,
+      "loss": 0.6307,
+      "step": 6807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7915899244331968e-06,
+      "loss": 0.5471,
+      "step": 6808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7890725470982717e-06,
+      "loss": 0.6508,
+      "step": 6809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7865561213744628e-06,
+      "loss": 0.6811,
+      "step": 6810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7840406475938553e-06,
+      "loss": 0.6455,
+      "step": 6811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7815261260884087e-06,
+      "loss": 0.5665,
+      "step": 6812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779012557189961e-06,
+      "loss": 0.6089,
+      "step": 6813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7764999412302207e-06,
+      "loss": 0.6447,
+      "step": 6814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7739882785407747e-06,
+      "loss": 0.6614,
+      "step": 6815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7714775694530782e-06,
+      "loss": 0.595,
+      "step": 6816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76896781429847e-06,
+      "loss": 0.623,
+      "step": 6817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766459013408154e-06,
+      "loss": 0.6008,
+      "step": 6818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763951167113209e-06,
+      "loss": 0.7083,
+      "step": 6819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761444275744596e-06,
+      "loss": 0.6672,
+      "step": 6820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7589383396331404e-06,
+      "loss": 0.6078,
+      "step": 6821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756433359109547e-06,
+      "loss": 0.5658,
+      "step": 6822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7539293345043928e-06,
+      "loss": 0.5849,
+      "step": 6823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7514262661481272e-06,
+      "loss": 0.6343,
+      "step": 6824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7489241543710756e-06,
+      "loss": 0.6346,
+      "step": 6825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7464229995034363e-06,
+      "loss": 0.644,
+      "step": 6826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743922801875283e-06,
+      "loss": 0.6493,
+      "step": 6827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7414235618165575e-06,
+      "loss": 0.7093,
+      "step": 6828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.738925279657084e-06,
+      "loss": 0.5576,
+      "step": 6829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736427955726555e-06,
+      "loss": 0.6832,
+      "step": 6830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733931590354536e-06,
+      "loss": 0.5922,
+      "step": 6831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7314361838704673e-06,
+      "loss": 0.665,
+      "step": 6832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7289417366036618e-06,
+      "loss": 0.6299,
+      "step": 6833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7264482488833075e-06,
+      "loss": 0.577,
+      "step": 6834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7239557210384615e-06,
+      "loss": 0.6867,
+      "step": 6835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7214641533980623e-06,
+      "loss": 0.6663,
+      "step": 6836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.718973546290916e-06,
+      "loss": 0.6125,
+      "step": 6837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716483900045699e-06,
+      "loss": 0.6644,
+      "step": 6838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713995214990971e-06,
+      "loss": 0.5651,
+      "step": 6839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7115074914551544e-06,
+      "loss": 0.6091,
+      "step": 6840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7090207297665504e-06,
+      "loss": 0.6592,
+      "step": 6841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7065349302533307e-06,
+      "loss": 0.7039,
+      "step": 6842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.704050093243541e-06,
+      "loss": 0.5821,
+      "step": 6843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7015662190651013e-06,
+      "loss": 0.66,
+      "step": 6844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699083308045802e-06,
+      "loss": 0.5821,
+      "step": 6845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696601360513309e-06,
+      "loss": 0.548,
+      "step": 6846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6941203767951598e-06,
+      "loss": 0.7131,
+      "step": 6847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.691640357218759e-06,
+      "loss": 0.6233,
+      "step": 6848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6891613021113984e-06,
+      "loss": 0.6892,
+      "step": 6849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686683211800231e-06,
+      "loss": 0.6368,
+      "step": 6850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684206086612282e-06,
+      "loss": 0.6399,
+      "step": 6851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6817299268744557e-06,
+      "loss": 0.6616,
+      "step": 6852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679254732913522e-06,
+      "loss": 0.5316,
+      "step": 6853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6767805050561323e-06,
+      "loss": 0.6155,
+      "step": 6854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674307243628803e-06,
+      "loss": 0.6478,
+      "step": 6855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6718349489579254e-06,
+      "loss": 0.6629,
+      "step": 6856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6693636213697615e-06,
+      "loss": 0.6685,
+      "step": 6857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6668932611904463e-06,
+      "loss": 0.6289,
+      "step": 6858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6644238687459934e-06,
+      "loss": 0.658,
+      "step": 6859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6619554443622796e-06,
+      "loss": 0.6409,
+      "step": 6860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659487988365059e-06,
+      "loss": 0.5555,
+      "step": 6861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657021501079956e-06,
+      "loss": 0.6634,
+      "step": 6862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6545559828324684e-06,
+      "loss": 0.5634,
+      "step": 6863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6520914339479643e-06,
+      "loss": 0.5596,
+      "step": 6864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6496278547516863e-06,
+      "loss": 0.5661,
+      "step": 6865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647165245568747e-06,
+      "loss": 0.6208,
+      "step": 6866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6447036067241314e-06,
+      "loss": 0.5445,
+      "step": 6867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642242938542696e-06,
+      "loss": 0.6049,
+      "step": 6868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6397832413491743e-06,
+      "loss": 0.5699,
+      "step": 6869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.637324515468164e-06,
+      "loss": 0.6783,
+      "step": 6870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6348667612241397e-06,
+      "loss": 0.5306,
+      "step": 6871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6324099789414415e-06,
+      "loss": 0.5863,
+      "step": 6872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629954168944293e-06,
+      "loss": 0.5873,
+      "step": 6873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6274993315567785e-06,
+      "loss": 0.6146,
+      "step": 6874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625045467102859e-06,
+      "loss": 0.6584,
+      "step": 6875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622592575906364e-06,
+      "loss": 0.6551,
+      "step": 6876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620140658290996e-06,
+      "loss": 0.6337,
+      "step": 6877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6176897145803325e-06,
+      "loss": 0.6325,
+      "step": 6878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6152397450978196e-06,
+      "loss": 0.6629,
+      "step": 6879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6127907501667726e-06,
+      "loss": 0.6933,
+      "step": 6880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6103427301103803e-06,
+      "loss": 0.6936,
+      "step": 6881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6078956852517034e-06,
+      "loss": 0.6393,
+      "step": 6882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6054496159136734e-06,
+      "loss": 0.6604,
+      "step": 6883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6030045224190924e-06,
+      "loss": 0.5747,
+      "step": 6884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600560405090635e-06,
+      "loss": 0.6187,
+      "step": 6885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5981172642508455e-06,
+      "loss": 0.6848,
+      "step": 6886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5956751002221393e-06,
+      "loss": 0.649,
+      "step": 6887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5932339133268057e-06,
+      "loss": 0.7149,
+      "step": 6888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5907937038870035e-06,
+      "loss": 0.6366,
+      "step": 6889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588354472224759e-06,
+      "loss": 0.5584,
+      "step": 6890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5859162186619756e-06,
+      "loss": 0.6583,
+      "step": 6891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5834789435204245e-06,
+      "loss": 0.6358,
+      "step": 6892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5810426471217463e-06,
+      "loss": 0.6432,
+      "step": 6893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5786073297874547e-06,
+      "loss": 0.7015,
+      "step": 6894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576172991838933e-06,
+      "loss": 0.6386,
+      "step": 6895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573739633597436e-06,
+      "loss": 0.5814,
+      "step": 6896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5713072553840867e-06,
+      "loss": 0.5575,
+      "step": 6897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5688758575198856e-06,
+      "loss": 0.5878,
+      "step": 6898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5664454403256966e-06,
+      "loss": 0.615,
+      "step": 6899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5640160041222573e-06,
+      "loss": 0.5955,
+      "step": 6900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5615875492301746e-06,
+      "loss": 0.6454,
+      "step": 6901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5591600759699274e-06,
+      "loss": 0.5829,
+      "step": 6902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5567335846618634e-06,
+      "loss": 0.7068,
+      "step": 6903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554308075626203e-06,
+      "loss": 0.6299,
+      "step": 6904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5518835491830354e-06,
+      "loss": 0.5717,
+      "step": 6905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5494600056523157e-06,
+      "loss": 0.5984,
+      "step": 6906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5470374453538803e-06,
+      "loss": 0.6349,
+      "step": 6907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5446158686074286e-06,
+      "loss": 0.6371,
+      "step": 6908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5421952757325254e-06,
+      "loss": 0.6632,
+      "step": 6909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5397756670486186e-06,
+      "loss": 0.6887,
+      "step": 6910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5373570428750162e-06,
+      "loss": 0.6786,
+      "step": 6911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534939403530898e-06,
+      "loss": 0.6844,
+      "step": 6912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5325227493353155e-06,
+      "loss": 0.6153,
+      "step": 6913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5301070806071905e-06,
+      "loss": 0.6678,
+      "step": 6914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527692397665311e-06,
+      "loss": 0.5855,
+      "step": 6915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5252787008283406e-06,
+      "loss": 0.6833,
+      "step": 6916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5228659904148057e-06,
+      "loss": 0.6313,
+      "step": 6917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520454266743112e-06,
+      "loss": 0.4709,
+      "step": 6918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5180435301315286e-06,
+      "loss": 0.689,
+      "step": 6919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5156337808981923e-06,
+      "loss": 0.7216,
+      "step": 6920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.513225019361116e-06,
+      "loss": 0.6359,
+      "step": 6921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5108172458381763e-06,
+      "loss": 0.6193,
+      "step": 6922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5084104606471237e-06,
+      "loss": 0.6167,
+      "step": 6923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506004664105577e-06,
+      "loss": 0.5224,
+      "step": 6924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5035998565310193e-06,
+      "loss": 0.5899,
+      "step": 6925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5011960382408162e-06,
+      "loss": 0.6705,
+      "step": 6926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4987932095521862e-06,
+      "loss": 0.6422,
+      "step": 6927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496391370782234e-06,
+      "loss": 0.6534,
+      "step": 6928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49399052224792e-06,
+      "loss": 0.5191,
+      "step": 6929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4915906642660814e-06,
+      "loss": 0.6036,
+      "step": 6930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4891917971534206e-06,
+      "loss": 0.5604,
+      "step": 6931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4867939212265135e-06,
+      "loss": 0.6208,
+      "step": 6932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4843970368018e-06,
+      "loss": 0.6244,
+      "step": 6933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.482001144195594e-06,
+      "loss": 0.6752,
+      "step": 6934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479606243724074e-06,
+      "loss": 0.6444,
+      "step": 6935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4772123357032908e-06,
+      "loss": 0.7756,
+      "step": 6936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474819420449166e-06,
+      "loss": 0.632,
+      "step": 6937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4724274982774866e-06,
+      "loss": 0.6299,
+      "step": 6938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4700365695039097e-06,
+      "loss": 0.6852,
+      "step": 6939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46764663444396e-06,
+      "loss": 0.6174,
+      "step": 6940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4652576934130324e-06,
+      "loss": 0.6073,
+      "step": 6941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4628697467263916e-06,
+      "loss": 0.5893,
+      "step": 6942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.460482794699166e-06,
+      "loss": 0.7234,
+      "step": 6943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4580968376463633e-06,
+      "loss": 0.5708,
+      "step": 6944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455711875882849e-06,
+      "loss": 0.5986,
+      "step": 6945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4533279097233596e-06,
+      "loss": 0.6898,
+      "step": 6946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.450944939482509e-06,
+      "loss": 0.6217,
+      "step": 6947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4485629654747687e-06,
+      "loss": 0.6477,
+      "step": 6948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446181988014482e-06,
+      "loss": 0.6746,
+      "step": 6949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4438020074158644e-06,
+      "loss": 0.5862,
+      "step": 6950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441423023992994e-06,
+      "loss": 0.6046,
+      "step": 6951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4390450380598217e-06,
+      "loss": 0.6543,
+      "step": 6952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4366680499301654e-06,
+      "loss": 0.6253,
+      "step": 6953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43429205991771e-06,
+      "loss": 0.6091,
+      "step": 6954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4319170683360125e-06,
+      "loss": 0.6081,
+      "step": 6955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4295430754984906e-06,
+      "loss": 0.6641,
+      "step": 6956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427170081718442e-06,
+      "loss": 0.6236,
+      "step": 6957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4247980873090218e-06,
+      "loss": 0.573,
+      "step": 6958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4224270925832583e-06,
+      "loss": 0.6637,
+      "step": 6959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420057097854046e-06,
+      "loss": 0.5912,
+      "step": 6960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4176881034341458e-06,
+      "loss": 0.5447,
+      "step": 6961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4153201096361946e-06,
+      "loss": 0.6083,
+      "step": 6962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4129531167726882e-06,
+      "loss": 0.5248,
+      "step": 6963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4105871251559943e-06,
+      "loss": 0.5541,
+      "step": 6964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408222135098348e-06,
+      "loss": 0.5579,
+      "step": 6965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4058581469118493e-06,
+      "loss": 0.5935,
+      "step": 6966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4034951609084743e-06,
+      "loss": 0.7494,
+      "step": 6967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401133177400058e-06,
+      "loss": 0.651,
+      "step": 6968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398772196698308e-06,
+      "loss": 0.6132,
+      "step": 6969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3964122191147967e-06,
+      "loss": 0.5803,
+      "step": 6970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3940532449609643e-06,
+      "loss": 0.6328,
+      "step": 6971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3916952745481226e-06,
+      "loss": 0.7324,
+      "step": 6972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3893383081874456e-06,
+      "loss": 0.5502,
+      "step": 6973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3869823461899788e-06,
+      "loss": 0.6749,
+      "step": 6974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3846273888666327e-06,
+      "loss": 0.6073,
+      "step": 6975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.382273436528184e-06,
+      "loss": 0.6116,
+      "step": 6976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3799204894852834e-06,
+      "loss": 0.5952,
+      "step": 6977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3775685480484432e-06,
+      "loss": 0.6917,
+      "step": 6978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3752176125280434e-06,
+      "loss": 0.6609,
+      "step": 6979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372867683234329e-06,
+      "loss": 0.6394,
+      "step": 6980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3705187604774215e-06,
+      "loss": 0.6452,
+      "step": 6981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3681708445673e-06,
+      "loss": 0.6935,
+      "step": 6982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365823935813815e-06,
+      "loss": 0.6406,
+      "step": 6983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3634780345266805e-06,
+      "loss": 0.6209,
+      "step": 6984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361133141015481e-06,
+      "loss": 0.6161,
+      "step": 6985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3587892555896707e-06,
+      "loss": 0.695,
+      "step": 6986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3564463785585656e-06,
+      "loss": 0.6395,
+      "step": 6987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3541045102313498e-06,
+      "loss": 0.6335,
+      "step": 6988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351763650917074e-06,
+      "loss": 0.5734,
+      "step": 6989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3494238009246574e-06,
+      "loss": 0.607,
+      "step": 6990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347084960562885e-06,
+      "loss": 0.5985,
+      "step": 6991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.34474713014041e-06,
+      "loss": 0.6565,
+      "step": 6992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342410309965748e-06,
+      "loss": 0.6311,
+      "step": 6993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340074500347287e-06,
+      "loss": 0.6239,
+      "step": 6994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3377397015932746e-06,
+      "loss": 0.6383,
+      "step": 6995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3354059140118358e-06,
+      "loss": 0.6393,
+      "step": 6996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3330731379109526e-06,
+      "loss": 0.6855,
+      "step": 6997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330741373598474e-06,
+      "loss": 0.5975,
+      "step": 6998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3284106213821236e-06,
+      "loss": 0.6485,
+      "step": 6999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3260808815694823e-06,
+      "loss": 0.6141,
+      "step": 7000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323752154468003e-06,
+      "loss": 0.6235,
+      "step": 7001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3214244403850007e-06,
+      "loss": 0.7178,
+      "step": 7002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3190977396276602e-06,
+      "loss": 0.623,
+      "step": 7003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3167720525030324e-06,
+      "loss": 0.6013,
+      "step": 7004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3144473793180277e-06,
+      "loss": 0.7012,
+      "step": 7005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312123720379437e-06,
+      "loss": 0.5929,
+      "step": 7006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.309801075993904e-06,
+      "loss": 0.6773,
+      "step": 7007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307479446467944e-06,
+      "loss": 0.6424,
+      "step": 7008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3051588321079375e-06,
+      "loss": 0.6987,
+      "step": 7009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302839233220131e-06,
+      "loss": 0.6175,
+      "step": 7010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300520650110638e-06,
+      "loss": 0.556,
+      "step": 7011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2982030830854353e-06,
+      "loss": 0.6895,
+      "step": 7012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2958865324503697e-06,
+      "loss": 0.6717,
+      "step": 7013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2935709985111467e-06,
+      "loss": 0.6467,
+      "step": 7014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2912564815733494e-06,
+      "loss": 0.6787,
+      "step": 7015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288942981942417e-06,
+      "loss": 0.6385,
+      "step": 7016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2866304999236535e-06,
+      "loss": 0.6385,
+      "step": 7017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2843190358222378e-06,
+      "loss": 0.6705,
+      "step": 7018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2820085899432077e-06,
+      "loss": 0.5828,
+      "step": 7019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2796991625914667e-06,
+      "loss": 0.5838,
+      "step": 7020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2773907540717866e-06,
+      "loss": 0.7186,
+      "step": 7021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2750833646888015e-06,
+      "loss": 0.5755,
+      "step": 7022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2727769947470134e-06,
+      "loss": 0.6092,
+      "step": 7023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270471644550789e-06,
+      "loss": 0.5118,
+      "step": 7024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2681673144043582e-06,
+      "loss": 0.6227,
+      "step": 7025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2658640046118242e-06,
+      "loss": 0.6006,
+      "step": 7026
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263561715477147e-06,
+      "loss": 0.6673,
+      "step": 7027
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2612604473041556e-06,
+      "loss": 0.7427,
+      "step": 7028
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2589602003965426e-06,
+      "loss": 0.6052,
+      "step": 7029
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256660975057867e-06,
+      "loss": 0.6442,
+      "step": 7030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.254362771591554e-06,
+      "loss": 0.7645,
+      "step": 7031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25206559030089e-06,
+      "loss": 0.6265,
+      "step": 7032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2497694314890296e-06,
+      "loss": 0.5334,
+      "step": 7033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247474295458996e-06,
+      "loss": 0.5916,
+      "step": 7034
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2451801825136675e-06,
+      "loss": 0.6454,
+      "step": 7035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2428870929558012e-06,
+      "loss": 0.6293,
+      "step": 7036
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240595027088006e-06,
+      "loss": 0.6892,
+      "step": 7037
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2383039852127617e-06,
+      "loss": 0.6593,
+      "step": 7038
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2360139676324145e-06,
+      "loss": 0.6997,
+      "step": 7039
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2337249746491695e-06,
+      "loss": 0.6333,
+      "step": 7040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2314370065651037e-06,
+      "loss": 0.5719,
+      "step": 7041
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2291500636821537e-06,
+      "loss": 0.5943,
+      "step": 7042
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2268641463021235e-06,
+      "loss": 0.6798,
+      "step": 7043
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2245792547266774e-06,
+      "loss": 0.6163,
+      "step": 7044
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222295389257354e-06,
+      "loss": 0.5059,
+      "step": 7045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2200125501955472e-06,
+      "loss": 0.6714,
+      "step": 7046
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2177307378425182e-06,
+      "loss": 0.6008,
+      "step": 7047
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215449952499393e-06,
+      "loss": 0.6418,
+      "step": 7048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2131701944671623e-06,
+      "loss": 0.5515,
+      "step": 7049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2108914640466815e-06,
+      "loss": 0.568,
+      "step": 7050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.208613761538667e-06,
+      "loss": 0.622,
+      "step": 7051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2063370872437082e-06,
+      "loss": 0.6562,
+      "step": 7052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2040614414622487e-06,
+      "loss": 0.5954,
+      "step": 7053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2017868244946006e-06,
+      "loss": 0.6401,
+      "step": 7054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1995132366409435e-06,
+      "loss": 0.6057,
+      "step": 7055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1972406782013167e-06,
+      "loss": 0.6513,
+      "step": 7056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194969149475624e-06,
+      "loss": 0.6086,
+      "step": 7057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192698650763635e-06,
+      "loss": 0.7138,
+      "step": 7058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1904291823649835e-06,
+      "loss": 0.612,
+      "step": 7059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188160744579165e-06,
+      "loss": 0.7156,
+      "step": 7060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185893337705541e-06,
+      "loss": 0.6571,
+      "step": 7061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183626962043337e-06,
+      "loss": 0.5829,
+      "step": 7062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181361617891642e-06,
+      "loss": 0.7148,
+      "step": 7063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1790973055494045e-06,
+      "loss": 0.6021,
+      "step": 7064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1768340253154483e-06,
+      "loss": 0.6163,
+      "step": 7065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17457177748845e-06,
+      "loss": 0.6457,
+      "step": 7066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.172310562366955e-06,
+      "loss": 0.5936,
+      "step": 7067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170050380249369e-06,
+      "loss": 0.596,
+      "step": 7068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1677912314339653e-06,
+      "loss": 0.6682,
+      "step": 7069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1655331162188764e-06,
+      "loss": 0.6305,
+      "step": 7070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1632760349021055e-06,
+      "loss": 0.6524,
+      "step": 7071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161019987781513e-06,
+      "loss": 0.5051,
+      "step": 7072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1587649751548246e-06,
+      "loss": 0.6042,
+      "step": 7073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156510997319626e-06,
+      "loss": 0.6858,
+      "step": 7074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1542580545733773e-06,
+      "loss": 0.5732,
+      "step": 7075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1520061472133903e-06,
+      "loss": 0.5611,
+      "step": 7076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1497552755368446e-06,
+      "loss": 0.5231,
+      "step": 7077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1475054398407845e-06,
+      "loss": 0.5253,
+      "step": 7078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.145256640422114e-06,
+      "loss": 0.692,
+      "step": 7079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143008877577605e-06,
+      "loss": 0.5919,
+      "step": 7080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140762151603888e-06,
+      "loss": 0.6643,
+      "step": 7081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138516462797459e-06,
+      "loss": 0.7126,
+      "step": 7082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1362718114546777e-06,
+      "loss": 0.6522,
+      "step": 7083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1340281978717637e-06,
+      "loss": 0.6737,
+      "step": 7084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131785622344806e-06,
+      "loss": 0.6199,
+      "step": 7085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.129544085169751e-06,
+      "loss": 0.6304,
+      "step": 7086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1273035866424097e-06,
+      "loss": 0.6219,
+      "step": 7087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1250641270584527e-06,
+      "loss": 0.6149,
+      "step": 7088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122825706713423e-06,
+      "loss": 0.6349,
+      "step": 7089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1205883259027173e-06,
+      "loss": 0.6578,
+      "step": 7090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.118351984921597e-06,
+      "loss": 0.6807,
+      "step": 7091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1161166840651893e-06,
+      "loss": 0.6255,
+      "step": 7092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1138824236284803e-06,
+      "loss": 0.5501,
+      "step": 7093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11164920390632e-06,
+      "loss": 0.5826,
+      "step": 7094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1094170251934254e-06,
+      "loss": 0.6169,
+      "step": 7095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107185887784371e-06,
+      "loss": 0.5342,
+      "step": 7096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104955791973594e-06,
+      "loss": 0.6663,
+      "step": 7097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1027267380553972e-06,
+      "loss": 0.6489,
+      "step": 7098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100498726323943e-06,
+      "loss": 0.6541,
+      "step": 7099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098271757073258e-06,
+      "loss": 0.6631,
+      "step": 7100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096045830597231e-06,
+      "loss": 0.6375,
+      "step": 7101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093820947189612e-06,
+      "loss": 0.5594,
+      "step": 7102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0915971071440123e-06,
+      "loss": 0.6639,
+      "step": 7103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0893743107539123e-06,
+      "loss": 0.6214,
+      "step": 7104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0871525583126472e-06,
+      "loss": 0.628,
+      "step": 7105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0849318501134176e-06,
+      "loss": 0.6419,
+      "step": 7106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0827121864492827e-06,
+      "loss": 0.6108,
+      "step": 7107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080493567613172e-06,
+      "loss": 0.7368,
+      "step": 7108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0782759938978703e-06,
+      "loss": 0.5885,
+      "step": 7109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0760594655960255e-06,
+      "loss": 0.6174,
+      "step": 7110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0738439830001465e-06,
+      "loss": 0.672,
+      "step": 7111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071629546402609e-06,
+      "loss": 0.644,
+      "step": 7112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0694161560956438e-06,
+      "loss": 0.6993,
+      "step": 7113
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067203812371352e-06,
+      "loss": 0.6628,
+      "step": 7114
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064992515521691e-06,
+      "loss": 0.6339,
+      "step": 7115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0627822658384787e-06,
+      "loss": 0.6318,
+      "step": 7116
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0605730636133993e-06,
+      "loss": 0.6923,
+      "step": 7117
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0583649091379954e-06,
+      "loss": 0.6376,
+      "step": 7118
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056157802703672e-06,
+      "loss": 0.5919,
+      "step": 7119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0539517446016975e-06,
+      "loss": 0.6439,
+      "step": 7120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517467351232014e-06,
+      "loss": 0.5832,
+      "step": 7121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04954277455917e-06,
+      "loss": 0.6526,
+      "step": 7122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0473398632004628e-06,
+      "loss": 0.6497,
+      "step": 7123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0451380013377876e-06,
+      "loss": 0.5533,
+      "step": 7124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042937189261719e-06,
+      "loss": 0.6561,
+      "step": 7125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040737427262699e-06,
+      "loss": 0.6645,
+      "step": 7126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0385387156310234e-06,
+      "loss": 0.5629,
+      "step": 7127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0363410546568494e-06,
+      "loss": 0.6475,
+      "step": 7128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0341444446301993e-06,
+      "loss": 0.6688,
+      "step": 7129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0319488858409552e-06,
+      "loss": 0.5658,
+      "step": 7130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0297543785788597e-06,
+      "loss": 0.567,
+      "step": 7131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027560923133519e-06,
+      "loss": 0.5487,
+      "step": 7132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025368519794394e-06,
+      "loss": 0.6535,
+      "step": 7133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0231771688508185e-06,
+      "loss": 0.656,
+      "step": 7134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0209868705919767e-06,
+      "loss": 0.5515,
+      "step": 7135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0187976253069196e-06,
+      "loss": 0.6209,
+      "step": 7136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0166094332845553e-06,
+      "loss": 0.668,
+      "step": 7137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014422294813656e-06,
+      "loss": 0.5696,
+      "step": 7138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012236210182853e-06,
+      "loss": 0.6082,
+      "step": 7139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010051179680641e-06,
+      "loss": 0.6532,
+      "step": 7140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0078672035953693e-06,
+      "loss": 0.6733,
+      "step": 7141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.005684282215259e-06,
+      "loss": 0.6098,
+      "step": 7142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00350241582838e-06,
+      "loss": 0.5835,
+      "step": 7143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0013216047226756e-06,
+      "loss": 0.645,
+      "step": 7144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9991418491859383e-06,
+      "loss": 0.613,
+      "step": 7145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996963149505826e-06,
+      "loss": 0.6869,
+      "step": 7146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9947855059698584e-06,
+      "loss": 0.4947,
+      "step": 7147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992608918865413e-06,
+      "loss": 0.5784,
+      "step": 7148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990433388479731e-06,
+      "loss": 0.5158,
+      "step": 7149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988258915099912e-06,
+      "loss": 0.6136,
+      "step": 7150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9860854990129176e-06,
+      "loss": 0.6385,
+      "step": 7151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983913140505567e-06,
+      "loss": 0.6823,
+      "step": 7152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9817418398645405e-06,
+      "loss": 0.7626,
+      "step": 7153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9795715973763853e-06,
+      "loss": 0.6121,
+      "step": 7154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9774024133275016e-06,
+      "loss": 0.6865,
+      "step": 7155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975234288004151e-06,
+      "loss": 0.6577,
+      "step": 7156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9730672216924573e-06,
+      "loss": 0.622,
+      "step": 7157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9709012146784035e-06,
+      "loss": 0.5775,
+      "step": 7158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9687362672478307e-06,
+      "loss": 0.6485,
+      "step": 7159
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9665723796864476e-06,
+      "loss": 0.6213,
+      "step": 7160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9644095522798158e-06,
+      "loss": 0.6306,
+      "step": 7161
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962247785313355e-06,
+      "loss": 0.6195,
+      "step": 7162
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960087079072356e-06,
+      "loss": 0.5559,
+      "step": 7163
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9579274338419586e-06,
+      "loss": 0.6194,
+      "step": 7164
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955768849907168e-06,
+      "loss": 0.733,
+      "step": 7165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9536113275528477e-06,
+      "loss": 0.5571,
+      "step": 7166
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951454867063721e-06,
+      "loss": 0.6913,
+      "step": 7167
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9492994687243715e-06,
+      "loss": 0.6198,
+      "step": 7168
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9471451328192437e-06,
+      "loss": 0.6058,
+      "step": 7169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94499185963264e-06,
+      "loss": 0.6283,
+      "step": 7170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9428396494487225e-06,
+      "loss": 0.6008,
+      "step": 7171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9406885025515132e-06,
+      "loss": 0.6565,
+      "step": 7172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938538419224898e-06,
+      "loss": 0.5796,
+      "step": 7173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9363893997526184e-06,
+      "loss": 0.5834,
+      "step": 7174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9342414444182734e-06,
+      "loss": 0.6539,
+      "step": 7175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932094553505326e-06,
+      "loss": 0.6002,
+      "step": 7176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929948727297096e-06,
+      "loss": 0.57,
+      "step": 7177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927803966076762e-06,
+      "loss": 0.6127,
+      "step": 7178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9256602701273685e-06,
+      "loss": 0.586,
+      "step": 7179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923517639731811e-06,
+      "loss": 0.6804,
+      "step": 7180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9213760751728494e-06,
+      "loss": 0.6068,
+      "step": 7181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919235576733098e-06,
+      "loss": 0.5973,
+      "step": 7182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9170961446950407e-06,
+      "loss": 0.5396,
+      "step": 7183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914957779341009e-06,
+      "loss": 0.6418,
+      "step": 7184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9128204809532004e-06,
+      "loss": 0.6003,
+      "step": 7185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910684249813668e-06,
+      "loss": 0.6164,
+      "step": 7186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9085490862043276e-06,
+      "loss": 0.5908,
+      "step": 7187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9064149904069507e-06,
+      "loss": 0.6626,
+      "step": 7188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9042819627031705e-06,
+      "loss": 0.6612,
+      "step": 7189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9021500033744777e-06,
+      "loss": 0.6022,
+      "step": 7190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9000191127022239e-06,
+      "loss": 0.6185,
+      "step": 7191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8978892909676138e-06,
+      "loss": 0.6119,
+      "step": 7192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8957605384517209e-06,
+      "loss": 0.6376,
+      "step": 7193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893632855435471e-06,
+      "loss": 0.5964,
+      "step": 7194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.89150624219965e-06,
+      "loss": 0.6118,
+      "step": 7195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8893806990248986e-06,
+      "loss": 0.6437,
+      "step": 7196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.887256226191726e-06,
+      "loss": 0.633,
+      "step": 7197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8851328239804922e-06,
+      "loss": 0.5586,
+      "step": 7198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8830104926714188e-06,
+      "loss": 0.6949,
+      "step": 7199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.880889232544585e-06,
+      "loss": 0.4727,
+      "step": 7200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8787690438799278e-06,
+      "loss": 0.5536,
+      "step": 7201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8766499269572425e-06,
+      "loss": 0.6255,
+      "step": 7202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874531882056191e-06,
+      "loss": 0.6787,
+      "step": 7203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8724149094562827e-06,
+      "loss": 0.575,
+      "step": 7204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8702990094368911e-06,
+      "loss": 0.6618,
+      "step": 7205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8681841822772461e-06,
+      "loss": 0.5727,
+      "step": 7206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8660704282564391e-06,
+      "loss": 0.6249,
+      "step": 7207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8639577476534154e-06,
+      "loss": 0.6558,
+      "step": 7208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8618461407469822e-06,
+      "loss": 0.6388,
+      "step": 7209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8597356078158046e-06,
+      "loss": 0.6623,
+      "step": 7210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857626149138403e-06,
+      "loss": 0.6561,
+      "step": 7211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.855517764993161e-06,
+      "loss": 0.6218,
+      "step": 7212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8534104556583166e-06,
+      "loss": 0.6338,
+      "step": 7213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8513042214119669e-06,
+      "loss": 0.5647,
+      "step": 7214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8491990625320654e-06,
+      "loss": 0.6291,
+      "step": 7215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8470949792964289e-06,
+      "loss": 0.6148,
+      "step": 7216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8449919719827282e-06,
+      "loss": 0.5822,
+      "step": 7217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8428900408684924e-06,
+      "loss": 0.6083,
+      "step": 7218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8407891862311078e-06,
+      "loss": 0.6872,
+      "step": 7219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8386894083478214e-06,
+      "loss": 0.5815,
+      "step": 7220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8365907074957324e-06,
+      "loss": 0.6173,
+      "step": 7221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8344930839518084e-06,
+      "loss": 0.6856,
+      "step": 7222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8323965379928654e-06,
+      "loss": 0.6462,
+      "step": 7223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8303010698955803e-06,
+      "loss": 0.6274,
+      "step": 7224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8282066799364874e-06,
+      "loss": 0.5705,
+      "step": 7225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8261133683919797e-06,
+      "loss": 0.6369,
+      "step": 7226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8240211355383053e-06,
+      "loss": 0.5937,
+      "step": 7227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8219299816515745e-06,
+      "loss": 0.5938,
+      "step": 7228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198399070077499e-06,
+      "loss": 0.6626,
+      "step": 7229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8177509118826542e-06,
+      "loss": 0.6901,
+      "step": 7230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8156629965519712e-06,
+      "loss": 0.6143,
+      "step": 7231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8135761612912372e-06,
+      "loss": 0.702,
+      "step": 7232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8114904063758432e-06,
+      "loss": 0.6725,
+      "step": 7233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8094057320810499e-06,
+      "loss": 0.663,
+      "step": 7234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8073221386819618e-06,
+      "loss": 0.6382,
+      "step": 7235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8052396264535487e-06,
+      "loss": 0.6218,
+      "step": 7236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8031581956706345e-06,
+      "loss": 0.6216,
+      "step": 7237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8010778466079015e-06,
+      "loss": 0.6615,
+      "step": 7238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.798998579539889e-06,
+      "loss": 0.6349,
+      "step": 7239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.796920394740993e-06,
+      "loss": 0.6463,
+      "step": 7240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7948432924854654e-06,
+      "loss": 0.6181,
+      "step": 7241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7927672730474222e-06,
+      "loss": 0.55,
+      "step": 7242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7906923367008277e-06,
+      "loss": 0.6197,
+      "step": 7243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7886184837195087e-06,
+      "loss": 0.601,
+      "step": 7244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7865457143771459e-06,
+      "loss": 0.5256,
+      "step": 7245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7844740289472785e-06,
+      "loss": 0.5973,
+      "step": 7246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782403427703303e-06,
+      "loss": 0.6425,
+      "step": 7247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7803339109184725e-06,
+      "loss": 0.6854,
+      "step": 7248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7782654788658937e-06,
+      "loss": 0.6059,
+      "step": 7249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7761981318185395e-06,
+      "loss": 0.5366,
+      "step": 7250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7741318700492294e-06,
+      "loss": 0.5391,
+      "step": 7251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772066693830642e-06,
+      "loss": 0.6844,
+      "step": 7252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.770002603435318e-06,
+      "loss": 0.56,
+      "step": 7253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7679395991356506e-06,
+      "loss": 0.6123,
+      "step": 7254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7658776812038892e-06,
+      "loss": 0.6594,
+      "step": 7255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7638168499121412e-06,
+      "loss": 0.6882,
+      "step": 7256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761757105532369e-06,
+      "loss": 0.6242,
+      "step": 7257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7596984483363944e-06,
+      "loss": 0.5946,
+      "step": 7258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7576408785958931e-06,
+      "loss": 0.6379,
+      "step": 7259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7555843965823992e-06,
+      "loss": 0.642,
+      "step": 7260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.753529002567299e-06,
+      "loss": 0.5906,
+      "step": 7261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7514746968218432e-06,
+      "loss": 0.6394,
+      "step": 7262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7494214796171338e-06,
+      "loss": 0.7259,
+      "step": 7263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7473693512241273e-06,
+      "loss": 0.6901,
+      "step": 7264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7453183119136397e-06,
+      "loss": 0.636,
+      "step": 7265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432683619563428e-06,
+      "loss": 0.6097,
+      "step": 7266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7412195016227607e-06,
+      "loss": 0.588,
+      "step": 7267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7391717311832834e-06,
+      "loss": 0.6043,
+      "step": 7268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7371250509081484e-06,
+      "loss": 0.6096,
+      "step": 7269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7350794610674482e-06,
+      "loss": 0.6145,
+      "step": 7270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7330349619311415e-06,
+      "loss": 0.663,
+      "step": 7271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7309915537690324e-06,
+      "loss": 0.5952,
+      "step": 7272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7289492368507876e-06,
+      "loss": 0.5875,
+      "step": 7273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7269080114459246e-06,
+      "loss": 0.4547,
+      "step": 7274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7248678778238215e-06,
+      "loss": 0.7173,
+      "step": 7275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7228288362537105e-06,
+      "loss": 0.5565,
+      "step": 7276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7207908870046786e-06,
+      "loss": 0.6326,
+      "step": 7277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7187540303456706e-06,
+      "loss": 0.5464,
+      "step": 7278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7167182665454852e-06,
+      "loss": 0.7114,
+      "step": 7279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714683595872777e-06,
+      "loss": 0.61,
+      "step": 7280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7126500185960614e-06,
+      "loss": 0.6687,
+      "step": 7281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7106175349837028e-06,
+      "loss": 0.6678,
+      "step": 7282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085861453039244e-06,
+      "loss": 0.6211,
+      "step": 7283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065558498248036e-06,
+      "loss": 0.5941,
+      "step": 7284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045266488142753e-06,
+      "loss": 0.6383,
+      "step": 7285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024985425401252e-06,
+      "loss": 0.636,
+      "step": 7286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.700471531270005e-06,
+      "loss": 0.6221,
+      "step": 7287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6984456152714102e-06,
+      "loss": 0.6207,
+      "step": 7288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6964207948116983e-06,
+      "loss": 0.6437,
+      "step": 7289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6943970701580792e-06,
+      "loss": 0.55,
+      "step": 7290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6923744415776223e-06,
+      "loss": 0.6645,
+      "step": 7291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6903529093372495e-06,
+      "loss": 0.6622,
+      "step": 7292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6883324737037355e-06,
+      "loss": 0.5961,
+      "step": 7293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6863131349437157e-06,
+      "loss": 0.6819,
+      "step": 7294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6842948933236758e-06,
+      "loss": 0.6356,
+      "step": 7295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6822777491099606e-06,
+      "loss": 0.587,
+      "step": 7296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6802617025687685e-06,
+      "loss": 0.5686,
+      "step": 7297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6782467539661518e-06,
+      "loss": 0.641,
+      "step": 7298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6762329035680192e-06,
+      "loss": 0.7136,
+      "step": 7299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6742201516401324e-06,
+      "loss": 0.7036,
+      "step": 7300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6722084984481147e-06,
+      "loss": 0.5877,
+      "step": 7301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670197944257438e-06,
+      "loss": 0.6298,
+      "step": 7302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668188489333431e-06,
+      "loss": 0.6639,
+      "step": 7303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.666180133941273e-06,
+      "loss": 0.7123,
+      "step": 7304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6641728783460097e-06,
+      "loss": 0.6501,
+      "step": 7305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6621667228125305e-06,
+      "loss": 0.6235,
+      "step": 7306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6601616676055843e-06,
+      "loss": 0.5806,
+      "step": 7307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.658157712989774e-06,
+      "loss": 0.6092,
+      "step": 7308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6561548592295573e-06,
+      "loss": 0.5349,
+      "step": 7309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6541531065892435e-06,
+      "loss": 0.6592,
+      "step": 7310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.652152455333006e-06,
+      "loss": 0.6874,
+      "step": 7311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6501529057248623e-06,
+      "loss": 0.6433,
+      "step": 7312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6481544580286902e-06,
+      "loss": 0.5818,
+      "step": 7313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.646157112508221e-06,
+      "loss": 0.5573,
+      "step": 7314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6441608694270394e-06,
+      "loss": 0.6423,
+      "step": 7315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6421657290485848e-06,
+      "loss": 0.6819,
+      "step": 7316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6401716916361542e-06,
+      "loss": 0.6874,
+      "step": 7317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.638178757452894e-06,
+      "loss": 0.6329,
+      "step": 7318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636186926761808e-06,
+      "loss": 0.7231,
+      "step": 7319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6341961998257528e-06,
+      "loss": 0.5611,
+      "step": 7320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6322065769074448e-06,
+      "loss": 0.6134,
+      "step": 7321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.630218058269447e-06,
+      "loss": 0.5751,
+      "step": 7322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6282306441741791e-06,
+      "loss": 0.6774,
+      "step": 7323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6262443348839196e-06,
+      "loss": 0.7238,
+      "step": 7324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.624259130660797e-06,
+      "loss": 0.6777,
+      "step": 7325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6222750317667923e-06,
+      "loss": 0.4635,
+      "step": 7326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6202920384637443e-06,
+      "loss": 0.5773,
+      "step": 7327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6183101510133436e-06,
+      "loss": 0.6721,
+      "step": 7328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6163293696771342e-06,
+      "loss": 0.6321,
+      "step": 7329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6143496947165204e-06,
+      "loss": 0.5345,
+      "step": 7330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6123711263927534e-06,
+      "loss": 0.6586,
+      "step": 7331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6103936649669406e-06,
+      "loss": 0.7145,
+      "step": 7332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6084173107000433e-06,
+      "loss": 0.7558,
+      "step": 7333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6064420638528777e-06,
+      "loss": 0.6615,
+      "step": 7334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6044679246861117e-06,
+      "loss": 0.5667,
+      "step": 7335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.602494893460268e-06,
+      "loss": 0.5661,
+      "step": 7336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600522970435726e-06,
+      "loss": 0.6001,
+      "step": 7337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5985521558727112e-06,
+      "loss": 0.662,
+      "step": 7338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5965824500313143e-06,
+      "loss": 0.6172,
+      "step": 7339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5946138531714705e-06,
+      "loss": 0.5623,
+      "step": 7340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592646365552969e-06,
+      "loss": 0.6759,
+      "step": 7341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5906799874354594e-06,
+      "loss": 0.6672,
+      "step": 7342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5887147190784391e-06,
+      "loss": 0.6695,
+      "step": 7343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5867505607412604e-06,
+      "loss": 0.5514,
+      "step": 7344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5847875126831269e-06,
+      "loss": 0.5797,
+      "step": 7345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5828255751631017e-06,
+      "loss": 0.6004,
+      "step": 7346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5808647484400942e-06,
+      "loss": 0.6503,
+      "step": 7347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5789050327728728e-06,
+      "loss": 0.6071,
+      "step": 7348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769464284200542e-06,
+      "loss": 0.6388,
+      "step": 7349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5749889356401148e-06,
+      "loss": 0.6625,
+      "step": 7350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5730325546913804e-06,
+      "loss": 0.5994,
+      "step": 7351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.57107728583203e-06,
+      "loss": 0.7155,
+      "step": 7352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569123129320096e-06,
+      "loss": 0.5807,
+      "step": 7353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5671700854134641e-06,
+      "loss": 0.6022,
+      "step": 7354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5652181543698742e-06,
+      "loss": 0.6853,
+      "step": 7355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5632673364469187e-06,
+      "loss": 0.5881,
+      "step": 7356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5613176319020396e-06,
+      "loss": 0.611,
+      "step": 7357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5593690409925411e-06,
+      "loss": 0.5848,
+      "step": 7358
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5574215639755707e-06,
+      "loss": 0.687,
+      "step": 7359
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5554752011081332e-06,
+      "loss": 0.6284,
+      "step": 7360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5535299526470881e-06,
+      "loss": 0.7287,
+      "step": 7361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5515858188491451e-06,
+      "loss": 0.5335,
+      "step": 7362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5496427999708663e-06,
+      "loss": 0.6636,
+      "step": 7363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.547700896268669e-06,
+      "loss": 0.6301,
+      "step": 7364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5457601079988226e-06,
+      "loss": 0.6343,
+      "step": 7365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.543820435417448e-06,
+      "loss": 0.6777,
+      "step": 7366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.541881878780519e-06,
+      "loss": 0.6831,
+      "step": 7367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5399444383438633e-06,
+      "loss": 0.612,
+      "step": 7368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5380081143631598e-06,
+      "loss": 0.6404,
+      "step": 7369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5360729070939451e-06,
+      "loss": 0.6415,
+      "step": 7370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5341388167916017e-06,
+      "loss": 0.5916,
+      "step": 7371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5322058437113686e-06,
+      "loss": 0.5747,
+      "step": 7372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530273988108335e-06,
+      "loss": 0.5762,
+      "step": 7373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5283432502374441e-06,
+      "loss": 0.6377,
+      "step": 7374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5264136303534893e-06,
+      "loss": 0.593,
+      "step": 7375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.524485128711123e-06,
+      "loss": 0.6093,
+      "step": 7376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5225577455648433e-06,
+      "loss": 0.6675,
+      "step": 7377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5206314811690037e-06,
+      "loss": 0.6272,
+      "step": 7378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5187063357778053e-06,
+      "loss": 0.5297,
+      "step": 7379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5167823096453126e-06,
+      "loss": 0.713,
+      "step": 7380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5148594030254304e-06,
+      "loss": 0.5684,
+      "step": 7381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5129376161719223e-06,
+      "loss": 0.6186,
+      "step": 7382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511016949338402e-06,
+      "loss": 0.501,
+      "step": 7383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5090974027783366e-06,
+      "loss": 0.5915,
+      "step": 7384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507178976745044e-06,
+      "loss": 0.6345,
+      "step": 7385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5052616714916946e-06,
+      "loss": 0.5998,
+      "step": 7386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5033454872713117e-06,
+      "loss": 0.5393,
+      "step": 7387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5014304243367671e-06,
+      "loss": 0.6259,
+      "step": 7388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4995164829407939e-06,
+      "loss": 0.634,
+      "step": 7389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497603663335967e-06,
+      "loss": 0.6218,
+      "step": 7390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4956919657747183e-06,
+      "loss": 0.6297,
+      "step": 7391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49378139050933e-06,
+      "loss": 0.622,
+      "step": 7392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4918719377919377e-06,
+      "loss": 0.6327,
+      "step": 7393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489963607874524e-06,
+      "loss": 0.6507,
+      "step": 7394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4880564010089327e-06,
+      "loss": 0.577,
+      "step": 7395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4861503174468517e-06,
+      "loss": 0.6582,
+      "step": 7396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.484245357439823e-06,
+      "loss": 0.6247,
+      "step": 7397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4823415212392378e-06,
+      "loss": 0.6989,
+      "step": 7398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.480438809096345e-06,
+      "loss": 0.6203,
+      "step": 7399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4785372212622407e-06,
+      "loss": 0.6862,
+      "step": 7400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4766367579878727e-06,
+      "loss": 0.6519,
+      "step": 7401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4747374195240416e-06,
+      "loss": 0.6852,
+      "step": 7402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4728392061213991e-06,
+      "loss": 0.6459,
+      "step": 7403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4709421180304484e-06,
+      "loss": 0.6286,
+      "step": 7404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4690461555015446e-06,
+      "loss": 0.5956,
+      "step": 7405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4671513187848928e-06,
+      "loss": 0.4867,
+      "step": 7406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4652576081305525e-06,
+      "loss": 0.5757,
+      "step": 7407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.463365023788429e-06,
+      "loss": 0.5503,
+      "step": 7408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4614735660082869e-06,
+      "loss": 0.6697,
+      "step": 7409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4595832350397377e-06,
+      "loss": 0.5996,
+      "step": 7410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4576940311322418e-06,
+      "loss": 0.6206,
+      "step": 7411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4558059545351144e-06,
+      "loss": 0.6203,
+      "step": 7412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4539190054975228e-06,
+      "loss": 0.5575,
+      "step": 7413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452033184268482e-06,
+      "loss": 0.5637,
+      "step": 7414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4501484910968611e-06,
+      "loss": 0.5757,
+      "step": 7415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4482649262313787e-06,
+      "loss": 0.6038,
+      "step": 7416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4463824899206047e-06,
+      "loss": 0.6352,
+      "step": 7417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4445011824129584e-06,
+      "loss": 0.66,
+      "step": 7418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4426210039567157e-06,
+      "loss": 0.5731,
+      "step": 7419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4407419547999978e-06,
+      "loss": 0.5758,
+      "step": 7420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4388640351907802e-06,
+      "loss": 0.6048,
+      "step": 7421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4369872453768862e-06,
+      "loss": 0.5247,
+      "step": 7422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4351115856059928e-06,
+      "loss": 0.5788,
+      "step": 7423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4332370561256271e-06,
+      "loss": 0.5888,
+      "step": 7424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4313636571831668e-06,
+      "loss": 0.6278,
+      "step": 7425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.42949138902584e-06,
+      "loss": 0.5646,
+      "step": 7426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4276202519007266e-06,
+      "loss": 0.6331,
+      "step": 7427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4257502460547544e-06,
+      "loss": 0.5717,
+      "step": 7428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4238813717347078e-06,
+      "loss": 0.5352,
+      "step": 7429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4220136291872166e-06,
+      "loss": 0.5807,
+      "step": 7430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4201470186587617e-06,
+      "loss": 0.6235,
+      "step": 7431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4182815403956785e-06,
+      "loss": 0.5871,
+      "step": 7432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4164171946441485e-06,
+      "loss": 0.6158,
+      "step": 7433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4145539816502062e-06,
+      "loss": 0.5866,
+      "step": 7434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4126919016597362e-06,
+      "loss": 0.7263,
+      "step": 7435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410830954918473e-06,
+      "loss": 0.6659,
+      "step": 7436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4089711416719987e-06,
+      "loss": 0.6904,
+      "step": 7437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407112462165754e-06,
+      "loss": 0.626,
+      "step": 7438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4052549166450224e-06,
+      "loss": 0.6631,
+      "step": 7439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4033985053549425e-06,
+      "loss": 0.6265,
+      "step": 7440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4015432285404984e-06,
+      "loss": 0.5374,
+      "step": 7441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3996890864465273e-06,
+      "loss": 0.6036,
+      "step": 7442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3978360793177181e-06,
+      "loss": 0.5954,
+      "step": 7443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3959842073986085e-06,
+      "loss": 0.4963,
+      "step": 7444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.394133470933584e-06,
+      "loss": 0.6271,
+      "step": 7445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3922838701668838e-06,
+      "loss": 0.5485,
+      "step": 7446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3904354053425927e-06,
+      "loss": 0.5585,
+      "step": 7447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3885880767046555e-06,
+      "loss": 0.5946,
+      "step": 7448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3867418844968527e-06,
+      "loss": 0.6442,
+      "step": 7449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3848968289628296e-06,
+      "loss": 0.5766,
+      "step": 7450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.383052910346071e-06,
+      "loss": 0.6416,
+      "step": 7451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.381210128889915e-06,
+      "loss": 0.6054,
+      "step": 7452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3793684848375499e-06,
+      "loss": 0.5739,
+      "step": 7453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3775279784320128e-06,
+      "loss": 0.5822,
+      "step": 7454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3756886099161926e-06,
+      "loss": 0.6404,
+      "step": 7455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3738503795328263e-06,
+      "loss": 0.6526,
+      "step": 7456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3720132875244995e-06,
+      "loss": 0.5782,
+      "step": 7457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3701773341336533e-06,
+      "loss": 0.5617,
+      "step": 7458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3683425196025734e-06,
+      "loss": 0.6062,
+      "step": 7459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3665088441733953e-06,
+      "loss": 0.5757,
+      "step": 7460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3646763080881065e-06,
+      "loss": 0.6231,
+      "step": 7461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3628449115885412e-06,
+      "loss": 0.6354,
+      "step": 7462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3610146549163871e-06,
+      "loss": 0.6375,
+      "step": 7463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359185538313179e-06,
+      "loss": 0.5848,
+      "step": 7464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3573575620202984e-06,
+      "loss": 0.5722,
+      "step": 7465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3555307262789841e-06,
+      "loss": 0.6818,
+      "step": 7466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3537050313303179e-06,
+      "loss": 0.6591,
+      "step": 7467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3518804774152305e-06,
+      "loss": 0.7816,
+      "step": 7468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3500570647745092e-06,
+      "loss": 0.6285,
+      "step": 7469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3482347936487838e-06,
+      "loss": 0.7128,
+      "step": 7470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3464136642785364e-06,
+      "loss": 0.6262,
+      "step": 7471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3445936769040969e-06,
+      "loss": 0.6782,
+      "step": 7472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3427748317656443e-06,
+      "loss": 0.613,
+      "step": 7473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3409571291032086e-06,
+      "loss": 0.6752,
+      "step": 7474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3391405691566684e-06,
+      "loss": 0.5101,
+      "step": 7475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3373251521657526e-06,
+      "loss": 0.6852,
+      "step": 7476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3355108783700332e-06,
+      "loss": 0.7001,
+      "step": 7477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3336977480089418e-06,
+      "loss": 0.561,
+      "step": 7478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3318857613217516e-06,
+      "loss": 0.6065,
+      "step": 7479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330074918547587e-06,
+      "loss": 0.6965,
+      "step": 7480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3282652199254197e-06,
+      "loss": 0.6054,
+      "step": 7481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3264566656940736e-06,
+      "loss": 0.6825,
+      "step": 7482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3246492560922154e-06,
+      "loss": 0.6636,
+      "step": 7483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3228429913583719e-06,
+      "loss": 0.6604,
+      "step": 7484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3210378717309092e-06,
+      "loss": 0.5851,
+      "step": 7485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3192338974480445e-06,
+      "loss": 0.719,
+      "step": 7486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3174310687478432e-06,
+      "loss": 0.6572,
+      "step": 7487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3156293858682246e-06,
+      "loss": 0.5414,
+      "step": 7488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3138288490469508e-06,
+      "loss": 0.6543,
+      "step": 7489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3120294585216353e-06,
+      "loss": 0.5982,
+      "step": 7490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3102312145297402e-06,
+      "loss": 0.6116,
+      "step": 7491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3084341173085746e-06,
+      "loss": 0.5134,
+      "step": 7492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066381670952999e-06,
+      "loss": 0.642,
+      "step": 7493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3048433641269221e-06,
+      "loss": 0.6458,
+      "step": 7494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030497086402993e-06,
+      "loss": 0.5922,
+      "step": 7495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012572008721325e-06,
+      "loss": 0.6318,
+      "step": 7496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29946584105898e-06,
+      "loss": 0.6483,
+      "step": 7497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976756294372427e-06,
+      "loss": 0.6064,
+      "step": 7498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958865662431697e-06,
+      "loss": 0.6211,
+      "step": 7499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2940986517128618e-06,
+      "loss": 0.6832,
+      "step": 7500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2923118860822647e-06,
+      "loss": 0.5259,
+      "step": 7501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2905262695871722e-06,
+      "loss": 0.5452,
+      "step": 7502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2887418024632326e-06,
+      "loss": 0.6454,
+      "step": 7503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2869584849459383e-06,
+      "loss": 0.6128,
+      "step": 7504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2851763172706266e-06,
+      "loss": 0.6034,
+      "step": 7505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2833952996724864e-06,
+      "loss": 0.669,
+      "step": 7506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2816154323865594e-06,
+      "loss": 0.5755,
+      "step": 7507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2798367156477277e-06,
+      "loss": 0.569,
+      "step": 7508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2780591496907258e-06,
+      "loss": 0.5643,
+      "step": 7509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276282734750135e-06,
+      "loss": 0.5862,
+      "step": 7510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2745074710603855e-06,
+      "loss": 0.5401,
+      "step": 7511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2727333588557544e-06,
+      "loss": 0.5322,
+      "step": 7512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2709603983703678e-06,
+      "loss": 0.5991,
+      "step": 7513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2691885898381995e-06,
+      "loss": 0.5119,
+      "step": 7514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2674179334930715e-06,
+      "loss": 0.6796,
+      "step": 7515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656484295686522e-06,
+      "loss": 0.5593,
+      "step": 7516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2638800782984629e-06,
+      "loss": 0.6881,
+      "step": 7517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2621128799158677e-06,
+      "loss": 0.6697,
+      "step": 7518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603468346540794e-06,
+      "loss": 0.5963,
+      "step": 7519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2585819427461564e-06,
+      "loss": 0.6561,
+      "step": 7520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2568182044250144e-06,
+      "loss": 0.6698,
+      "step": 7521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2550556199234064e-06,
+      "loss": 0.6262,
+      "step": 7522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2532941894739381e-06,
+      "loss": 0.6339,
+      "step": 7523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251533913309061e-06,
+      "loss": 0.5873,
+      "step": 7524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2497747916610747e-06,
+      "loss": 0.6192,
+      "step": 7525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2480168247621261e-06,
+      "loss": 0.6253,
+      "step": 7526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2462600128442136e-06,
+      "loss": 0.5783,
+      "step": 7527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244504356139179e-06,
+      "loss": 0.6566,
+      "step": 7528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.242749854878711e-06,
+      "loss": 0.6595,
+      "step": 7529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2409965092943488e-06,
+      "loss": 0.5715,
+      "step": 7530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392443196174763e-06,
+      "loss": 0.5624,
+      "step": 7531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2374932860793287e-06,
+      "loss": 0.5893,
+      "step": 7532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2357434089109844e-06,
+      "loss": 0.5946,
+      "step": 7533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2339946883433719e-06,
+      "loss": 0.6307,
+      "step": 7534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2322471246072653e-06,
+      "loss": 0.519,
+      "step": 7535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2305007179332851e-06,
+      "loss": 0.6261,
+      "step": 7536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2287554685519066e-06,
+      "loss": 0.703,
+      "step": 7537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2270113766934433e-06,
+      "loss": 0.6952,
+      "step": 7538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2252684425880567e-06,
+      "loss": 0.6535,
+      "step": 7539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2235266664657641e-06,
+      "loss": 0.6595,
+      "step": 7540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2217860485564203e-06,
+      "loss": 0.569,
+      "step": 7541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2200465890897329e-06,
+      "loss": 0.6402,
+      "step": 7542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2183082882952535e-06,
+      "loss": 0.5657,
+      "step": 7543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216571146402381e-06,
+      "loss": 0.6397,
+      "step": 7544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2148351636403643e-06,
+      "loss": 0.6749,
+      "step": 7545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2131003402382956e-06,
+      "loss": 0.6031,
+      "step": 7546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2113666764251186e-06,
+      "loss": 0.5204,
+      "step": 7547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209634172429619e-06,
+      "loss": 0.6477,
+      "step": 7548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.207902828480434e-06,
+      "loss": 0.6643,
+      "step": 7549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.206172644806043e-06,
+      "loss": 0.6287,
+      "step": 7550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2044436216347754e-06,
+      "loss": 0.6088,
+      "step": 7551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.202715759194808e-06,
+      "loss": 0.5477,
+      "step": 7552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2009890577141625e-06,
+      "loss": 0.6957,
+      "step": 7553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.199263517420708e-06,
+      "loss": 0.5577,
+      "step": 7554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1975391385421575e-06,
+      "loss": 0.6694,
+      "step": 7555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1958159213060793e-06,
+      "loss": 0.5924,
+      "step": 7556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1940938659398781e-06,
+      "loss": 0.6031,
+      "step": 7557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1923729726708143e-06,
+      "loss": 0.5986,
+      "step": 7558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1906532417259885e-06,
+      "loss": 0.5981,
+      "step": 7559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1889346733323493e-06,
+      "loss": 0.5937,
+      "step": 7560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1872172677166937e-06,
+      "loss": 0.46,
+      "step": 7561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185501025105663e-06,
+      "loss": 0.6897,
+      "step": 7562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1837859457257473e-06,
+      "loss": 0.6977,
+      "step": 7563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1820720298032807e-06,
+      "loss": 0.6981,
+      "step": 7564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1803592775644445e-06,
+      "loss": 0.529,
+      "step": 7565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1786476892352705e-06,
+      "loss": 0.5746,
+      "step": 7566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1769372650416322e-06,
+      "loss": 0.6432,
+      "step": 7567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1752280052092492e-06,
+      "loss": 0.6489,
+      "step": 7568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1735199099636897e-06,
+      "loss": 0.6218,
+      "step": 7569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.171812979530368e-06,
+      "loss": 0.5915,
+      "step": 7570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1701072141345427e-06,
+      "loss": 0.5868,
+      "step": 7571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1684026140013227e-06,
+      "loss": 0.6144,
+      "step": 7572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1666991793556559e-06,
+      "loss": 0.5892,
+      "step": 7573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1649969104223457e-06,
+      "loss": 0.6445,
+      "step": 7574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.163295807426037e-06,
+      "loss": 0.6014,
+      "step": 7575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1615958705912155e-06,
+      "loss": 0.6631,
+      "step": 7576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1598971001422244e-06,
+      "loss": 0.6342,
+      "step": 7577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158199496303245e-06,
+      "loss": 0.6584,
+      "step": 7578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1565030592983062e-06,
+      "loss": 0.6794,
+      "step": 7579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1548077893512833e-06,
+      "loss": 0.6625,
+      "step": 7580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1531136866858971e-06,
+      "loss": 0.6636,
+      "step": 7581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1514207515257147e-06,
+      "loss": 0.6224,
+      "step": 7582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1497289840941506e-06,
+      "loss": 0.6444,
+      "step": 7583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1480383846144627e-06,
+      "loss": 0.7088,
+      "step": 7584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463489533097538e-06,
+      "loss": 0.6812,
+      "step": 7585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1446606904029788e-06,
+      "loss": 0.6533,
+      "step": 7586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1429735961169319e-06,
+      "loss": 0.6649,
+      "step": 7587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141287670674257e-06,
+      "loss": 0.6043,
+      "step": 7588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1396029142974408e-06,
+      "loss": 0.4602,
+      "step": 7589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1379193272088174e-06,
+      "loss": 0.6578,
+      "step": 7590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1362369096305658e-06,
+      "loss": 0.5973,
+      "step": 7591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1345556617847099e-06,
+      "loss": 0.5581,
+      "step": 7592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1328755838931239e-06,
+      "loss": 0.6049,
+      "step": 7593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1311966761775227e-06,
+      "loss": 0.576,
+      "step": 7594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1295189388594664e-06,
+      "loss": 0.6298,
+      "step": 7595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.127842372160366e-06,
+      "loss": 0.6431,
+      "step": 7596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1261669763014725e-06,
+      "loss": 0.5952,
+      "step": 7597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1244927515038851e-06,
+      "loss": 0.6024,
+      "step": 7598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1228196979885474e-06,
+      "loss": 0.6817,
+      "step": 7599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.121147815976248e-06,
+      "loss": 0.6454,
+      "step": 7600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.119477105687623e-06,
+      "loss": 0.4946,
+      "step": 7601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117807567343152e-06,
+      "loss": 0.5886,
+      "step": 7602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.11613920116316e-06,
+      "loss": 0.6635,
+      "step": 7603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1144720073678184e-06,
+      "loss": 0.6764,
+      "step": 7604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.11280598617714e-06,
+      "loss": 0.6088,
+      "step": 7605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1111411378109926e-06,
+      "loss": 0.5762,
+      "step": 7606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1094774624890793e-06,
+      "loss": 0.5989,
+      "step": 7607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1078149604309518e-06,
+      "loss": 0.6069,
+      "step": 7608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1061536318560085e-06,
+      "loss": 0.6342,
+      "step": 7609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.104493476983487e-06,
+      "loss": 0.6899,
+      "step": 7610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10283449603248e-06,
+      "loss": 0.6392,
+      "step": 7611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1011766892219177e-06,
+      "loss": 0.6712,
+      "step": 7612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0995200567705765e-06,
+      "loss": 0.6163,
+      "step": 7613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0978645988970793e-06,
+      "loss": 0.6673,
+      "step": 7614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0962103158198945e-06,
+      "loss": 0.6321,
+      "step": 7615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094557207757333e-06,
+      "loss": 0.6839,
+      "step": 7616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0929052749275536e-06,
+      "loss": 0.6608,
+      "step": 7617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0912545175485567e-06,
+      "loss": 0.6451,
+      "step": 7618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0896049358381911e-06,
+      "loss": 0.5454,
+      "step": 7619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.087956530014146e-06,
+      "loss": 0.6495,
+      "step": 7620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0863093002939606e-06,
+      "loss": 0.6052,
+      "step": 7621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0846632468950157e-06,
+      "loss": 0.6686,
+      "step": 7622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0830183700345375e-06,
+      "loss": 0.5546,
+      "step": 7623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0813746699295946e-06,
+      "loss": 0.667,
+      "step": 7624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0797321467971067e-06,
+      "loss": 0.6276,
+      "step": 7625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0780908008538327e-06,
+      "loss": 0.6326,
+      "step": 7626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076450632316377e-06,
+      "loss": 0.6124,
+      "step": 7627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.074811641401189e-06,
+      "loss": 0.6244,
+      "step": 7628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073173828324562e-06,
+      "loss": 0.6619,
+      "step": 7629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0715371933026387e-06,
+      "loss": 0.6512,
+      "step": 7630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0699017365513997e-06,
+      "loss": 0.6395,
+      "step": 7631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0682674582866725e-06,
+      "loss": 0.6409,
+      "step": 7632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0666343587241302e-06,
+      "loss": 0.5386,
+      "step": 7633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.065002438079288e-06,
+      "loss": 0.6515,
+      "step": 7634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06337169656751e-06,
+      "loss": 0.5573,
+      "step": 7635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0617421344040003e-06,
+      "loss": 0.534,
+      "step": 7636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0601137518038085e-06,
+      "loss": 0.6021,
+      "step": 7637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0584865489818285e-06,
+      "loss": 0.5361,
+      "step": 7638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0568605261528009e-06,
+      "loss": 0.5808,
+      "step": 7639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055235683531306e-06,
+      "loss": 0.6116,
+      "step": 7640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0536120213317725e-06,
+      "loss": 0.6237,
+      "step": 7641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051989539768471e-06,
+      "loss": 0.6132,
+      "step": 7642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0503682390555182e-06,
+      "loss": 0.643,
+      "step": 7643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0487481194068693e-06,
+      "loss": 0.6073,
+      "step": 7644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.047129181036335e-06,
+      "loss": 0.6045,
+      "step": 7645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0455114241575593e-06,
+      "loss": 0.6516,
+      "step": 7646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0438948489840327e-06,
+      "loss": 0.6953,
+      "step": 7647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042279455729096e-06,
+      "loss": 0.5823,
+      "step": 7648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0406652446059263e-06,
+      "loss": 0.7101,
+      "step": 7649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0390522158275483e-06,
+      "loss": 0.5821,
+      "step": 7650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0374403696068292e-06,
+      "loss": 0.5693,
+      "step": 7651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0358297061564826e-06,
+      "loss": 0.6053,
+      "step": 7652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0342202256890632e-06,
+      "loss": 0.5286,
+      "step": 7653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0326119284169688e-06,
+      "loss": 0.6614,
+      "step": 7654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0310048145524476e-06,
+      "loss": 0.5738,
+      "step": 7655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0293988843075842e-06,
+      "loss": 0.6557,
+      "step": 7656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0277941378943102e-06,
+      "loss": 0.5427,
+      "step": 7657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0261905755244017e-06,
+      "loss": 0.6402,
+      "step": 7658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0245881974094762e-06,
+      "loss": 0.6616,
+      "step": 7659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0229870037609968e-06,
+      "loss": 0.6258,
+      "step": 7660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.02138699479027e-06,
+      "loss": 0.6391,
+      "step": 7661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0197881707084444e-06,
+      "loss": 0.5479,
+      "step": 7662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0181905317265128e-06,
+      "loss": 0.6937,
+      "step": 7663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165940780553164e-06,
+      "loss": 0.5978,
+      "step": 7664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0149988099055308e-06,
+      "loss": 0.56,
+      "step": 7665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0134047274876845e-06,
+      "loss": 0.7753,
+      "step": 7666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011811831012144e-06,
+      "loss": 0.5281,
+      "step": 7667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0102201206891193e-06,
+      "loss": 0.6877,
+      "step": 7668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0086295967286664e-06,
+      "loss": 0.6557,
+      "step": 7669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0070402593406835e-06,
+      "loss": 0.5971,
+      "step": 7670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.00545210873491e-06,
+      "loss": 0.6862,
+      "step": 7671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0038651451209336e-06,
+      "loss": 0.5694,
+      "step": 7672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0022793687081789e-06,
+      "loss": 0.6255,
+      "step": 7673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000694779705922e-06,
+      "loss": 0.614,
+      "step": 7674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.991113783232753e-07,
+      "loss": 0.5736,
+      "step": 7675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.975291647691988e-07,
+      "loss": 0.6815,
+      "step": 7676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.959481392524907e-07,
+      "loss": 0.6524,
+      "step": 7677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.943683019817985e-07,
+      "loss": 0.6012,
+      "step": 7678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9278965316561e-07,
+      "loss": 0.603,
+      "step": 7679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.912121930122542e-07,
+      "loss": 0.6917,
+      "step": 7680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.896359217299046e-07,
+      "loss": 0.7357,
+      "step": 7681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.880608395265822e-07,
+      "loss": 0.6998,
+      "step": 7682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.864869466101457e-07,
+      "loss": 0.6173,
+      "step": 7683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.849142431882952e-07,
+      "loss": 0.6754,
+      "step": 7684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.833427294685827e-07,
+      "loss": 0.6519,
+      "step": 7685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.817724056583943e-07,
+      "loss": 0.6223,
+      "step": 7686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.802032719649635e-07,
+      "loss": 0.5441,
+      "step": 7687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.786353285953654e-07,
+      "loss": 0.6548,
+      "step": 7688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.770685757565169e-07,
+      "loss": 0.6057,
+      "step": 7689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.755030136551813e-07,
+      "loss": 0.6222,
+      "step": 7690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739386424979602e-07,
+      "loss": 0.5687,
+      "step": 7691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.723754624913017e-07,
+      "loss": 0.5778,
+      "step": 7692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.708134738414944e-07,
+      "loss": 0.6752,
+      "step": 7693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.692526767546727e-07,
+      "loss": 0.6636,
+      "step": 7694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.676930714368115e-07,
+      "loss": 0.6633,
+      "step": 7695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66134658093727e-07,
+      "loss": 0.6647,
+      "step": 7696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.645774369310812e-07,
+      "loss": 0.6232,
+      "step": 7697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.630214081543765e-07,
+      "loss": 0.6371,
+      "step": 7698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.614665719689597e-07,
+      "loss": 0.653,
+      "step": 7699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.599129285800158e-07,
+      "loss": 0.5993,
+      "step": 7700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.583604781925804e-07,
+      "loss": 0.6011,
+      "step": 7701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.568092210115264e-07,
+      "loss": 0.6668,
+      "step": 7702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55259157241567e-07,
+      "loss": 0.6188,
+      "step": 7703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.53710287087265e-07,
+      "loss": 0.6006,
+      "step": 7704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.52162610753019e-07,
+      "loss": 0.6372,
+      "step": 7705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.506161284430737e-07,
+      "loss": 0.6813,
+      "step": 7706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.490708403615146e-07,
+      "loss": 0.7095,
+      "step": 7707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475267467122695e-07,
+      "loss": 0.5982,
+      "step": 7708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.459838476991101e-07,
+      "loss": 0.6605,
+      "step": 7709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444421435256501e-07,
+      "loss": 0.6932,
+      "step": 7710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429016343953434e-07,
+      "loss": 0.6207,
+      "step": 7711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.413623205114896e-07,
+      "loss": 0.6509,
+      "step": 7712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.398242020772253e-07,
+      "loss": 0.6919,
+      "step": 7713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.382872792955377e-07,
+      "loss": 0.6332,
+      "step": 7714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.367515523692483e-07,
+      "loss": 0.5665,
+      "step": 7715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.352170215010248e-07,
+      "loss": 0.6754,
+      "step": 7716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336836868933751e-07,
+      "loss": 0.671,
+      "step": 7717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.321515487486499e-07,
+      "loss": 0.6359,
+      "step": 7718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.306206072690449e-07,
+      "loss": 0.6531,
+      "step": 7719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290908626565931e-07,
+      "loss": 0.5276,
+      "step": 7720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.275623151131719e-07,
+      "loss": 0.6407,
+      "step": 7721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.260349648404998e-07,
+      "loss": 0.6213,
+      "step": 7722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245088120401402e-07,
+      "loss": 0.6201,
+      "step": 7723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.229838569134953e-07,
+      "loss": 0.6452,
+      "step": 7724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.214600996618095e-07,
+      "loss": 0.6132,
+      "step": 7725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.19937540486171e-07,
+      "loss": 0.5982,
+      "step": 7726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.18416179587508e-07,
+      "loss": 0.5638,
+      "step": 7727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.168960171665919e-07,
+      "loss": 0.6758,
+      "step": 7728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.153770534240347e-07,
+      "loss": 0.584,
+      "step": 7729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.138592885602915e-07,
+      "loss": 0.6076,
+      "step": 7730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.123427227756587e-07,
+      "loss": 0.652,
+      "step": 7731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.108273562702719e-07,
+      "loss": 0.7,
+      "step": 7732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.093131892441143e-07,
+      "loss": 0.5884,
+      "step": 7733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.078002218970062e-07,
+      "loss": 0.6828,
+      "step": 7734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.06288454428611e-07,
+      "loss": 0.6505,
+      "step": 7735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.047778870384327e-07,
+      "loss": 0.6096,
+      "step": 7736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.032685199258163e-07,
+      "loss": 0.5591,
+      "step": 7737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.017603532899533e-07,
+      "loss": 0.6242,
+      "step": 7738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.002533873298724e-07,
+      "loss": 0.5213,
+      "step": 7739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.987476222444436e-07,
+      "loss": 0.6005,
+      "step": 7740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.972430582323788e-07,
+      "loss": 0.6152,
+      "step": 7741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957396954922326e-07,
+      "loss": 0.536,
+      "step": 7742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.942375342224041e-07,
+      "loss": 0.6053,
+      "step": 7743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.92736574621127e-07,
+      "loss": 0.6682,
+      "step": 7744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.912368168864804e-07,
+      "loss": 0.639,
+      "step": 7745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.897382612163852e-07,
+      "loss": 0.6395,
+      "step": 7746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.882409078086007e-07,
+      "loss": 0.6575,
+      "step": 7747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.867447568607312e-07,
+      "loss": 0.6252,
+      "step": 7748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.852498085702199e-07,
+      "loss": 0.6087,
+      "step": 7749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.837560631343534e-07,
+      "loss": 0.6591,
+      "step": 7750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.822635207502561e-07,
+      "loss": 0.6396,
+      "step": 7751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.807721816148939e-07,
+      "loss": 0.5977,
+      "step": 7752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.792820459250817e-07,
+      "loss": 0.5573,
+      "step": 7753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.777931138774665e-07,
+      "loss": 0.5773,
+      "step": 7754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.763053856685367e-07,
+      "loss": 0.6438,
+      "step": 7755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.748188614946295e-07,
+      "loss": 0.6781,
+      "step": 7756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.733335415519162e-07,
+      "loss": 0.5885,
+      "step": 7757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.718494260364118e-07,
+      "loss": 0.5698,
+      "step": 7758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703665151439722e-07,
+      "loss": 0.5914,
+      "step": 7759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.688848090702928e-07,
+      "loss": 0.5974,
+      "step": 7760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.674043080109129e-07,
+      "loss": 0.5492,
+      "step": 7761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.659250121612072e-07,
+      "loss": 0.683,
+      "step": 7762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.644469217164009e-07,
+      "loss": 0.6523,
+      "step": 7763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629700368715521e-07,
+      "loss": 0.6315,
+      "step": 7764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.614943578215618e-07,
+      "loss": 0.7127,
+      "step": 7765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.60019884761173e-07,
+      "loss": 0.6436,
+      "step": 7766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.585466178849688e-07,
+      "loss": 0.5632,
+      "step": 7767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.570745573873717e-07,
+      "loss": 0.5853,
+      "step": 7768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.556037034626485e-07,
+      "loss": 0.6406,
+      "step": 7769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.541340563049039e-07,
+      "loss": 0.6184,
+      "step": 7770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.526656161080815e-07,
+      "loss": 0.6243,
+      "step": 7771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.511983830659731e-07,
+      "loss": 0.7034,
+      "step": 7772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.497323573722038e-07,
+      "loss": 0.5618,
+      "step": 7773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.48267539220241e-07,
+      "loss": 0.5848,
+      "step": 7774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.46803928803398e-07,
+      "loss": 0.5278,
+      "step": 7775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.453415263148202e-07,
+      "loss": 0.6526,
+      "step": 7776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438803319474997e-07,
+      "loss": 0.638,
+      "step": 7777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.424203458942659e-07,
+      "loss": 0.5273,
+      "step": 7778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.409615683477912e-07,
+      "loss": 0.6011,
+      "step": 7779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.395039995005871e-07,
+      "loss": 0.6571,
+      "step": 7780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.380476395450054e-07,
+      "loss": 0.6247,
+      "step": 7781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.365924886732401e-07,
+      "loss": 0.6435,
+      "step": 7782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.351385470773243e-07,
+      "loss": 0.5715,
+      "step": 7783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.336858149491312e-07,
+      "loss": 0.6384,
+      "step": 7784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.32234292480375e-07,
+      "loss": 0.5624,
+      "step": 7785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.307839798626094e-07,
+      "loss": 0.6457,
+      "step": 7786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.293348772872301e-07,
+      "loss": 0.6314,
+      "step": 7787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.278869849454718e-07,
+      "loss": 0.5614,
+      "step": 7788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.264403030284074e-07,
+      "loss": 0.5131,
+      "step": 7789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.249948317269574e-07,
+      "loss": 0.6086,
+      "step": 7790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.235505712318747e-07,
+      "loss": 0.6649,
+      "step": 7791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.221075217337537e-07,
+      "loss": 0.6265,
+      "step": 7792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.206656834230353e-07,
+      "loss": 0.6546,
+      "step": 7793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.192250564899929e-07,
+      "loss": 0.5962,
+      "step": 7794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.177856411247443e-07,
+      "loss": 0.6365,
+      "step": 7795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.163474375172442e-07,
+      "loss": 0.6347,
+      "step": 7796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.14910445857291e-07,
+      "loss": 0.6462,
+      "step": 7797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.134746663345206e-07,
+      "loss": 0.6234,
+      "step": 7798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.120400991384103e-07,
+      "loss": 0.6272,
+      "step": 7799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.106067444582777e-07,
+      "loss": 0.5617,
+      "step": 7800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.091746024832758e-07,
+      "loss": 0.6671,
+      "step": 7801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.077436734024069e-07,
+      "loss": 0.5539,
+      "step": 7802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.063139574045054e-07,
+      "loss": 0.6621,
+      "step": 7803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.048854546782481e-07,
+      "loss": 0.5906,
+      "step": 7804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.03458165412151e-07,
+      "loss": 0.6107,
+      "step": 7805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.020320897945699e-07,
+      "loss": 0.5823,
+      "step": 7806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.006072280137023e-07,
+      "loss": 0.6359,
+      "step": 7807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.99183580257582e-07,
+      "loss": 0.6292,
+      "step": 7808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.977611467140889e-07,
+      "loss": 0.6714,
+      "step": 7809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.96339927570936e-07,
+      "loss": 0.6637,
+      "step": 7810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.949199230156779e-07,
+      "loss": 0.5995,
+      "step": 7811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.935011332357113e-07,
+      "loss": 0.5795,
+      "step": 7812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.920835584182718e-07,
+      "loss": 0.6756,
+      "step": 7813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.906671987504311e-07,
+      "loss": 0.5663,
+      "step": 7814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.892520544191052e-07,
+      "loss": 0.6054,
+      "step": 7815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.878381256110467e-07,
+      "loss": 0.7245,
+      "step": 7816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.864254125128501e-07,
+      "loss": 0.6532,
+      "step": 7817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85013915310947e-07,
+      "loss": 0.6372,
+      "step": 7818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.836036341916098e-07,
+      "loss": 0.5786,
+      "step": 7819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.821945693409506e-07,
+      "loss": 0.5855,
+      "step": 7820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.807867209449194e-07,
+      "loss": 0.6661,
+      "step": 7821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.793800891893099e-07,
+      "loss": 0.6306,
+      "step": 7822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779746742597505e-07,
+      "loss": 0.6376,
+      "step": 7823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.765704763417125e-07,
+      "loss": 0.6818,
+      "step": 7824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.751674956205024e-07,
+      "loss": 0.5803,
+      "step": 7825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.737657322812697e-07,
+      "loss": 0.5959,
+      "step": 7826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.723651865090032e-07,
+      "loss": 0.6267,
+      "step": 7827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.709658584885304e-07,
+      "loss": 0.6571,
+      "step": 7828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.695677484045161e-07,
+      "loss": 0.5946,
+      "step": 7829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.681708564414658e-07,
+      "loss": 0.5505,
+      "step": 7830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.667751827837244e-07,
+      "loss": 0.7138,
+      "step": 7831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.65380727615479e-07,
+      "loss": 0.6377,
+      "step": 7832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.639874911207501e-07,
+      "loss": 0.6162,
+      "step": 7833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.625954734834006e-07,
+      "loss": 0.601,
+      "step": 7834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.5474,
+      "step": 7835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.59815095515487e-07,
+      "loss": 0.5487,
+      "step": 7836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584267355518438e-07,
+      "loss": 0.6379,
+      "step": 7837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.570395951794219e-07,
+      "loss": 0.7005,
+      "step": 7838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.556536745812782e-07,
+      "loss": 0.6093,
+      "step": 7839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.542689739403097e-07,
+      "loss": 0.6018,
+      "step": 7840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.528854934392549e-07,
+      "loss": 0.5927,
+      "step": 7841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.515032332606865e-07,
+      "loss": 0.5865,
+      "step": 7842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.501221935870207e-07,
+      "loss": 0.5628,
+      "step": 7843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.487423746005085e-07,
+      "loss": 0.6068,
+      "step": 7844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.473637764832409e-07,
+      "loss": 0.6142,
+      "step": 7845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.459863994171513e-07,
+      "loss": 0.5876,
+      "step": 7846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.446102435840086e-07,
+      "loss": 0.6651,
+      "step": 7847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.43235309165421e-07,
+      "loss": 0.6109,
+      "step": 7848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.418615963428344e-07,
+      "loss": 0.6425,
+      "step": 7849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40489105297535e-07,
+      "loss": 0.6115,
+      "step": 7850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3911783621065e-07,
+      "loss": 0.6276,
+      "step": 7851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.377477892631412e-07,
+      "loss": 0.7141,
+      "step": 7852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363789646358111e-07,
+      "loss": 0.6751,
+      "step": 7853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.350113625093002e-07,
+      "loss": 0.5983,
+      "step": 7854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.33644983064089e-07,
+      "loss": 0.6344,
+      "step": 7855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322798264804954e-07,
+      "loss": 0.605,
+      "step": 7856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.309158929386762e-07,
+      "loss": 0.5622,
+      "step": 7857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.295531826186264e-07,
+      "loss": 0.5625,
+      "step": 7858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281916957001811e-07,
+      "loss": 0.6287,
+      "step": 7859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.268314323630099e-07,
+      "loss": 0.5772,
+      "step": 7860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.254723927866281e-07,
+      "loss": 0.5834,
+      "step": 7861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.241145771503832e-07,
+      "loss": 0.757,
+      "step": 7862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.227579856334621e-07,
+      "loss": 0.5806,
+      "step": 7863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.214026184148948e-07,
+      "loss": 0.6981,
+      "step": 7864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.200484756735449e-07,
+      "loss": 0.6287,
+      "step": 7865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.186955575881138e-07,
+      "loss": 0.5788,
+      "step": 7866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.173438643371466e-07,
+      "loss": 0.5962,
+      "step": 7867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.159933960990207e-07,
+      "loss": 0.6431,
+      "step": 7868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146441530519555e-07,
+      "loss": 0.6129,
+      "step": 7869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.132961353740054e-07,
+      "loss": 0.6291,
+      "step": 7870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119493432430702e-07,
+      "loss": 0.6377,
+      "step": 7871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.106037768368812e-07,
+      "loss": 0.5567,
+      "step": 7872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092594363330097e-07,
+      "loss": 0.5591,
+      "step": 7873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079163219088659e-07,
+      "loss": 0.6688,
+      "step": 7874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.065744337416968e-07,
+      "loss": 0.5962,
+      "step": 7875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0523377200859e-07,
+      "loss": 0.5493,
+      "step": 7876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.038943368864681e-07,
+      "loss": 0.5473,
+      "step": 7877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025561285520954e-07,
+      "loss": 0.6693,
+      "step": 7878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.012191471820706e-07,
+      "loss": 0.5604,
+      "step": 7879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.998833929528337e-07,
+      "loss": 0.6368,
+      "step": 7880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.985488660406625e-07,
+      "loss": 0.5533,
+      "step": 7881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.972155666216684e-07,
+      "loss": 0.6736,
+      "step": 7882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.958834948718074e-07,
+      "loss": 0.5877,
+      "step": 7883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.945526509668688e-07,
+      "loss": 0.6982,
+      "step": 7884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.93223035082482e-07,
+      "loss": 0.5854,
+      "step": 7885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.918946473941124e-07,
+      "loss": 0.5975,
+      "step": 7886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.905674880770664e-07,
+      "loss": 0.6468,
+      "step": 7887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892415573064848e-07,
+      "loss": 0.5917,
+      "step": 7888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.879168552573478e-07,
+      "loss": 0.6606,
+      "step": 7889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.865933821044734e-07,
+      "loss": 0.6308,
+      "step": 7890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.852711380225197e-07,
+      "loss": 0.6362,
+      "step": 7891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839501231859802e-07,
+      "loss": 0.598,
+      "step": 7892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.826303377691846e-07,
+      "loss": 0.5947,
+      "step": 7893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.813117819463033e-07,
+      "loss": 0.5563,
+      "step": 7894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.79994455891344e-07,
+      "loss": 0.5781,
+      "step": 7895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.786783597781499e-07,
+      "loss": 0.6495,
+      "step": 7896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.77363493780403e-07,
+      "loss": 0.6976,
+      "step": 7897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.760498580716268e-07,
+      "loss": 0.6111,
+      "step": 7898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.74737452825176e-07,
+      "loss": 0.6886,
+      "step": 7899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.734262782142464e-07,
+      "loss": 0.6459,
+      "step": 7900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.721163344118731e-07,
+      "loss": 0.6431,
+      "step": 7901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.708076215909243e-07,
+      "loss": 0.6153,
+      "step": 7902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.695001399241096e-07,
+      "loss": 0.665,
+      "step": 7903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.681938895839746e-07,
+      "loss": 0.6425,
+      "step": 7904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.668888707429011e-07,
+      "loss": 0.6472,
+      "step": 7905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.655850835731104e-07,
+      "loss": 0.628,
+      "step": 7906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.642825282466614e-07,
+      "loss": 0.52,
+      "step": 7907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629812049354478e-07,
+      "loss": 0.6129,
+      "step": 7908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.616811138112034e-07,
+      "loss": 0.5979,
+      "step": 7909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.603822550454997e-07,
+      "loss": 0.6236,
+      "step": 7910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.59084628809743e-07,
+      "loss": 0.5693,
+      "step": 7911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.577882352751785e-07,
+      "loss": 0.5709,
+      "step": 7912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.564930746128895e-07,
+      "loss": 0.6474,
+      "step": 7913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.551991469937958e-07,
+      "loss": 0.5892,
+      "step": 7914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.539064525886518e-07,
+      "loss": 0.5346,
+      "step": 7915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.526149915680535e-07,
+      "loss": 0.6285,
+      "step": 7916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.513247641024323e-07,
+      "loss": 0.6838,
+      "step": 7917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.500357703620586e-07,
+      "loss": 0.6221,
+      "step": 7918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.48748010517033e-07,
+      "loss": 0.5918,
+      "step": 7919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.474614847373051e-07,
+      "loss": 0.7035,
+      "step": 7920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.461761931926502e-07,
+      "loss": 0.6551,
+      "step": 7921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448921360526883e-07,
+      "loss": 0.6501,
+      "step": 7922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436093134868726e-07,
+      "loss": 0.5925,
+      "step": 7923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.423277256644944e-07,
+      "loss": 0.7326,
+      "step": 7924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.410473727546828e-07,
+      "loss": 0.6425,
+      "step": 7925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.397682549264028e-07,
+      "loss": 0.6451,
+      "step": 7926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.384903723484559e-07,
+      "loss": 0.5713,
+      "step": 7927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.372137251894839e-07,
+      "loss": 0.5918,
+      "step": 7928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.359383136179598e-07,
+      "loss": 0.5111,
+      "step": 7929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346641378022012e-07,
+      "loss": 0.5302,
+      "step": 7930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.333911979103569e-07,
+      "loss": 0.5783,
+      "step": 7931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.321194941104136e-07,
+      "loss": 0.6611,
+      "step": 7932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.30849026570195e-07,
+      "loss": 0.6353,
+      "step": 7933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.295797954573612e-07,
+      "loss": 0.5856,
+      "step": 7934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.28311800939414e-07,
+      "loss": 0.6963,
+      "step": 7935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270450431836839e-07,
+      "loss": 0.5834,
+      "step": 7936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.257795223573448e-07,
+      "loss": 0.5857,
+      "step": 7937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245152386274034e-07,
+      "loss": 0.6195,
+      "step": 7938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232521921607038e-07,
+      "loss": 0.6558,
+      "step": 7939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.219903831239304e-07,
+      "loss": 0.7008,
+      "step": 7940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.207298116836002e-07,
+      "loss": 0.648,
+      "step": 7941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194704780060678e-07,
+      "loss": 0.6997,
+      "step": 7942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.182123822575258e-07,
+      "loss": 0.5721,
+      "step": 7943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.169555246040005e-07,
+      "loss": 0.584,
+      "step": 7944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.156999052113588e-07,
+      "loss": 0.6787,
+      "step": 7945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144455242453007e-07,
+      "loss": 0.6452,
+      "step": 7946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.131923818713659e-07,
+      "loss": 0.6394,
+      "step": 7947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.119404782549276e-07,
+      "loss": 0.6978,
+      "step": 7948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106898135611961e-07,
+      "loss": 0.727,
+      "step": 7949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.094403879552213e-07,
+      "loss": 0.6009,
+      "step": 7950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08192201601886e-07,
+      "loss": 0.5813,
+      "step": 7951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.069452546659116e-07,
+      "loss": 0.5943,
+      "step": 7952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056995473118532e-07,
+      "loss": 0.5444,
+      "step": 7953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.044550797041071e-07,
+      "loss": 0.6315,
+      "step": 7954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.032118520069008e-07,
+      "loss": 0.5622,
+      "step": 7955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.01969864384302e-07,
+      "loss": 0.6775,
+      "step": 7956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.007291170002116e-07,
+      "loss": 0.638,
+      "step": 7957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.994896100183689e-07,
+      "loss": 0.6179,
+      "step": 7958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.982513436023507e-07,
+      "loss": 0.6123,
+      "step": 7959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970143179155675e-07,
+      "loss": 0.6378,
+      "step": 7960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.957785331212662e-07,
+      "loss": 0.5833,
+      "step": 7961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.945439893825311e-07,
+      "loss": 0.6084,
+      "step": 7962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933106868622828e-07,
+      "loss": 0.5804,
+      "step": 7963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.920786257232769e-07,
+      "loss": 0.5937,
+      "step": 7964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.908478061281076e-07,
+      "loss": 0.6716,
+      "step": 7965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896182282392016e-07,
+      "loss": 0.6238,
+      "step": 7966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.883898922188247e-07,
+      "loss": 0.6427,
+      "step": 7967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871627982290773e-07,
+      "loss": 0.6346,
+      "step": 7968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.859369464318976e-07,
+      "loss": 0.607,
+      "step": 7969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.847123369890572e-07,
+      "loss": 0.6535,
+      "step": 7970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83488970062166e-07,
+      "loss": 0.5803,
+      "step": 7971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.822668458126701e-07,
+      "loss": 0.5721,
+      "step": 7972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.810459644018506e-07,
+      "loss": 0.6631,
+      "step": 7973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.798263259908232e-07,
+      "loss": 0.6724,
+      "step": 7974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.786079307405423e-07,
+      "loss": 0.5531,
+      "step": 7975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.77390778811796e-07,
+      "loss": 0.6451,
+      "step": 7976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.761748703652093e-07,
+      "loss": 0.5491,
+      "step": 7977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.749602055612436e-07,
+      "loss": 0.6647,
+      "step": 7978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737467845601962e-07,
+      "loss": 0.6456,
+      "step": 7979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.725346075221994e-07,
+      "loss": 0.6496,
+      "step": 7980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.713236746072215e-07,
+      "loss": 0.5592,
+      "step": 7981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.70113985975067e-07,
+      "loss": 0.6127,
+      "step": 7982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.689055417853762e-07,
+      "loss": 0.5877,
+      "step": 7983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.676983421976245e-07,
+      "loss": 0.6653,
+      "step": 7984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.664923873711237e-07,
+      "loss": 0.5748,
+      "step": 7985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.652876774650207e-07,
+      "loss": 0.5995,
+      "step": 7986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.640842126382973e-07,
+      "loss": 0.6475,
+      "step": 7987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.628819930497753e-07,
+      "loss": 0.6828,
+      "step": 7988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.61681018858109e-07,
+      "loss": 0.5853,
+      "step": 7989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604812902217849e-07,
+      "loss": 0.5542,
+      "step": 7990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.592828072991319e-07,
+      "loss": 0.6689,
+      "step": 7991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.580855702483113e-07,
+      "loss": 0.6571,
+      "step": 7992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.568895792273188e-07,
+      "loss": 0.6578,
+      "step": 7993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.556948343939861e-07,
+      "loss": 0.6107,
+      "step": 7994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.545013359059836e-07,
+      "loss": 0.596,
+      "step": 7995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533090839208133e-07,
+      "loss": 0.5824,
+      "step": 7996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521180785958136e-07,
+      "loss": 0.5395,
+      "step": 7997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509283200881588e-07,
+      "loss": 0.6011,
+      "step": 7998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.497398085548611e-07,
+      "loss": 0.6228,
+      "step": 7999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.485525441527651e-07,
+      "loss": 0.6176,
+      "step": 8000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.473665270385498e-07,
+      "loss": 0.617,
+      "step": 8001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.461817573687345e-07,
+      "loss": 0.6452,
+      "step": 8002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.449982352996675e-07,
+      "loss": 0.6353,
+      "step": 8003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.438159609875382e-07,
+      "loss": 0.5181,
+      "step": 8004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.426349345883675e-07,
+      "loss": 0.5887,
+      "step": 8005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.414551562580117e-07,
+      "loss": 0.5039,
+      "step": 8006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.402766261521675e-07,
+      "loss": 0.5985,
+      "step": 8007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.390993444263604e-07,
+      "loss": 0.6548,
+      "step": 8008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379233112359549e-07,
+      "loss": 0.5486,
+      "step": 8009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367485267361494e-07,
+      "loss": 0.6472,
+      "step": 8010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355749910819775e-07,
+      "loss": 0.6767,
+      "step": 8011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.344027044283074e-07,
+      "loss": 0.6364,
+      "step": 8012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332316669298454e-07,
+      "loss": 0.5943,
+      "step": 8013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.320618787411291e-07,
+      "loss": 0.643,
+      "step": 8014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30893340016535e-07,
+      "loss": 0.7101,
+      "step": 8015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.297260509102697e-07,
+      "loss": 0.636,
+      "step": 8016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.285600115763789e-07,
+      "loss": 0.6202,
+      "step": 8017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.27395222168745e-07,
+      "loss": 0.6334,
+      "step": 8018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262316828410807e-07,
+      "loss": 0.6734,
+      "step": 8019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.250693937469354e-07,
+      "loss": 0.6464,
+      "step": 8020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.239083550396962e-07,
+      "loss": 0.5242,
+      "step": 8021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227485668725795e-07,
+      "loss": 0.6332,
+      "step": 8022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.215900293986431e-07,
+      "loss": 0.5885,
+      "step": 8023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204327427707745e-07,
+      "loss": 0.6062,
+      "step": 8024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.192767071417016e-07,
+      "loss": 0.5097,
+      "step": 8025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.18121922663981e-07,
+      "loss": 0.5522,
+      "step": 8026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169683894900068e-07,
+      "loss": 0.6461,
+      "step": 8027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158161077720125e-07,
+      "loss": 0.6407,
+      "step": 8028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146650776620587e-07,
+      "loss": 0.5854,
+      "step": 8029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.135152993120463e-07,
+      "loss": 0.5665,
+      "step": 8030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.12366772873708e-07,
+      "loss": 0.6132,
+      "step": 8031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.112194984986118e-07,
+      "loss": 0.5977,
+      "step": 8032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100734763381632e-07,
+      "loss": 0.6363,
+      "step": 8033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.089287065435988e-07,
+      "loss": 0.6573,
+      "step": 8034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.077851892659924e-07,
+      "loss": 0.6343,
+      "step": 8035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06642924656251e-07,
+      "loss": 0.5527,
+      "step": 8036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.055019128651151e-07,
+      "loss": 0.6724,
+      "step": 8037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.043621540431654e-07,
+      "loss": 0.5541,
+      "step": 8038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032236483408115e-07,
+      "loss": 0.622,
+      "step": 8039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.020863959083e-07,
+      "loss": 0.608,
+      "step": 8040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.009503968957108e-07,
+      "loss": 0.6586,
+      "step": 8041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.998156514529595e-07,
+      "loss": 0.6176,
+      "step": 8042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.986821597297986e-07,
+      "loss": 0.6318,
+      "step": 8043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975499218758096e-07,
+      "loss": 0.6009,
+      "step": 8044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.964189380404128e-07,
+      "loss": 0.6119,
+      "step": 8045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.952892083728622e-07,
+      "loss": 0.6983,
+      "step": 8046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.941607330222442e-07,
+      "loss": 0.608,
+      "step": 8047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93033512137483e-07,
+      "loss": 0.5525,
+      "step": 8048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.919075458673362e-07,
+      "loss": 0.593,
+      "step": 8049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.907828343603938e-07,
+      "loss": 0.6237,
+      "step": 8050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896593777650815e-07,
+      "loss": 0.7065,
+      "step": 8051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.885371762296609e-07,
+      "loss": 0.6443,
+      "step": 8052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.874162299022244e-07,
+      "loss": 0.6336,
+      "step": 8053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.862965389307028e-07,
+      "loss": 0.6171,
+      "step": 8054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8517810346286e-07,
+      "loss": 0.6407,
+      "step": 8055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.840609236462912e-07,
+      "loss": 0.5418,
+      "step": 8056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829449996284286e-07,
+      "loss": 0.7302,
+      "step": 8057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.818303315565409e-07,
+      "loss": 0.6243,
+      "step": 8058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.807169195777261e-07,
+      "loss": 0.735,
+      "step": 8059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.796047638389212e-07,
+      "loss": 0.5075,
+      "step": 8060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.784938644868909e-07,
+      "loss": 0.6561,
+      "step": 8061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.773842216682434e-07,
+      "loss": 0.6092,
+      "step": 8062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.762758355294128e-07,
+      "loss": 0.6213,
+      "step": 8063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.751687062166721e-07,
+      "loss": 0.557,
+      "step": 8064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.740628338761255e-07,
+      "loss": 0.6327,
+      "step": 8065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7295821865371183e-07,
+      "loss": 0.6592,
+      "step": 8066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7185486069520783e-07,
+      "loss": 0.5607,
+      "step": 8067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707527601462192e-07,
+      "loss": 0.6675,
+      "step": 8068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696519171521885e-07,
+      "loss": 0.6909,
+      "step": 8069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6855233185839175e-07,
+      "loss": 0.5489,
+      "step": 8070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.674540044099396e-07,
+      "loss": 0.6151,
+      "step": 8071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.663569349517749e-07,
+      "loss": 0.6586,
+      "step": 8072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6526112362867523e-07,
+      "loss": 0.583,
+      "step": 8073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641665705852538e-07,
+      "loss": 0.6351,
+      "step": 8074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630732759659562e-07,
+      "loss": 0.5455,
+      "step": 8075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.619812399150603e-07,
+      "loss": 0.7116,
+      "step": 8076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6089046257668303e-07,
+      "loss": 0.6743,
+      "step": 8077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.598009440947704e-07,
+      "loss": 0.7034,
+      "step": 8078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5871268461310294e-07,
+      "loss": 0.6,
+      "step": 8079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5762568427529795e-07,
+      "loss": 0.6528,
+      "step": 8080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.565399432248041e-07,
+      "loss": 0.4919,
+      "step": 8081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554554616049045e-07,
+      "loss": 0.6264,
+      "step": 8082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5437223955871467e-07,
+      "loss": 0.625,
+      "step": 8083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5329027722918695e-07,
+      "loss": 0.5524,
+      "step": 8084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5220957475910377e-07,
+      "loss": 0.6356,
+      "step": 8085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.511301322910833e-07,
+      "loss": 0.7632,
+      "step": 8086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.500519499675793e-07,
+      "loss": 0.553,
+      "step": 8087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4897502793087576e-07,
+      "loss": 0.639,
+      "step": 8088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4789936632309236e-07,
+      "loss": 0.5826,
+      "step": 8089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4682496528618224e-07,
+      "loss": 0.6212,
+      "step": 8090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.457518249619308e-07,
+      "loss": 0.6245,
+      "step": 8091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4467994549195816e-07,
+      "loss": 0.5861,
+      "step": 8092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4360932701771885e-07,
+      "loss": 0.5913,
+      "step": 8093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4253996968050107e-07,
+      "loss": 0.7381,
+      "step": 8094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4147187362142187e-07,
+      "loss": 0.6009,
+      "step": 8095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.404050389814396e-07,
+      "loss": 0.6412,
+      "step": 8096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3933946590134166e-07,
+      "loss": 0.5603,
+      "step": 8097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3827515452174675e-07,
+      "loss": 0.5771,
+      "step": 8098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.372121049831135e-07,
+      "loss": 0.5657,
+      "step": 8099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3615031742572864e-07,
+      "loss": 0.6245,
+      "step": 8100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.350897919897146e-07,
+      "loss": 0.5634,
+      "step": 8101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3403052881502483e-07,
+      "loss": 0.5666,
+      "step": 8102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3297252804145094e-07,
+      "loss": 0.5253,
+      "step": 8103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3191578980861237e-07,
+      "loss": 0.635,
+      "step": 8104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3086031425596754e-07,
+      "loss": 0.641,
+      "step": 8105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2980610152280165e-07,
+      "loss": 0.5689,
+      "step": 8106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287531517482402e-07,
+      "loss": 0.6369,
+      "step": 8107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2770146507123745e-07,
+      "loss": 0.6019,
+      "step": 8108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2665104163058246e-07,
+      "loss": 0.528,
+      "step": 8109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.256018815648977e-07,
+      "loss": 0.5733,
+      "step": 8110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.24553985012639e-07,
+      "loss": 0.6177,
+      "step": 8111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.235073521120936e-07,
+      "loss": 0.6121,
+      "step": 8112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.224619830013854e-07,
+      "loss": 0.6273,
+      "step": 8113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2141787781846634e-07,
+      "loss": 0.5825,
+      "step": 8114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2037503670112946e-07,
+      "loss": 0.5708,
+      "step": 8115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1933345978699137e-07,
+      "loss": 0.6133,
+      "step": 8116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1829314721351213e-07,
+      "loss": 0.6398,
+      "step": 8117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.172540991179752e-07,
+      "loss": 0.5834,
+      "step": 8118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.162163156375043e-07,
+      "loss": 0.7097,
+      "step": 8119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1517979690905205e-07,
+      "loss": 0.5947,
+      "step": 8120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.141445430694069e-07,
+      "loss": 0.6511,
+      "step": 8121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1311055425518723e-07,
+      "loss": 0.588,
+      "step": 8122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.120778306028483e-07,
+      "loss": 0.6208,
+      "step": 8123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1104637224867573e-07,
+      "loss": 0.6613,
+      "step": 8124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1001617932878825e-07,
+      "loss": 0.6023,
+      "step": 8125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.089872519791394e-07,
+      "loss": 0.6101,
+      "step": 8126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.079595903355138e-07,
+      "loss": 0.6869,
+      "step": 8127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0693319453353086e-07,
+      "loss": 0.5896,
+      "step": 8128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.059080647086411e-07,
+      "loss": 0.6073,
+      "step": 8129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.048842009961296e-07,
+      "loss": 0.572,
+      "step": 8130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.038616035311116e-07,
+      "loss": 0.6241,
+      "step": 8131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.028402724485381e-07,
+      "loss": 0.5994,
+      "step": 8132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.018202078831934e-07,
+      "loss": 0.6437,
+      "step": 8133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.008014099696922e-07,
+      "loss": 0.6099,
+      "step": 8134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9978387884248235e-07,
+      "loss": 0.6554,
+      "step": 8135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9876761463584655e-07,
+      "loss": 0.6667,
+      "step": 8136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.977526174838997e-07,
+      "loss": 0.5527,
+      "step": 8137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9673888752058685e-07,
+      "loss": 0.5905,
+      "step": 8138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9572642487968994e-07,
+      "loss": 0.5799,
+      "step": 8139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.947152296948198e-07,
+      "loss": 0.5732,
+      "step": 8140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.937053020994219e-07,
+      "loss": 0.5959,
+      "step": 8141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.926966422267753e-07,
+      "loss": 0.5755,
+      "step": 8142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9168925020999007e-07,
+      "loss": 0.5343,
+      "step": 8143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9068312618200987e-07,
+      "loss": 0.5736,
+      "step": 8144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8967827027560965e-07,
+      "loss": 0.6095,
+      "step": 8145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8867468262340115e-07,
+      "loss": 0.7129,
+      "step": 8146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8767236335782274e-07,
+      "loss": 0.5866,
+      "step": 8147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.866713126111499e-07,
+      "loss": 0.649,
+      "step": 8148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.85671530515489e-07,
+      "loss": 0.6651,
+      "step": 8149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8467301720277905e-07,
+      "loss": 0.5805,
+      "step": 8150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836757728047913e-07,
+      "loss": 0.662,
+      "step": 8151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8267979745313043e-07,
+      "loss": 0.6612,
+      "step": 8152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.816850912792336e-07,
+      "loss": 0.5855,
+      "step": 8153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.806916544143702e-07,
+      "loss": 0.6339,
+      "step": 8154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.796994869896409e-07,
+      "loss": 0.5077,
+      "step": 8155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7870858913598097e-07,
+      "loss": 0.6102,
+      "step": 8156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7771896098415695e-07,
+      "loss": 0.6243,
+      "step": 8157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.767306026647688e-07,
+      "loss": 0.6018,
+      "step": 8158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.757435143082466e-07,
+      "loss": 0.7189,
+      "step": 8159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.747576960448551e-07,
+      "loss": 0.6045,
+      "step": 8160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.737731480046902e-07,
+      "loss": 0.6139,
+      "step": 8161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7278987031768134e-07,
+      "loss": 0.5884,
+      "step": 8162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7180786311359016e-07,
+      "loss": 0.5326,
+      "step": 8163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.708271265220087e-07,
+      "loss": 0.616,
+      "step": 8164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6984766067236223e-07,
+      "loss": 0.6121,
+      "step": 8165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.688694656939107e-07,
+      "loss": 0.5608,
+      "step": 8166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6789254171574416e-07,
+      "loss": 0.6762,
+      "step": 8167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.66916888866784e-07,
+      "loss": 0.6342,
+      "step": 8168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.659425072757861e-07,
+      "loss": 0.632,
+      "step": 8169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.649693970713375e-07,
+      "loss": 0.5434,
+      "step": 8170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.639975583818578e-07,
+      "loss": 0.5775,
+      "step": 8171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.630269913355977e-07,
+      "loss": 0.6299,
+      "step": 8172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.620576960606414e-07,
+      "loss": 0.6259,
+      "step": 8173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.610896726849045e-07,
+      "loss": 0.6424,
+      "step": 8174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.601229213361346e-07,
+      "loss": 0.6012,
+      "step": 8175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5915744214191416e-07,
+      "loss": 0.5842,
+      "step": 8176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5819323522965243e-07,
+      "loss": 0.6171,
+      "step": 8177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5723030072659646e-07,
+      "loss": 0.5788,
+      "step": 8178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.562686387598213e-07,
+      "loss": 0.5412,
+      "step": 8179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.553082494562354e-07,
+      "loss": 0.5732,
+      "step": 8180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5434913294257855e-07,
+      "loss": 0.6623,
+      "step": 8181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.533912893454261e-07,
+      "loss": 0.6273,
+      "step": 8182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5243471879117917e-07,
+      "loss": 0.7013,
+      "step": 8183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5147942140607684e-07,
+      "loss": 0.6046,
+      "step": 8184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5052539731618706e-07,
+      "loss": 0.6752,
+      "step": 8185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.495726466474103e-07,
+      "loss": 0.6243,
+      "step": 8186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.486211695254793e-07,
+      "loss": 0.6366,
+      "step": 8187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4767096607595805e-07,
+      "loss": 0.7152,
+      "step": 8188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4672203642424407e-07,
+      "loss": 0.6723,
+      "step": 8189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.457743806955638e-07,
+      "loss": 0.6908,
+      "step": 8190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4482799901497945e-07,
+      "loss": 0.5884,
+      "step": 8191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.438828915073822e-07,
+      "loss": 0.708,
+      "step": 8192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4293905829749565e-07,
+      "loss": 0.5811,
+      "step": 8193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.419964995098746e-07,
+      "loss": 0.7317,
+      "step": 8194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.410552152689073e-07,
+      "loss": 0.6101,
+      "step": 8195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.401152056988155e-07,
+      "loss": 0.6872,
+      "step": 8196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.391764709236467e-07,
+      "loss": 0.6493,
+      "step": 8197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3823901106728507e-07,
+      "loss": 0.6424,
+      "step": 8198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3730282625344614e-07,
+      "loss": 0.6039,
+      "step": 8199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3636791660567546e-07,
+      "loss": 0.71,
+      "step": 8200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.354342822473511e-07,
+      "loss": 0.7201,
+      "step": 8201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.345019233016833e-07,
+      "loss": 0.6549,
+      "step": 8202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3357083989171257e-07,
+      "loss": 0.6682,
+      "step": 8203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3264103214031284e-07,
+      "loss": 0.5672,
+      "step": 8204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.317125001701882e-07,
+      "loss": 0.6837,
+      "step": 8205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.307852441038761e-07,
+      "loss": 0.5492,
+      "step": 8206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2985926406374436e-07,
+      "loss": 0.6116,
+      "step": 8207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.289345601719918e-07,
+      "loss": 0.6211,
+      "step": 8208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2801113255065073e-07,
+      "loss": 0.7029,
+      "step": 8209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2708898132158385e-07,
+      "loss": 0.5987,
+      "step": 8210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.261681066064859e-07,
+      "loss": 0.58,
+      "step": 8211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.252485085268808e-07,
+      "loss": 0.6105,
+      "step": 8212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2433018720412825e-07,
+      "loss": 0.641,
+      "step": 8213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.234131427594145e-07,
+      "loss": 0.6139,
+      "step": 8214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.224973753137628e-07,
+      "loss": 0.5421,
+      "step": 8215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2158288498802426e-07,
+      "loss": 0.6635,
+      "step": 8216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.206696719028823e-07,
+      "loss": 0.6488,
+      "step": 8217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1975773617885063e-07,
+      "loss": 0.6971,
+      "step": 8218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1884707793627736e-07,
+      "loss": 0.6086,
+      "step": 8219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1793769729533854e-07,
+      "loss": 0.6016,
+      "step": 8220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1702959437604487e-07,
+      "loss": 0.6453,
+      "step": 8221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1612276929823383e-07,
+      "loss": 0.5612,
+      "step": 8222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.152172221815808e-07,
+      "loss": 0.5654,
+      "step": 8223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.143129531455857e-07,
+      "loss": 0.6579,
+      "step": 8224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1340996230958654e-07,
+      "loss": 0.5696,
+      "step": 8225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1250824979274675e-07,
+      "loss": 0.6302,
+      "step": 8226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1160781571406453e-07,
+      "loss": 0.6242,
+      "step": 8227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1070866019236813e-07,
+      "loss": 0.6078,
+      "step": 8228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0981078334631596e-07,
+      "loss": 0.6909,
+      "step": 8229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.089141852944011e-07,
+      "loss": 0.6516,
+      "step": 8230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0801886615494323e-07,
+      "loss": 0.611,
+      "step": 8231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0712482604609796e-07,
+      "loss": 0.689,
+      "step": 8232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0623206508584857e-07,
+      "loss": 0.5566,
+      "step": 8233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.053405833920109e-07,
+      "loss": 0.6205,
+      "step": 8234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.044503810822319e-07,
+      "loss": 0.6917,
+      "step": 8235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0356145827399207e-07,
+      "loss": 0.5931,
+      "step": 8236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.026738150845976e-07,
+      "loss": 0.6297,
+      "step": 8237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.017874516311903e-07,
+      "loss": 0.5995,
+      "step": 8238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.009023680307421e-07,
+      "loss": 0.6359,
+      "step": 8239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0001856440005307e-07,
+      "loss": 0.6874,
+      "step": 8240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.991360408557609e-07,
+      "loss": 0.5844,
+      "step": 8241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9825479751432793e-07,
+      "loss": 0.7086,
+      "step": 8242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9737483449205016e-07,
+      "loss": 0.5758,
+      "step": 8243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9649615190505465e-07,
+      "loss": 0.548,
+      "step": 8244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.956187498693008e-07,
+      "loss": 0.6164,
+      "step": 8245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.94742628500575e-07,
+      "loss": 0.6423,
+      "step": 8246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.938677879144991e-07,
+      "loss": 0.5769,
+      "step": 8247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.929942282265241e-07,
+      "loss": 0.5867,
+      "step": 8248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.921219495519301e-07,
+      "loss": 0.646,
+      "step": 8249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912509520058304e-07,
+      "loss": 0.7162,
+      "step": 8250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9038123570316987e-07,
+      "loss": 0.6257,
+      "step": 8251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8951280075872335e-07,
+      "loss": 0.4934,
+      "step": 8252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.886456472870935e-07,
+      "loss": 0.6645,
+      "step": 8253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8777977540272005e-07,
+      "loss": 0.7132,
+      "step": 8254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.869151852198682e-07,
+      "loss": 0.6329,
+      "step": 8255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.860518768526377e-07,
+      "loss": 0.6481,
+      "step": 8256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.851898504149575e-07,
+      "loss": 0.6985,
+      "step": 8257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.843291060205855e-07,
+      "loss": 0.6784,
+      "step": 8258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8346964378311416e-07,
+      "loss": 0.6828,
+      "step": 8259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826114638159638e-07,
+      "loss": 0.6128,
+      "step": 8260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.817545662323884e-07,
+      "loss": 0.6137,
+      "step": 8261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8089895114546964e-07,
+      "loss": 0.5739,
+      "step": 8262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.800446186681205e-07,
+      "loss": 0.6558,
+      "step": 8263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.791915689130864e-07,
+      "loss": 0.6222,
+      "step": 8264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7833980199294285e-07,
+      "loss": 0.5821,
+      "step": 8265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.774893180200966e-07,
+      "loss": 0.5738,
+      "step": 8266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.766401171067823e-07,
+      "loss": 0.6874,
+      "step": 8267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7579219936506807e-07,
+      "loss": 0.5698,
+      "step": 8268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.749455649068511e-07,
+      "loss": 0.618,
+      "step": 8269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.74100213843862e-07,
+      "loss": 0.6374,
+      "step": 8270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7325614628765815e-07,
+      "loss": 0.6308,
+      "step": 8271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7241336234962943e-07,
+      "loss": 0.6073,
+      "step": 8272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7157186214099575e-07,
+      "loss": 0.6025,
+      "step": 8273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707316457728115e-07,
+      "loss": 0.6462,
+      "step": 8274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.698927133559548e-07,
+      "loss": 0.7066,
+      "step": 8275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.690550650011392e-07,
+      "loss": 0.6126,
+      "step": 8276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.682187008189063e-07,
+      "loss": 0.6323,
+      "step": 8277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.673836209196312e-07,
+      "loss": 0.6194,
+      "step": 8278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6654982541351573e-07,
+      "loss": 0.5529,
+      "step": 8279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6571731441059623e-07,
+      "loss": 0.6009,
+      "step": 8280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64886088020736e-07,
+      "loss": 0.6911,
+      "step": 8281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6405614635363064e-07,
+      "loss": 0.6761,
+      "step": 8282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6322748951880583e-07,
+      "loss": 0.6248,
+      "step": 8283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6240011762561744e-07,
+      "loss": 0.6562,
+      "step": 8284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.615740307832526e-07,
+      "loss": 0.5469,
+      "step": 8285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.607492291007274e-07,
+      "loss": 0.6548,
+      "step": 8286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.599257126868904e-07,
+      "loss": 0.5758,
+      "step": 8287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5910348165041787e-07,
+      "loss": 0.6468,
+      "step": 8288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.582825360998187e-07,
+      "loss": 0.6623,
+      "step": 8289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5746287614343167e-07,
+      "loss": 0.5866,
+      "step": 8290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.566445018894259e-07,
+      "loss": 0.694,
+      "step": 8291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5582741344579833e-07,
+      "loss": 0.639,
+      "step": 8292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550116109203815e-07,
+      "loss": 0.6613,
+      "step": 8293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5419709442083274e-07,
+      "loss": 0.647,
+      "step": 8294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.533838640546438e-07,
+      "loss": 0.6251,
+      "step": 8295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.525719199291332e-07,
+      "loss": 0.6188,
+      "step": 8296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5176126215145425e-07,
+      "loss": 0.5705,
+      "step": 8297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.509518908285857e-07,
+      "loss": 0.5884,
+      "step": 8298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.501438060673389e-07,
+      "loss": 0.6289,
+      "step": 8299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.493370079743551e-07,
+      "loss": 0.5638,
+      "step": 8300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.485314966561059e-07,
+      "loss": 0.5978,
+      "step": 8301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.477272722188917e-07,
+      "loss": 0.5866,
+      "step": 8302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.469243347688466e-07,
+      "loss": 0.5939,
+      "step": 8303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4612268441193243e-07,
+      "loss": 0.7026,
+      "step": 8304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.453223212539391e-07,
+      "loss": 0.5964,
+      "step": 8305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.445232454004909e-07,
+      "loss": 0.6099,
+      "step": 8306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.437254569570402e-07,
+      "loss": 0.5311,
+      "step": 8307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4292895602886836e-07,
+      "loss": 0.5446,
+      "step": 8308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.421337427210879e-07,
+      "loss": 0.6183,
+      "step": 8309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4133981713864276e-07,
+      "loss": 0.5762,
+      "step": 8310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4054717938630346e-07,
+      "loss": 0.6787,
+      "step": 8311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3975582956867524e-07,
+      "loss": 0.6838,
+      "step": 8312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.389657677901902e-07,
+      "loss": 0.539,
+      "step": 8313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3817699415510932e-07,
+      "loss": 0.6648,
+      "step": 8314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.373895087675282e-07,
+      "loss": 0.6252,
+      "step": 8315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.366033117313682e-07,
+      "loss": 0.5556,
+      "step": 8316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3581840315038295e-07,
+      "loss": 0.6153,
+      "step": 8317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3503478312815298e-07,
+      "loss": 0.5445,
+      "step": 8318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.342524517680933e-07,
+      "loss": 0.6303,
+      "step": 8319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3347140917344579e-07,
+      "loss": 0.5995,
+      "step": 8320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3269165544728355e-07,
+      "loss": 0.6464,
+      "step": 8321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.319131906925065e-07,
+      "loss": 0.5534,
+      "step": 8322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3113601501185024e-07,
+      "loss": 0.5452,
+      "step": 8323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3036012850787603e-07,
+      "loss": 0.6012,
+      "step": 8324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2958553128297644e-07,
+      "loss": 0.5585,
+      "step": 8325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2881222343937194e-07,
+      "loss": 0.6752,
+      "step": 8326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2804020507911528e-07,
+      "loss": 0.6356,
+      "step": 8327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2726947630408835e-07,
+      "loss": 0.6736,
+      "step": 8328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.26500037216002e-07,
+      "loss": 0.6313,
+      "step": 8329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2573188791639723e-07,
+      "loss": 0.6632,
+      "step": 8330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2496502850664624e-07,
+      "loss": 0.6364,
+      "step": 8331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2419945908794926e-07,
+      "loss": 0.5848,
+      "step": 8332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2343517976133656e-07,
+      "loss": 0.5189,
+      "step": 8333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2267219062766965e-07,
+      "loss": 0.6535,
+      "step": 8334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.21910491787638e-07,
+      "loss": 0.6142,
+      "step": 8335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2115008334176125e-07,
+      "loss": 0.5714,
+      "step": 8336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2039096539038908e-07,
+      "loss": 0.6085,
+      "step": 8337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1963313803370023e-07,
+      "loss": 0.5786,
+      "step": 8338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1887660137170364e-07,
+      "loss": 0.6876,
+      "step": 8339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1812135550423942e-07,
+      "loss": 0.6334,
+      "step": 8340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.173674005309734e-07,
+      "loss": 0.5727,
+      "step": 8341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1661473655140485e-07,
+      "loss": 0.5451,
+      "step": 8342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1586336366486104e-07,
+      "loss": 0.5323,
+      "step": 8343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.151132819704993e-07,
+      "loss": 0.5724,
+      "step": 8344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.14364491567306e-07,
+      "loss": 0.6281,
+      "step": 8345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1361699255409872e-07,
+      "loss": 0.6522,
+      "step": 8346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1287078502952086e-07,
+      "loss": 0.6203,
+      "step": 8347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1212586909204914e-07,
+      "loss": 0.5834,
+      "step": 8348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1138224483998936e-07,
+      "loss": 0.5628,
+      "step": 8349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.106399123714764e-07,
+      "loss": 0.6649,
+      "step": 8350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0989887178447078e-07,
+      "loss": 0.6651,
+      "step": 8351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.091591231767709e-07,
+      "loss": 0.6128,
+      "step": 8352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0842066664599758e-07,
+      "loss": 0.6461,
+      "step": 8353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0768350228960288e-07,
+      "loss": 0.6168,
+      "step": 8354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0694763020486896e-07,
+      "loss": 0.6929,
+      "step": 8355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0621305048890815e-07,
+      "loss": 0.6332,
+      "step": 8356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.054797632386607e-07,
+      "loss": 0.6018,
+      "step": 8357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0474776855089806e-07,
+      "loss": 0.6372,
+      "step": 8358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0401706652221854e-07,
+      "loss": 0.7497,
+      "step": 8359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0328765724905274e-07,
+      "loss": 0.5718,
+      "step": 8360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0255954082765816e-07,
+      "loss": 0.5699,
+      "step": 8361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0183271735412347e-07,
+      "loss": 0.6062,
+      "step": 8362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0110718692436527e-07,
+      "loss": 0.6589,
+      "step": 8363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0038294963413251e-07,
+      "loss": 0.6276,
+      "step": 8364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.996600055789988e-07,
+      "loss": 0.6164,
+      "step": 8365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9893835485436996e-07,
+      "loss": 0.6421,
+      "step": 8366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9821799755548211e-07,
+      "loss": 0.5731,
+      "step": 8367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9749893377739803e-07,
+      "loss": 0.6738,
+      "step": 8368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9678116361501189e-07,
+      "loss": 0.6532,
+      "step": 8369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9606468716304562e-07,
+      "loss": 0.5005,
+      "step": 8370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9534950451605028e-07,
+      "loss": 0.5524,
+      "step": 8371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.946356157684093e-07,
+      "loss": 0.65,
+      "step": 8372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.939230210143328e-07,
+      "loss": 0.6761,
+      "step": 8373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9321172034785895e-07,
+      "loss": 0.6793,
+      "step": 8374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.925017138628571e-07,
+      "loss": 0.6199,
+      "step": 8375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9179300165302562e-07,
+      "loss": 0.5409,
+      "step": 8376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.910855838118919e-07,
+      "loss": 0.5393,
+      "step": 8377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.903794604328124e-07,
+      "loss": 0.656,
+      "step": 8378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8967463160897258e-07,
+      "loss": 0.5427,
+      "step": 8379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8897109743338694e-07,
+      "loss": 0.6671,
+      "step": 8380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8826885799889893e-07,
+      "loss": 0.588,
+      "step": 8381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8756791339818225e-07,
+      "loss": 0.5741,
+      "step": 8382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8686826372374066e-07,
+      "loss": 0.5125,
+      "step": 8383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8616990906790366e-07,
+      "loss": 0.6213,
+      "step": 8384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8547284952283084e-07,
+      "loss": 0.5954,
+      "step": 8385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.847770851805142e-07,
+      "loss": 0.6756,
+      "step": 8386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8408261613277022e-07,
+      "loss": 0.5478,
+      "step": 8387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.833894424712479e-07,
+      "loss": 0.6165,
+      "step": 8388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8269756428742403e-07,
+      "loss": 0.5993,
+      "step": 8389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8200698167260333e-07,
+      "loss": 0.6704,
+      "step": 8390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.813176947179196e-07,
+      "loss": 0.5568,
+      "step": 8391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8062970351433895e-07,
+      "loss": 0.6992,
+      "step": 8392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.799430081526532e-07,
+      "loss": 0.5997,
+      "step": 8393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.792576087234832e-07,
+      "loss": 0.6808,
+      "step": 8394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7857350531728213e-07,
+      "loss": 0.6123,
+      "step": 8395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7789069802432668e-07,
+      "loss": 0.6168,
+      "step": 8396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.77209186934727e-07,
+      "loss": 0.5658,
+      "step": 8397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7652897213842113e-07,
+      "loss": 0.5089,
+      "step": 8398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7585005372517504e-07,
+      "loss": 0.6682,
+      "step": 8399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7517243178458486e-07,
+      "loss": 0.6182,
+      "step": 8400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.744961064060724e-07,
+      "loss": 0.5349,
+      "step": 8401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.73821077678894e-07,
+      "loss": 0.5868,
+      "step": 8402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7314734569213065e-07,
+      "loss": 0.5915,
+      "step": 8403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7247491053469233e-07,
+      "loss": 0.6415,
+      "step": 8404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.718037722953203e-07,
+      "loss": 0.5999,
+      "step": 8405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7113393106258368e-07,
+      "loss": 0.6499,
+      "step": 8406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7046538692487957e-07,
+      "loss": 0.7306,
+      "step": 8407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6979813997043294e-07,
+      "loss": 0.6375,
+      "step": 8408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.691321902873011e-07,
+      "loss": 0.5393,
+      "step": 8409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6846753796336491e-07,
+      "loss": 0.5087,
+      "step": 8410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6780418308634083e-07,
+      "loss": 0.5175,
+      "step": 8411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6714212574376886e-07,
+      "loss": 0.6624,
+      "step": 8412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6648136602301912e-07,
+      "loss": 0.6307,
+      "step": 8413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6582190401129183e-07,
+      "loss": 0.6296,
+      "step": 8414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6516373979561295e-07,
+      "loss": 0.5942,
+      "step": 8415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6450687346283966e-07,
+      "loss": 0.5251,
+      "step": 8416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.638513050996582e-07,
+      "loss": 0.5667,
+      "step": 8417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6319703479258153e-07,
+      "loss": 0.6249,
+      "step": 8418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6254406262795176e-07,
+      "loss": 0.6173,
+      "step": 8419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6189238869194324e-07,
+      "loss": 0.6036,
+      "step": 8420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6124201307055277e-07,
+      "loss": 0.4424,
+      "step": 8421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059293584961056e-07,
+      "loss": 0.6892,
+      "step": 8422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5994515711477366e-07,
+      "loss": 0.6108,
+      "step": 8423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.592986769515281e-07,
+      "loss": 0.6405,
+      "step": 8424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5865349544518904e-07,
+      "loss": 0.7445,
+      "step": 8425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5800961268089942e-07,
+      "loss": 0.6032,
+      "step": 8426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.573670287436313e-07,
+      "loss": 0.6774,
+      "step": 8427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5672574371818462e-07,
+      "loss": 0.5531,
+      "step": 8428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.560857576891883e-07,
+      "loss": 0.6172,
+      "step": 8429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5544707074109932e-07,
+      "loss": 0.5599,
+      "step": 8430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5480968295820576e-07,
+      "loss": 0.6611,
+      "step": 8431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5417359442462143e-07,
+      "loss": 0.6109,
+      "step": 8432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5353880522429032e-07,
+      "loss": 0.6714,
+      "step": 8433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.529053154409832e-07,
+      "loss": 0.5656,
+      "step": 8434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5227312515830096e-07,
+      "loss": 0.6182,
+      "step": 8435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5164223445967242e-07,
+      "loss": 0.5916,
+      "step": 8436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.510126434283543e-07,
+      "loss": 0.6472,
+      "step": 8437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5038435214743242e-07,
+      "loss": 0.5883,
+      "step": 8438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.497573606998215e-07,
+      "loss": 0.5517,
+      "step": 8439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4913166916826538e-07,
+      "loss": 0.5417,
+      "step": 8440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4850727763533246e-07,
+      "loss": 0.6329,
+      "step": 8441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4788418618342571e-07,
+      "loss": 0.6488,
+      "step": 8442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4726239489477158e-07,
+      "loss": 0.6251,
+      "step": 8443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4664190385142663e-07,
+      "loss": 0.6152,
+      "step": 8444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4602271313527537e-07,
+      "loss": 0.5002,
+      "step": 8445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4540482282803136e-07,
+      "loss": 0.6885,
+      "step": 8446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4478823301123712e-07,
+      "loss": 0.5954,
+      "step": 8447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.44172943766262e-07,
+      "loss": 0.5755,
+      "step": 8448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4355895517430442e-07,
+      "loss": 0.6423,
+      "step": 8449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.429462673163906e-07,
+      "loss": 0.5965,
+      "step": 8450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4233488027337705e-07,
+      "loss": 0.6825,
+      "step": 8451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4172479412594586e-07,
+      "loss": 0.6247,
+      "step": 8452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4111600895461043e-07,
+      "loss": 0.565,
+      "step": 8453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.405085248397098e-07,
+      "loss": 0.5583,
+      "step": 8454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3990234186141204e-07,
+      "loss": 0.6792,
+      "step": 8455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3929746009971434e-07,
+      "loss": 0.5815,
+      "step": 8456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3869387963444281e-07,
+      "loss": 0.6153,
+      "step": 8457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3809160054524818e-07,
+      "loss": 0.6523,
+      "step": 8458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3749062291161463e-07,
+      "loss": 0.6437,
+      "step": 8459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3689094681284987e-07,
+      "loss": 0.6526,
+      "step": 8460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3629257232809278e-07,
+      "loss": 0.6535,
+      "step": 8461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3569549953631022e-07,
+      "loss": 0.6416,
+      "step": 8462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3509972851629584e-07,
+      "loss": 0.5661,
+      "step": 8463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3450525934667337e-07,
+      "loss": 0.5668,
+      "step": 8464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3391209210589117e-07,
+      "loss": 0.4963,
+      "step": 8465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3332022687223112e-07,
+      "loss": 0.5708,
+      "step": 8466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3272966372379846e-07,
+      "loss": 0.611,
+      "step": 8467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.321404027385298e-07,
+      "loss": 0.5638,
+      "step": 8468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3155244399418732e-07,
+      "loss": 0.5225,
+      "step": 8469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3096578756836343e-07,
+      "loss": 0.649,
+      "step": 8470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.303804335384795e-07,
+      "loss": 0.6468,
+      "step": 8471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2979638198178158e-07,
+      "loss": 0.6251,
+      "step": 8472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2921363297534684e-07,
+      "loss": 0.6698,
+      "step": 8473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2863218659607823e-07,
+      "loss": 0.6218,
+      "step": 8474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2805204292070772e-07,
+      "loss": 0.5529,
+      "step": 8475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2747320202579738e-07,
+      "loss": 0.5838,
+      "step": 8476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2689566398773613e-07,
+      "loss": 0.6264,
+      "step": 8477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.263194288827385e-07,
+      "loss": 0.6546,
+      "step": 8478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.257444967868482e-07,
+      "loss": 0.5961,
+      "step": 8479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2517086777594112e-07,
+      "loss": 0.5894,
+      "step": 8480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2459854192571563e-07,
+      "loss": 0.6419,
+      "step": 8481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2402751931170133e-07,
+      "loss": 0.5879,
+      "step": 8482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2345780000925456e-07,
+      "loss": 0.5853,
+      "step": 8483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2288938409355966e-07,
+      "loss": 0.6891,
+      "step": 8484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2232227163962995e-07,
+      "loss": 0.5823,
+      "step": 8485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2175646272230558e-07,
+      "loss": 0.6575,
+      "step": 8486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2119195741625456e-07,
+      "loss": 0.7092,
+      "step": 8487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.206287557959751e-07,
+      "loss": 0.6214,
+      "step": 8488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2006685793578998e-07,
+      "loss": 0.5426,
+      "step": 8489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.195062639098521e-07,
+      "loss": 0.669,
+      "step": 8490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1894697379214337e-07,
+      "loss": 0.5892,
+      "step": 8491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1838898765647144e-07,
+      "loss": 0.6635,
+      "step": 8492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1783230557647075e-07,
+      "loss": 0.6473,
+      "step": 8493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1727692762560805e-07,
+      "loss": 0.6545,
+      "step": 8494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1672285387717475e-07,
+      "loss": 0.5973,
+      "step": 8495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1617008440429012e-07,
+      "loss": 0.5864,
+      "step": 8496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1561861927990248e-07,
+      "loss": 0.5723,
+      "step": 8497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1506845857678694e-07,
+      "loss": 0.6536,
+      "step": 8498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1451960236754656e-07,
+      "loss": 0.6184,
+      "step": 8499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.139720507246156e-07,
+      "loss": 0.6417,
+      "step": 8500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1342580372024958e-07,
+      "loss": 0.5604,
+      "step": 8501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1288086142653865e-07,
+      "loss": 0.5913,
+      "step": 8502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1233722391539636e-07,
+      "loss": 0.5752,
+      "step": 8503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1179489125856535e-07,
+      "loss": 0.5684,
+      "step": 8504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1125386352761614e-07,
+      "loss": 0.5339,
+      "step": 8505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1071414079394716e-07,
+      "loss": 0.5965,
+      "step": 8506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1017572312878478e-07,
+      "loss": 0.5184,
+      "step": 8507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0963861060318326e-07,
+      "loss": 0.633,
+      "step": 8508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0910280328802148e-07,
+      "loss": 0.6056,
+      "step": 8509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0856830125401285e-07,
+      "loss": 0.6357,
+      "step": 8510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0803510457169208e-07,
+      "loss": 0.6517,
+      "step": 8511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0750321331142288e-07,
+      "loss": 0.5989,
+      "step": 8512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0697262754340132e-07,
+      "loss": 0.568,
+      "step": 8513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0644334733764583e-07,
+      "loss": 0.6179,
+      "step": 8514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0591537276400387e-07,
+      "loss": 0.5146,
+      "step": 8515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0538870389215194e-07,
+      "loss": 0.5201,
+      "step": 8516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0486334079159332e-07,
+      "loss": 0.6668,
+      "step": 8517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0433928353165812e-07,
+      "loss": 0.6398,
+      "step": 8518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0381653218150655e-07,
+      "loss": 0.7783,
+      "step": 8519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0329508681012457e-07,
+      "loss": 0.6615,
+      "step": 8520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0277494748632599e-07,
+      "loss": 0.5797,
+      "step": 8521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0225611427875259e-07,
+      "loss": 0.5591,
+      "step": 8522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0173858725587405e-07,
+      "loss": 0.7408,
+      "step": 8523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0122236648598572e-07,
+      "loss": 0.6254,
+      "step": 8524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0070745203721532e-07,
+      "loss": 0.6177,
+      "step": 8525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0019384397751186e-07,
+      "loss": 0.6125,
+      "step": 8526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.968154237465666e-08,
+      "loss": 0.6009,
+      "step": 8527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.917054729625674e-08,
+      "loss": 0.6458,
+      "step": 8528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.866085880974819e-08,
+      "loss": 0.6692,
+      "step": 8529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.81524769823905e-08,
+      "loss": 0.537,
+      "step": 8530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.76454018812778e-08,
+      "loss": 0.6378,
+      "step": 8531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.713963357332545e-08,
+      "loss": 0.6869,
+      "step": 8532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.663517212527896e-08,
+      "loss": 0.568,
+      "step": 8533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.613201760371061e-08,
+      "loss": 0.534,
+      "step": 8534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.563017007502173e-08,
+      "loss": 0.6293,
+      "step": 8535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.512962960543936e-08,
+      "loss": 0.629,
+      "step": 8536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.463039626101845e-08,
+      "loss": 0.6031,
+      "step": 8537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.413247010764182e-08,
+      "loss": 0.6293,
+      "step": 8538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.363585121102025e-08,
+      "loss": 0.6303,
+      "step": 8539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.314053963669245e-08,
+      "loss": 0.67,
+      "step": 8540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.264653545002278e-08,
+      "loss": 0.5076,
+      "step": 8541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.215383871620242e-08,
+      "loss": 0.5447,
+      "step": 8542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.166244950025383e-08,
+      "loss": 0.6441,
+      "step": 8543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.1172367867024e-08,
+      "loss": 0.4931,
+      "step": 8544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.068359388118674e-08,
+      "loss": 0.5053,
+      "step": 8545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.019612760724605e-08,
+      "loss": 0.6407,
+      "step": 8546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.970996910953045e-08,
+      "loss": 0.6229,
+      "step": 8547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.922511845219972e-08,
+      "loss": 0.4833,
+      "step": 8548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.874157569923492e-08,
+      "loss": 0.6744,
+      "step": 8549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.825934091445165e-08,
+      "loss": 0.6794,
+      "step": 8550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.777841416148792e-08,
+      "loss": 0.5544,
+      "step": 8551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.72987955038107e-08,
+      "loss": 0.6725,
+      "step": 8552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.682048500471385e-08,
+      "loss": 0.5409,
+      "step": 8553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.634348272732019e-08,
+      "loss": 0.6386,
+      "step": 8554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.586778873457713e-08,
+      "loss": 0.6321,
+      "step": 8555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.539340308926225e-08,
+      "loss": 0.6177,
+      "step": 8556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.49203258539788e-08,
+      "loss": 0.6678,
+      "step": 8557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.444855709115796e-08,
+      "loss": 0.5526,
+      "step": 8558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.397809686305769e-08,
+      "loss": 0.6823,
+      "step": 8559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.35089452317639e-08,
+      "loss": 0.5369,
+      "step": 8560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.304110225919037e-08,
+      "loss": 0.6343,
+      "step": 8561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.257456800707663e-08,
+      "loss": 0.609,
+      "step": 8562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.210934253698899e-08,
+      "loss": 0.5662,
+      "step": 8563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.164542591032388e-08,
+      "loss": 0.6349,
+      "step": 8564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.118281818830343e-08,
+      "loss": 0.6016,
+      "step": 8565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.072151943197659e-08,
+      "loss": 0.5738,
+      "step": 8566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.026152970222023e-08,
+      "loss": 0.6673,
+      "step": 8567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.98028490597369e-08,
+      "loss": 0.5897,
+      "step": 8568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.934547756505928e-08,
+      "loss": 0.6669,
+      "step": 8569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.888941527854576e-08,
+      "loss": 0.6511,
+      "step": 8570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.843466226038155e-08,
+      "loss": 0.4978,
+      "step": 8571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.798121857057972e-08,
+      "loss": 0.5564,
+      "step": 8572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.752908426897909e-08,
+      "loss": 0.5193,
+      "step": 8573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.707825941524861e-08,
+      "loss": 0.5944,
+      "step": 8574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.662874406888177e-08,
+      "loss": 0.6141,
+      "step": 8575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.618053828920003e-08,
+      "loss": 0.6197,
+      "step": 8576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.573364213535161e-08,
+      "loss": 0.5087,
+      "step": 8577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.52880556663138e-08,
+      "loss": 0.6278,
+      "step": 8578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.484377894088956e-08,
+      "loss": 0.5916,
+      "step": 8579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.440081201770865e-08,
+      "loss": 0.6156,
+      "step": 8580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.395915495522876e-08,
+      "loss": 0.519,
+      "step": 8581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.351880781173437e-08,
+      "loss": 0.489,
+      "step": 8582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.307977064533678e-08,
+      "loss": 0.6338,
+      "step": 8583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264204351397519e-08,
+      "loss": 0.5775,
+      "step": 8584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.220562647541562e-08,
+      "loss": 0.616,
+      "step": 8585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.17705195872509e-08,
+      "loss": 0.6712,
+      "step": 8586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.133672290690064e-08,
+      "loss": 0.6074,
+      "step": 8587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.090423649161348e-08,
+      "loss": 0.5629,
+      "step": 8588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.04730603984638e-08,
+      "loss": 0.6269,
+      "step": 8589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.004319468435162e-08,
+      "loss": 0.6418,
+      "step": 8590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.961463940600599e-08,
+      "loss": 0.6082,
+      "step": 8591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.91873946199817e-08,
+      "loss": 0.7052,
+      "step": 8592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.876146038266363e-08,
+      "loss": 0.6168,
+      "step": 8593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.833683675025904e-08,
+      "loss": 0.6675,
+      "step": 8594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.791352377880533e-08,
+      "loss": 0.5069,
+      "step": 8595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.749152152416783e-08,
+      "loss": 0.6359,
+      "step": 8596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.707083004203418e-08,
+      "loss": 0.5838,
+      "step": 8597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.665144938792445e-08,
+      "loss": 0.676,
+      "step": 8598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.623337961718323e-08,
+      "loss": 0.5809,
+      "step": 8599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.581662078498197e-08,
+      "loss": 0.5965,
+      "step": 8600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.540117294631887e-08,
+      "loss": 0.6479,
+      "step": 8601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.49870361560212e-08,
+      "loss": 0.7177,
+      "step": 8602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.45742104687408e-08,
+      "loss": 0.6143,
+      "step": 8603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.41626959389574e-08,
+      "loss": 0.5823,
+      "step": 8604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.37524926209776e-08,
+      "loss": 0.5963,
+      "step": 8605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.334360056893585e-08,
+      "loss": 0.5947,
+      "step": 8606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.293601983679121e-08,
+      "loss": 0.5915,
+      "step": 8607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.252975047833287e-08,
+      "loss": 0.551,
+      "step": 8608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.212479254717462e-08,
+      "loss": 0.6372,
+      "step": 8609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.172114609675816e-08,
+      "loss": 0.5822,
+      "step": 8610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.131881118035199e-08,
+      "loss": 0.6768,
+      "step": 8611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.091778785105029e-08,
+      "loss": 0.6237,
+      "step": 8612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.05180761617774e-08,
+      "loss": 0.6145,
+      "step": 8613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.011967616528003e-08,
+      "loss": 0.7137,
+      "step": 8614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.97225879141361e-08,
+      "loss": 0.6199,
+      "step": 8615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9326811460748146e-08,
+      "loss": 0.65,
+      "step": 8616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8932346857344395e-08,
+      "loss": 0.6456,
+      "step": 8617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.853919415598319e-08,
+      "loss": 0.6847,
+      "step": 8618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.814735340854749e-08,
+      "loss": 0.6118,
+      "step": 8619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.775682466674815e-08,
+      "loss": 0.6328,
+      "step": 8620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.736760798212282e-08,
+      "loss": 0.5547,
+      "step": 8621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.697970340603376e-08,
+      "loss": 0.5974,
+      "step": 8622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.659311098967335e-08,
+      "loss": 0.6264,
+      "step": 8623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.620783078405967e-08,
+      "loss": 0.6774,
+      "step": 8624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.582386284003649e-08,
+      "loss": 0.6508,
+      "step": 8625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.544120720827662e-08,
+      "loss": 0.6537,
+      "step": 8626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.50598639392752e-08,
+      "loss": 0.5775,
+      "step": 8627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.467983308336089e-08,
+      "loss": 0.5751,
+      "step": 8628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.430111469068356e-08,
+      "loss": 0.5876,
+      "step": 8629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.392370881122322e-08,
+      "loss": 0.6271,
+      "step": 8630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.354761549478338e-08,
+      "loss": 0.5736,
+      "step": 8631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3172834790997664e-08,
+      "loss": 0.5641,
+      "step": 8632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.27993667493254e-08,
+      "loss": 0.633,
+      "step": 8633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2427211419051605e-08,
+      "loss": 0.6689,
+      "step": 8634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.205636884929033e-08,
+      "loss": 0.5696,
+      "step": 8635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.168683908897798e-08,
+      "loss": 0.6775,
+      "step": 8636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.131862218688222e-08,
+      "loss": 0.6802,
+      "step": 8637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0951718191596386e-08,
+      "loss": 0.6115,
+      "step": 8638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.058612715154065e-08,
+      "loss": 0.5215,
+      "step": 8639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.022184911495864e-08,
+      "loss": 0.5379,
+      "step": 8640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9858884129926343e-08,
+      "loss": 0.6246,
+      "step": 8641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.949723224434211e-08,
+      "loss": 0.6605,
+      "step": 8642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9136893505932205e-08,
+      "loss": 0.6935,
+      "step": 8643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8777867962249706e-08,
+      "loss": 0.6099,
+      "step": 8644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.842015566067559e-08,
+      "loss": 0.5578,
+      "step": 8645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8063756648414337e-08,
+      "loss": 0.6142,
+      "step": 8646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.770867097250054e-08,
+      "loss": 0.5772,
+      "step": 8647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.735489867979559e-08,
+      "loss": 0.6259,
+      "step": 8648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.700243981698327e-08,
+      "loss": 0.6561,
+      "step": 8649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6651294430578584e-08,
+      "loss": 0.554,
+      "step": 8650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.630146256692003e-08,
+      "loss": 0.6643,
+      "step": 8651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.595294427217623e-08,
+      "loss": 0.5725,
+      "step": 8652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.560573959233927e-08,
+      "loss": 0.6436,
+      "step": 8653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.525984857322807e-08,
+      "loss": 0.5643,
+      "step": 8654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.491527126049167e-08,
+      "loss": 0.584,
+      "step": 8655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.457200769960035e-08,
+      "loss": 0.6519,
+      "step": 8656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.423005793585566e-08,
+      "loss": 0.7089,
+      "step": 8657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.388942201438373e-08,
+      "loss": 0.6568,
+      "step": 8658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.355009998013748e-08,
+      "loss": 0.5451,
+      "step": 8659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3212091877896653e-08,
+      "loss": 0.5793,
+      "step": 8660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.287539775226668e-08,
+      "loss": 0.5611,
+      "step": 8661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.254001764768201e-08,
+      "loss": 0.5578,
+      "step": 8662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2205951608400574e-08,
+      "loss": 0.5372,
+      "step": 8663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1873199678508226e-08,
+      "loss": 0.5659,
+      "step": 8664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.154176190191872e-08,
+      "loss": 0.6298,
+      "step": 8665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1211638322370406e-08,
+      "loss": 0.6287,
+      "step": 8666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0882828983429545e-08,
+      "loss": 0.5983,
+      "step": 8667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0555333928488094e-08,
+      "loss": 0.6991,
+      "step": 8668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0229153200764816e-08,
+      "loss": 0.6105,
+      "step": 8669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.990428684330638e-08,
+      "loss": 0.671,
+      "step": 8670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9580734898981845e-08,
+      "loss": 0.5581,
+      "step": 8671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.925849741049259e-08,
+      "loss": 0.5511,
+      "step": 8672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.893757442036239e-08,
+      "loss": 0.6936,
+      "step": 8673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8617965970941805e-08,
+      "loss": 0.6696,
+      "step": 8674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.829967210441155e-08,
+      "loss": 0.5787,
+      "step": 8675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7982692862773563e-08,
+      "loss": 0.7151,
+      "step": 8676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.766702828785995e-08,
+      "loss": 0.6682,
+      "step": 8677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.73526784213285e-08,
+      "loss": 0.4948,
+      "step": 8678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.703964330466381e-08,
+      "loss": 0.5662,
+      "step": 8679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.672792297917505e-08,
+      "loss": 0.6564,
+      "step": 8680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.641751748600042e-08,
+      "loss": 0.5946,
+      "step": 8681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.610842686610272e-08,
+      "loss": 0.6047,
+      "step": 8682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.580065116027265e-08,
+      "loss": 0.6887,
+      "step": 8683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.54941904091266e-08,
+      "loss": 0.596,
+      "step": 8684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.51890446531078e-08,
+      "loss": 0.6099,
+      "step": 8685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.488521393248401e-08,
+      "loss": 0.631,
+      "step": 8686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4582698287354275e-08,
+      "loss": 0.6206,
+      "step": 8687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.428149775763778e-08,
+      "loss": 0.6224,
+      "step": 8688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.398161238308606e-08,
+      "loss": 0.5802,
+      "step": 8689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.36830422032719e-08,
+      "loss": 0.6905,
+      "step": 8690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3385787257598224e-08,
+      "loss": 0.6087,
+      "step": 8691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3089847585293653e-08,
+      "loss": 0.6221,
+      "step": 8692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.27952232254114e-08,
+      "loss": 0.6272,
+      "step": 8693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2501914216833686e-08,
+      "loss": 0.6426,
+      "step": 8694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.220992059826622e-08,
+      "loss": 0.5935,
+      "step": 8695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.191924240824484e-08,
+      "loss": 0.5665,
+      "step": 8696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.162987968512887e-08,
+      "loss": 0.5825,
+      "step": 8697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.134183246710554e-08,
+      "loss": 0.6717,
+      "step": 8698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.105510079218665e-08,
+      "loss": 0.6566,
+      "step": 8699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.076968469821084e-08,
+      "loss": 0.6771,
+      "step": 8700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.048558422284686e-08,
+      "loss": 0.5915,
+      "step": 8701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0202799403584724e-08,
+      "loss": 0.5918,
+      "step": 8702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9921330277742355e-08,
+      "loss": 0.5769,
+      "step": 8703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9641176882466705e-08,
+      "loss": 0.5456,
+      "step": 8704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.93623392547282e-08,
+      "loss": 0.6078,
+      "step": 8705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9084817431322964e-08,
+      "loss": 0.5916,
+      "step": 8706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8808611448877254e-08,
+      "loss": 0.6331,
+      "step": 8707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8533721343840802e-08,
+      "loss": 0.6447,
+      "step": 8708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8260147152489035e-08,
+      "loss": 0.6787,
+      "step": 8709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7987888910925297e-08,
+      "loss": 0.5706,
+      "step": 8710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7716946655080844e-08,
+      "loss": 0.5567,
+      "step": 8711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.74473204207093e-08,
+      "loss": 0.647,
+      "step": 8712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.717901024339331e-08,
+      "loss": 0.634,
+      "step": 8713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.691201615854233e-08,
+      "loss": 0.6829,
+      "step": 8714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.664633820138818e-08,
+      "loss": 0.5844,
+      "step": 8715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6381976406995023e-08,
+      "loss": 0.5786,
+      "step": 8716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.61189308102483e-08,
+      "loss": 0.6805,
+      "step": 8717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5857201445861347e-08,
+      "loss": 0.5579,
+      "step": 8718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5596788348375423e-08,
+      "loss": 0.6581,
+      "step": 8719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5337691552156372e-08,
+      "loss": 0.5831,
+      "step": 8720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5079911091395737e-08,
+      "loss": 0.5966,
+      "step": 8721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4823447000112967e-08,
+      "loss": 0.5533,
+      "step": 8722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.456829931215321e-08,
+      "loss": 0.5376,
+      "step": 8723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.43144680611862e-08,
+      "loss": 0.5975,
+      "step": 8724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4061953280711813e-08,
+      "loss": 0.5858,
+      "step": 8725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.381075500405339e-08,
+      "loss": 0.6345,
+      "step": 8726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3560873264359963e-08,
+      "loss": 0.4923,
+      "step": 8727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3312308094607382e-08,
+      "loss": 0.6671,
+      "step": 8728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.306505952760163e-08,
+      "loss": 0.6225,
+      "step": 8729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2819127595967715e-08,
+      "loss": 0.5679,
+      "step": 8730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2574512332163013e-08,
+      "loss": 0.5682,
+      "step": 8731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2331213768468363e-08,
+      "loss": 0.7055,
+      "step": 8732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2089231936991417e-08,
+      "loss": 0.5242,
+      "step": 8733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1848566869665522e-08,
+      "loss": 0.548,
+      "step": 8734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1609218598250826e-08,
+      "loss": 0.569,
+      "step": 8735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1371187154334283e-08,
+      "loss": 0.6251,
+      "step": 8736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1134472569328544e-08,
+      "loss": 0.6212,
+      "step": 8737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.089907487447196e-08,
+      "loss": 0.6606,
+      "step": 8738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.066499410082856e-08,
+      "loss": 0.6184,
+      "step": 8739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0432230279290312e-08,
+      "loss": 0.6343,
+      "step": 8740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0200783440574857e-08,
+      "loss": 0.6111,
+      "step": 8741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9970653615225542e-08,
+      "loss": 0.674,
+      "step": 8742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9741840833611413e-08,
+      "loss": 0.658,
+      "step": 8743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9514345125929423e-08,
+      "loss": 0.579,
+      "step": 8744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9288166522201113e-08,
+      "loss": 0.6139,
+      "step": 8745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9063305052275937e-08,
+      "loss": 0.5943,
+      "step": 8746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.883976074582683e-08,
+      "loss": 0.5756,
+      "step": 8747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8617533632354636e-08,
+      "loss": 0.5853,
+      "step": 8748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8396623741187004e-08,
+      "loss": 0.6583,
+      "step": 8749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.817703110147728e-08,
+      "loss": 0.5427,
+      "step": 8750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7958755742203405e-08,
+      "loss": 0.5591,
+      "step": 8751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7741797692171213e-08,
+      "loss": 0.7119,
+      "step": 8752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7526156980012254e-08,
+      "loss": 0.608,
+      "step": 8753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.731183363418376e-08,
+      "loss": 0.6668,
+      "step": 8754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7098827682970885e-08,
+      "loss": 0.5451,
+      "step": 8755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6887139154483367e-08,
+      "loss": 0.6309,
+      "step": 8756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6676768076655525e-08,
+      "loss": 0.6682,
+      "step": 8757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6467714477250706e-08,
+      "loss": 0.6548,
+      "step": 8758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6259978383857955e-08,
+      "loss": 0.5601,
+      "step": 8759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6053559823890896e-08,
+      "loss": 0.6254,
+      "step": 8760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5848458824589963e-08,
+      "loss": 0.6581,
+      "step": 8761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.564467541302239e-08,
+      "loss": 0.6274,
+      "step": 8762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5442209616081117e-08,
+      "loss": 0.582,
+      "step": 8763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5241061460484764e-08,
+      "loss": 0.6276,
+      "step": 8764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5041230972778763e-08,
+      "loss": 0.6127,
+      "step": 8765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4842718179334248e-08,
+      "loss": 0.6151,
+      "step": 8766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4645523106349147e-08,
+      "loss": 0.5429,
+      "step": 8767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4449645779844868e-08,
+      "loss": 0.6948,
+      "step": 8768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4255086225674064e-08,
+      "loss": 0.6464,
+      "step": 8769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4061844469509534e-08,
+      "loss": 0.6015,
+      "step": 8770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3869920536854209e-08,
+      "loss": 0.5681,
+      "step": 8771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3679314453035609e-08,
+      "loss": 0.6515,
+      "step": 8772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3490026243208054e-08,
+      "loss": 0.654,
+      "step": 8773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3302055932350455e-08,
+      "loss": 0.6624,
+      "step": 8774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3115403545270744e-08,
+      "loss": 0.599,
+      "step": 8775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2930069106599218e-08,
+      "loss": 0.6065,
+      "step": 8776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2746052640795203e-08,
+      "loss": 0.6268,
+      "step": 8777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2563354172142605e-08,
+      "loss": 0.6729,
+      "step": 8778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2381973724752139e-08,
+      "loss": 0.6651,
+      "step": 8779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2201911322559102e-08,
+      "loss": 0.6251,
+      "step": 8780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2023166989326706e-08,
+      "loss": 0.582,
+      "step": 8781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1845740748642754e-08,
+      "loss": 0.4942,
+      "step": 8782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1669632623922955e-08,
+      "loss": 0.6418,
+      "step": 8783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1494842638407611e-08,
+      "loss": 0.7002,
+      "step": 8784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1321370815163824e-08,
+      "loss": 0.5719,
+      "step": 8785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1149217177083283e-08,
+      "loss": 0.5325,
+      "step": 8786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0978381746884481e-08,
+      "loss": 0.6104,
+      "step": 8787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0808864547113828e-08,
+      "loss": 0.5531,
+      "step": 8788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.064066560014121e-08,
+      "loss": 0.5595,
+      "step": 8789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.047378492816331e-08,
+      "loss": 0.5626,
+      "step": 8790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0308222553202518e-08,
+      "loss": 0.6081,
+      "step": 8791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.014397849711024e-08,
+      "loss": 0.6447,
+      "step": 8792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.981052781558032e-09,
+      "loss": 0.6328,
+      "step": 8793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.81944542804869e-09,
+      "loss": 0.5862,
+      "step": 8794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.659156457909602e-09,
+      "loss": 0.4852,
+      "step": 8795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.500185892292735e-09,
+      "loss": 0.6666,
+      "step": 8796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.342533752177974e-09,
+      "loss": 0.6406,
+      "step": 8797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.186200058369788e-09,
+      "loss": 0.6086,
+      "step": 8798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.03118483149834e-09,
+      "loss": 0.6879,
+      "step": 8799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.877488092022823e-09,
+      "loss": 0.6022,
+      "step": 8800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.72510986022479e-09,
+      "loss": 0.6396,
+      "step": 8801
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.57405015621371e-09,
+      "loss": 0.5883,
+      "step": 8802
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.424308999923635e-09,
+      "loss": 0.6182,
+      "step": 8803
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.275886411117651e-09,
+      "loss": 0.5339,
+      "step": 8804
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.128782409381197e-09,
+      "loss": 0.6064,
+      "step": 8805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.982997014127636e-09,
+      "loss": 0.6261,
+      "step": 8806
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.83853024459602e-09,
+      "loss": 0.6458,
+      "step": 8807
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.695382119851102e-09,
+      "loss": 0.6467,
+      "step": 8808
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.553552658784435e-09,
+      "loss": 0.5186,
+      "step": 8809
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.4130418801121595e-09,
+      "loss": 0.5779,
+      "step": 8810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.27384980237833e-09,
+      "loss": 0.6715,
+      "step": 8811
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.135976443950477e-09,
+      "loss": 0.6294,
+      "step": 8812
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.999421823024044e-09,
+      "loss": 0.6364,
+      "step": 8813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.864185957620173e-09,
+      "loss": 0.6416,
+      "step": 8814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7302688655856984e-09,
+      "loss": 0.6166,
+      "step": 8815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.597670564593151e-09,
+      "loss": 0.6399,
+      "step": 8816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.466391072140754e-09,
+      "loss": 0.5556,
+      "step": 8817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.33643040555354e-09,
+      "loss": 0.7776,
+      "step": 8818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2077885819822325e-09,
+      "loss": 0.5556,
+      "step": 8819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.080465618403253e-09,
+      "loss": 0.5852,
+      "step": 8820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9544615316198266e-09,
+      "loss": 0.7122,
+      "step": 8821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8297763382597625e-09,
+      "loss": 0.5053,
+      "step": 8822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.706410054777678e-09,
+      "loss": 0.5955,
+      "step": 8823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.584362697453882e-09,
+      "loss": 0.659,
+      "step": 8824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.463634282395492e-09,
+      "loss": 0.6921,
+      "step": 8825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3442248255330954e-09,
+      "loss": 0.6562,
+      "step": 8826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.22613434262631e-09,
+      "loss": 0.6417,
+      "step": 8827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.109362849259336e-09,
+      "loss": 0.5744,
+      "step": 8828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.993910360840959e-09,
+      "loss": 0.5925,
+      "step": 8829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8797768926089896e-09,
+      "loss": 0.6525,
+      "step": 8830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.766962459623603e-09,
+      "loss": 0.736,
+      "step": 8831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.655467076774001e-09,
+      "loss": 0.6201,
+      "step": 8832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.545290758772858e-09,
+      "loss": 0.6904,
+      "step": 8833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.436433520161876e-09,
+      "loss": 0.6787,
+      "step": 8834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.32889537530512e-09,
+      "loss": 0.5735,
+      "step": 8835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.222676338393461e-09,
+      "loss": 0.571,
+      "step": 8836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.117776423446795e-09,
+      "loss": 0.6436,
+      "step": 8837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.014195644306273e-09,
+      "loss": 0.5556,
+      "step": 8838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9119340146420715e-09,
+      "loss": 0.5357,
+      "step": 8839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.81099154795006e-09,
+      "loss": 0.5611,
+      "step": 8840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.711368257550696e-09,
+      "loss": 0.6496,
+      "step": 8841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6130641565912394e-09,
+      "loss": 0.64,
+      "step": 8842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.516079258044647e-09,
+      "loss": 0.6273,
+      "step": 8843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.420413574709569e-09,
+      "loss": 0.5925,
+      "step": 8844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3260671192114622e-09,
+      "loss": 0.5553,
+      "step": 8845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.233039904001478e-09,
+      "loss": 0.6085,
+      "step": 8846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1413319413542422e-09,
+      "loss": 0.708,
+      "step": 8847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0509432433734053e-09,
+      "loss": 0.6235,
+      "step": 8848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9618738219872045e-09,
+      "loss": 0.7475,
+      "step": 8849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8741236889506805e-09,
+      "loss": 0.6334,
+      "step": 8850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7876928558434603e-09,
+      "loss": 0.5693,
+      "step": 8851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7025813340719743e-09,
+      "loss": 0.6098,
+      "step": 8852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.618789134867239e-09,
+      "loss": 0.5965,
+      "step": 8853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5363162692881858e-09,
+      "loss": 0.5803,
+      "step": 8854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4551627482183316e-09,
+      "loss": 0.6412,
+      "step": 8855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.375328582366887e-09,
+      "loss": 0.5906,
+      "step": 8856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2968137822698687e-09,
+      "loss": 0.6007,
+      "step": 8857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2196183582889884e-09,
+      "loss": 0.7469,
+      "step": 8858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1437423206105423e-09,
+      "loss": 0.5905,
+      "step": 8859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.069185679248742e-09,
+      "loss": 0.5653,
+      "step": 8860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9959484440412736e-09,
+      "loss": 0.6476,
+      "step": 8861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.924030624654849e-09,
+      "loss": 0.566,
+      "step": 8862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8534322305796548e-09,
+      "loss": 0.6934,
+      "step": 8863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7841532711326826e-09,
+      "loss": 0.6761,
+      "step": 8864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7161937554555086e-09,
+      "loss": 0.6093,
+      "step": 8865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6495536925176248e-09,
+      "loss": 0.5909,
+      "step": 8866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5842330911131077e-09,
+      "loss": 0.5564,
+      "step": 8867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5202319598617288e-09,
+      "loss": 0.5805,
+      "step": 8868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4575503072100649e-09,
+      "loss": 0.5827,
+      "step": 8869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3961881414292776e-09,
+      "loss": 0.6655,
+      "step": 8870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3361454706195542e-09,
+      "loss": 0.5631,
+      "step": 8871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2774223027012255e-09,
+      "loss": 0.6125,
+      "step": 8872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2200186454269792e-09,
+      "loss": 0.58,
+      "step": 8873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1639345063696461e-09,
+      "loss": 0.6482,
+      "step": 8874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.109169892933304e-09,
+      "loss": 0.5657,
+      "step": 8875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0557248123421738e-09,
+      "loss": 0.6674,
+      "step": 8876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.003599271651723e-09,
+      "loss": 0.6421,
+      "step": 8877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.527932777386728e-10,
+      "loss": 0.5259,
+      "step": 8878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.033068373098808e-10,
+      "loss": 0.6659,
+      "step": 8879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.551399568945684e-10,
+      "loss": 0.5369,
+      "step": 8880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.082926428498727e-10,
+      "loss": 0.5153,
+      "step": 8881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.627649013586258e-10,
+      "loss": 0.7264,
+      "step": 8882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.185567384271342e-10,
+      "loss": 0.5337,
+      "step": 8883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.756681598907299e-10,
+      "loss": 0.6446,
+      "step": 8884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.340991714104405e-10,
+      "loss": 0.497,
+      "step": 8885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.938497784696573e-10,
+      "loss": 0.6729,
+      "step": 8886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.549199863819077e-10,
+      "loss": 0.6344,
+      "step": 8887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.173098002830834e-10,
+      "loss": 0.5411,
+      "step": 8888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.810192251381019e-10,
+      "loss": 0.6113,
+      "step": 8889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.460482657353549e-10,
+      "loss": 0.6968,
+      "step": 8890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.123969266911498e-10,
+      "loss": 0.5771,
+      "step": 8891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.800652124441584e-10,
+      "loss": 0.614,
+      "step": 8892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4905312726207786e-10,
+      "loss": 0.5375,
+      "step": 8893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.193606752371903e-10,
+      "loss": 0.6098,
+      "step": 8894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9098786028969315e-10,
+      "loss": 0.5986,
+      "step": 8895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.639346861610381e-10,
+      "loss": 0.5376,
+      "step": 8896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.382011564239228e-10,
+      "loss": 0.5695,
+      "step": 8897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.137872744734093e-10,
+      "loss": 0.5892,
+      "step": 8898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9069304353136474e-10,
+      "loss": 0.6288,
+      "step": 8899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6891846664535138e-10,
+      "loss": 0.6677,
+      "step": 8900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.484635466886264e-10,
+      "loss": 0.6306,
+      "step": 8901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2932828636125217e-10,
+      "loss": 0.6817,
+      "step": 8902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1151268818787586e-10,
+      "loss": 0.5825,
+      "step": 8903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.501675451994985e-11,
+      "loss": 0.6295,
+      "step": 8904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.984048753462148e-11,
+      "loss": 0.6689,
+      "step": 8905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.598388923362286e-11,
+      "loss": 0.5707,
+      "step": 8906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3446961447711775e-11,
+      "loss": 0.6463,
+      "step": 8907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.222970582890007e-11,
+      "loss": 0.6514,
+      "step": 8908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.233212385822526e-11,
+      "loss": 0.5555,
+      "step": 8909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3754216843530076e-11,
+      "loss": 0.6418,
+      "step": 8910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6495985913911327e-11,
+      "loss": 0.7159,
+      "step": 8911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0557432029711934e-11,
+      "loss": 0.5421,
+      "step": 8912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9385559736391265e-12,
+      "loss": 0.5628,
+      "step": 8913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.639358356315569e-12,
+      "loss": 0.5462,
+      "step": 8914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.598396107282412e-13,
+      "loss": 0.5021,
+      "step": 8915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6876,
+      "step": 8916
+    },
+    {
+      "epoch": 1.0,
+      "step": 8916,
+      "total_flos": 3255481882148864.0,
+      "train_loss": 0.6779399599667727,
+      "train_runtime": 87169.059,
+      "train_samples_per_second": 13.092,
+      "train_steps_per_second": 0.102
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8916,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 10000,
+  "total_flos": 3255481882148864.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}