Model save

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 5.0,
-    "train_loss": 0.1620459199016551,
-    "train_runtime": 43631.9674,
+    "epoch": 4.07,
+    "train_loss": 5.854813561474184e-06,
+    "train_runtime": 574.8329,
     "train_samples": 246408,
-    "train_samples_per_second": 28.237,
-    "train_steps_per_second": 0.441
+    "train_samples_per_second": 2143.301,
+    "train_steps_per_second": 33.497
 }

emissions.csv

@@ -1,3 +1,4 @@
 timestamp,experiment_id,project_name,duration,emissions,energy_consumed,country_name,country_iso_code,region,on_cloud,cloud_provider,cloud_region
 2022-04-07T05:24:20,9d6f504f-f106-4b54-a3c9-d2cc9c100918,codecarbon,43632.017745018005,9.320191315050605,16.458045762052986,USA,USA,Iowa,Y,gcp,us-central1
 2022-04-12T17:58:32,2ca859a2-6a59-473a-bf34-566c1ed805fb,codecarbon,574.8996481895447,0.12711368634643946,0.22446351111855814,USA,USA,Iowa,Y,gcp,us-central1
+2022-04-12T18:27:42,01e54279-0c0b-4c7b-9048-6e3d6739fae8,codecarbon,568.9987049102783,0.13174037435842878,0.23263354115915383,USA,USA,Iowa,Y,gcp,us-central1

runs/Apr12_17-48-07_anton-xtreme-s/events.out.tfevents.1649786354.anton-xtreme-s.19165.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f705f4114104cc28f687298b4b63607ed440f69f41dd5b38ea3a570018a088b
+size 17148

runs/Apr12_18-17-31_anton-xtreme-s/1649787493.4871702/events.out.tfevents.1649787493.anton-xtreme-s.23063.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73dfcda3e498e330cc9234085c6538dc61b38909edcc6934b8fab3f8ae066332
+size 5092

runs/Apr12_18-17-31_anton-xtreme-s/events.out.tfevents.1649787493.anton-xtreme-s.23063.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bff3ff6cb011e00f2d5b95a99babd33d3013ad22e6fc9ed8658928033ae83dd1
+size 49840

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 5.0,
-    "train_loss": 0.1620459199016551,
-    "train_runtime": 43631.9674,
+    "epoch": 4.07,
+    "train_loss": 5.854813561474184e-06,
+    "train_runtime": 574.8329,
     "train_samples": 246408,
-    "train_samples_per_second": 28.237,
-    "train_steps_per_second": 0.441
+    "train_samples_per_second": 2143.301,
+    "train_steps_per_second": 33.497
 }

trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": 0.7236466414789606,
   "best_model_checkpoint": "xtreme_s_xlsr_300m_fleurs_langid_truncated/checkpoint-19000",
-  "epoch": 4.999870146734191,
-  "global_step": 19250,
+  "epoch": 4.066216567125422,
+  "global_step": 19255,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -114179,1518 +114179,1548 @@
       "step": 19000
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.48695652173913e-06,
+      "epoch": 4.0,
+      "learning_rate": 4.572587655751956e-06,
       "loss": 0.0001,
       "step": 19001
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.469565217391304e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.555201390901187e-06,
+      "loss": 0.0002,
       "step": 19002
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.452173913043477e-06,
+      "epoch": 4.0,
+      "learning_rate": 4.53781512605042e-06,
       "loss": 0.0001,
       "step": 19003
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.434782608695652e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.520428861199652e-06,
+      "loss": 0.0002,
       "step": 19004
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.417391304347826e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.503042596348884e-06,
+      "loss": 0.0001,
       "step": 19005
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.399999999999999e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.4856563314981165e-06,
+      "loss": 0.0001,
       "step": 19006
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.382608695652174e-06,
-      "loss": 0.0002,
+      "epoch": 4.0,
+      "learning_rate": 4.468270066647348e-06,
+      "loss": 0.0,
       "step": 19007
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.365217391304348e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.45088380179658e-06,
+      "loss": 0.0001,
       "step": 19008
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.347826086956521e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.433497536945813e-06,
+      "loss": 0.0,
       "step": 19009
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.330434782608695e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.416111272095045e-06,
+      "loss": 0.0002,
       "step": 19010
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.313043478260869e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.398725007244277e-06,
+      "loss": 0.0001,
       "step": 19011
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.295652173913043e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.381338742393509e-06,
+      "loss": 0.0,
       "step": 19012
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.278260869565217e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.363952477542741e-06,
+      "loss": 0.0,
       "step": 19013
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.260869565217391e-06,
+      "epoch": 4.0,
+      "learning_rate": 4.346566212691972e-06,
       "loss": 0.0001,
       "step": 19014
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.2434782608695645e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.3291799478412056e-06,
+      "loss": 0.0002,
       "step": 19015
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.226086956521739e-06,
+      "epoch": 4.0,
+      "learning_rate": 4.311793682990437e-06,
       "loss": 0.0001,
       "step": 19016
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.208695652173913e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.294407418139669e-06,
+      "loss": 0.0,
       "step": 19017
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.1913043478260865e-06,
-      "loss": 0.0001,
+      "epoch": 4.0,
+      "learning_rate": 4.277021153288902e-06,
+      "loss": 0.0,
       "step": 19018
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.173913043478261e-06,
-      "loss": 0.0,
+      "epoch": 4.0,
+      "learning_rate": 4.259634888438133e-06,
+      "loss": 0.0001,
       "step": 19019
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.156521739130434e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 4.242248623587365e-06,
+      "loss": 0.0001,
       "step": 19020
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.1391304347826084e-06,
-      "loss": 0.0001,
+      "epoch": 4.01,
+      "learning_rate": 4.224862358736598e-06,
+      "loss": 0.0002,
       "step": 19021
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.121739130434782e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.20747609388583e-06,
       "loss": 0.0001,
       "step": 19022
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.104347826086956e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.190089829035061e-06,
       "loss": 0.0001,
       "step": 19023
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.08695652173913e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.172703564184294e-06,
       "loss": 0.0001,
       "step": 19024
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.069565217391304e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.155317299333526e-06,
       "loss": 0.0001,
       "step": 19025
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.052173913043478e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 4.137931034482758e-06,
+      "loss": 0.0001,
       "step": 19026
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.0347826086956515e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 4.120544769631991e-06,
+      "loss": 0.002,
       "step": 19027
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4.017391304347826e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.103158504781222e-06,
       "loss": 0.0001,
       "step": 19028
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 4e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.085772239930454e-06,
       "loss": 0.0,
       "step": 19029
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.9826086956521735e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.068385975079687e-06,
       "loss": 0.0001,
       "step": 19030
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.965217391304347e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 4.050999710228919e-06,
+      "loss": 0.0001,
       "step": 19031
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.947826086956521e-06,
-      "loss": 0.0001,
+      "epoch": 4.01,
+      "learning_rate": 4.033613445378151e-06,
+      "loss": 0.0,
       "step": 19032
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.9304347826086955e-06,
+      "epoch": 4.01,
+      "learning_rate": 4.016227180527383e-06,
       "loss": 0.0,
       "step": 19033
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.913043478260869e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.998840915676615e-06,
       "loss": 0.0001,
       "step": 19034
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.895652173913043e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.981454650825847e-06,
       "loss": 0.0001,
       "step": 19035
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.8782608695652175e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 3.96406838597508e-06,
+      "loss": 0.0001,
       "step": 19036
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.860869565217391e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.946682121124311e-06,
       "loss": 0.0001,
       "step": 19037
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 3.843478260869565e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.929295856273543e-06,
       "loss": 0.0001,
       "step": 19038
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.826086956521739e-06,
-      "loss": 0.0017,
+      "epoch": 4.01,
+      "learning_rate": 3.911909591422776e-06,
+      "loss": 0.0007,
       "step": 19039
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.8086956521739128e-06,
-      "loss": 0.0001,
+      "epoch": 4.01,
+      "learning_rate": 3.894523326572008e-06,
+      "loss": 0.0,
       "step": 19040
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.7913043478260865e-06,
-      "loss": 0.0002,
+      "epoch": 4.01,
+      "learning_rate": 3.87713706172124e-06,
+      "loss": 0.0,
       "step": 19041
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.7739130434782606e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.859750796870472e-06,
       "loss": 0.0001,
       "step": 19042
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.7565217391304347e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.842364532019704e-06,
       "loss": 0.0001,
       "step": 19043
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.7391304347826085e-06,
-      "loss": 0.0001,
+      "epoch": 4.01,
+      "learning_rate": 3.824978267168936e-06,
+      "loss": 0.0002,
       "step": 19044
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.7217391304347826e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.807592002318168e-06,
       "loss": 0.0001,
       "step": 19045
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.704347826086956e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.7902057374674005e-06,
       "loss": 0.0001,
       "step": 19046
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.68695652173913e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 3.7728194726166324e-06,
+      "loss": 0.0001,
       "step": 19047
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.669565217391304e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.7554332077658647e-06,
       "loss": 0.0,
       "step": 19048
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.652173913043478e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.7380469429150966e-06,
       "loss": 0.0,
       "step": 19049
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.634782608695652e-06,
-      "loss": 0.0002,
+      "epoch": 4.01,
+      "learning_rate": 3.720660678064329e-06,
+      "loss": 0.0,
       "step": 19050
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.6173913043478257e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 3.7032744132135607e-06,
+      "loss": 0.0001,
       "step": 19051
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.6e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 3.685888148362793e-06,
+      "loss": 0.0001,
       "step": 19052
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.5826086956521735e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 3.6685018835120253e-06,
+      "loss": 0.0001,
       "step": 19053
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.5652173913043477e-06,
-      "loss": 0.0001,
+      "epoch": 4.01,
+      "learning_rate": 3.6511156186612572e-06,
+      "loss": 0.0,
       "step": 19054
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.547826086956522e-06,
-      "loss": 0.0,
+      "epoch": 4.01,
+      "learning_rate": 3.6337293538104895e-06,
+      "loss": 0.0001,
       "step": 19055
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.530434782608695e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.6163430889597214e-06,
       "loss": 0.0001,
       "step": 19056
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.5130434782608692e-06,
+      "epoch": 4.01,
+      "learning_rate": 3.5989568241089537e-06,
       "loss": 0.0001,
       "step": 19057
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.495652173913043e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.5815705592581856e-06,
       "loss": 0.0001,
       "step": 19058
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.478260869565217e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.564184294407418e-06,
+      "loss": 0.0,
       "step": 19059
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.460869565217391e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.54679802955665e-06,
+      "loss": 0.0002,
       "step": 19060
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.443478260869565e-06,
-      "loss": 0.0002,
+      "epoch": 4.02,
+      "learning_rate": 3.529411764705882e-06,
+      "loss": 0.0001,
       "step": 19061
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.426086956521739e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.5120254998551143e-06,
+      "loss": 0.0832,
       "step": 19062
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.4086956521739128e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.4946392350043462e-06,
+      "loss": 0.0,
       "step": 19063
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.391304347826087e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.4772529701535785e-06,
+      "loss": 0.0001,
       "step": 19064
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.37391304347826e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.4598667053028104e-06,
       "loss": 0.0001,
       "step": 19065
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.3565217391304343e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.4424804404520427e-06,
+      "loss": 0.0002,
       "step": 19066
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.3391304347826085e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.425094175601275e-06,
       "loss": 0.0001,
       "step": 19067
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.321739130434782e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.407707910750507e-06,
+      "loss": 0.0,
       "step": 19068
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.3043478260869563e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.390321645899739e-06,
       "loss": 0.0001,
       "step": 19069
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.28695652173913e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.372935381048971e-06,
       "loss": 0.0,
       "step": 19070
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.269565217391304e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.3555491161982034e-06,
+      "loss": 0.0001,
       "step": 19071
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.252173913043478e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.3381628513474352e-06,
       "loss": 0.0,
       "step": 19072
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.234782608695652e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.3207765864966675e-06,
       "loss": 0.0001,
       "step": 19073
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.217391304347826e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.3033903216459e-06,
+      "loss": 0.0,
       "step": 19074
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.1999999999999994e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.2860040567951317e-06,
       "loss": 0.0001,
       "step": 19075
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.1826086956521736e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.268617791944364e-06,
       "loss": 0.0,
       "step": 19076
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 3.1652173913043473e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.2512315270935955e-06,
+      "loss": 0.0,
       "step": 19077
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.1478260869565214e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.233845262242828e-06,
       "loss": 0.0,
       "step": 19078
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.1304347826086955e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.2164589973920596e-06,
       "loss": 0.0,
       "step": 19079
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.1130434782608693e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.199072732541292e-06,
+      "loss": 0.0,
       "step": 19080
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.0956521739130434e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.1816864676905247e-06,
       "loss": 0.0001,
       "step": 19081
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.078260869565217e-06,
-      "loss": 0.0032,
+      "epoch": 4.02,
+      "learning_rate": 3.164300202839756e-06,
+      "loss": 0.0,
       "step": 19082
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.0608695652173912e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.1469139379889884e-06,
+      "loss": 0.0,
       "step": 19083
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.0434782608695645e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.1295276731382203e-06,
+      "loss": 0.0,
       "step": 19084
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.0260869565217387e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.1121414082874526e-06,
+      "loss": 0.0002,
       "step": 19085
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 3.008695652173913e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.0947551434366845e-06,
+      "loss": 0.0,
       "step": 19086
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.9913043478260865e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.0773688785859168e-06,
+      "loss": 0.0002,
       "step": 19087
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.9739130434782606e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 3.059982613735149e-06,
+      "loss": 0.0,
       "step": 19088
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.9565217391304344e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.042596348884381e-06,
+      "loss": 0.0002,
       "step": 19089
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.9391304347826085e-06,
+      "epoch": 4.02,
+      "learning_rate": 3.0252100840336132e-06,
       "loss": 0.0002,
       "step": 19090
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.921739130434782e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 3.007823819182845e-06,
+      "loss": 0.0002,
       "step": 19091
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.9043478260869563e-06,
-      "loss": 0.0002,
+      "epoch": 4.02,
+      "learning_rate": 2.9904375543320774e-06,
+      "loss": 0.0001,
       "step": 19092
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.8869565217391305e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 2.9730512894813093e-06,
+      "loss": 0.0002,
       "step": 19093
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.869565217391304e-06,
-      "loss": 0.0001,
+      "epoch": 4.02,
+      "learning_rate": 2.9556650246305416e-06,
+      "loss": 0.0017,
       "step": 19094
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.8521739130434783e-06,
+      "epoch": 4.02,
+      "learning_rate": 2.938278759779774e-06,
       "loss": 0.0001,
       "step": 19095
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.8347826086956516e-06,
-      "loss": 0.0,
+      "epoch": 4.02,
+      "learning_rate": 2.9208924949290058e-06,
+      "loss": 0.0001,
       "step": 19096
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.8173913043478257e-06,
-      "loss": 0.0004,
+      "epoch": 4.03,
+      "learning_rate": 2.903506230078238e-06,
+      "loss": 0.0002,
       "step": 19097
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.8e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.88611996522747e-06,
+      "loss": 0.0021,
       "step": 19098
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.7826086956521736e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.8687337003767022e-06,
+      "loss": 0.0001,
       "step": 19099
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.7652173913043477e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.851347435525934e-06,
+      "loss": 0.0002,
       "step": 19100
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.7478260869565214e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.8339611706751664e-06,
       "loss": 0.0001,
       "step": 19101
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.7304347826086956e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.8165749058243987e-06,
+      "loss": 0.0001,
       "step": 19102
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.7130434782608693e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.7991886409736306e-06,
+      "loss": 0.0001,
       "step": 19103
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.6956521739130434e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.781802376122863e-06,
+      "loss": 0.0,
       "step": 19104
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.6782608695652175e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.7644161112720948e-06,
       "loss": 0.0001,
       "step": 19105
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.660869565217391e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.747029846421327e-06,
+      "loss": 0.0,
       "step": 19106
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.643478260869565e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.729643581570559e-06,
       "loss": 0.0001,
       "step": 19107
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.6260869565217387e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.7122573167197913e-06,
       "loss": 0.0001,
       "step": 19108
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.608695652173913e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.6948710518690236e-06,
       "loss": 0.0001,
       "step": 19109
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.591304347826087e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.6774847870182554e-06,
+      "loss": 0.0,
       "step": 19110
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.5739130434782607e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.6600985221674877e-06,
       "loss": 0.0,
       "step": 19111
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.556521739130435e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.6427122573167196e-06,
       "loss": 0.0001,
       "step": 19112
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.5391304347826085e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.625325992465952e-06,
       "loss": 0.0001,
       "step": 19113
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.5217391304347826e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.6079397276151838e-06,
+      "loss": 0.0001,
       "step": 19114
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 2.504347826086956e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.590553462764416e-06,
       "loss": 0.0001,
       "step": 19115
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.48695652173913e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.5731671979136475e-06,
+      "loss": 0.0004,
       "step": 19116
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.469565217391304e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.5557809330628803e-06,
+      "loss": 0.0,
       "step": 19117
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.452173913043478e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.5383946682121126e-06,
+      "loss": 0.0001,
       "step": 19118
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.434782608695652e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.521008403361344e-06,
+      "loss": 0.0,
       "step": 19119
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.4173913043478258e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.5036221385105767e-06,
+      "loss": 0.0001,
       "step": 19120
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.4e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.486235873659808e-06,
+      "loss": 0.0,
       "step": 19121
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.3826086956521736e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.4688496088090405e-06,
+      "loss": 0.0001,
       "step": 19122
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.3652173913043477e-06,
-      "loss": 0.0092,
+      "epoch": 4.03,
+      "learning_rate": 2.4514633439582724e-06,
+      "loss": 0.0002,
       "step": 19123
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.347826086956522e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.4340770791075047e-06,
+      "loss": 0.0001,
       "step": 19124
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.330434782608695e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.416690814256737e-06,
+      "loss": 0.0012,
       "step": 19125
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.3130434782608693e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.399304549405969e-06,
+      "loss": 0.0001,
       "step": 19126
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.2956521739130434e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.381918284555201e-06,
       "loss": 0.0,
       "step": 19127
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.278260869565217e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.364532019704433e-06,
+      "loss": 0.0001,
       "step": 19128
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.2608695652173913e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.3471457548536653e-06,
       "loss": 0.0,
       "step": 19129
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.243478260869565e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.3297594900028976e-06,
+      "loss": 0.0002,
       "step": 19130
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.2260869565217387e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.3123732251521295e-06,
       "loss": 0.0001,
       "step": 19131
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.208695652173913e-06,
-      "loss": 0.0001,
+      "epoch": 4.03,
+      "learning_rate": 2.294986960301362e-06,
+      "loss": 0.0,
       "step": 19132
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.191304347826087e-06,
-      "loss": 0.0,
+      "epoch": 4.03,
+      "learning_rate": 2.2776006954505937e-06,
+      "loss": 0.0002,
       "step": 19133
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.1739130434782607e-06,
+      "epoch": 4.03,
+      "learning_rate": 2.260214430599826e-06,
       "loss": 0.0001,
       "step": 19134
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.1565217391304344e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.2428281657490583e-06,
+      "loss": 0.0001,
       "step": 19135
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.1391304347826085e-06,
+      "epoch": 4.04,
+      "learning_rate": 2.22544190089829e-06,
       "loss": 0.0001,
       "step": 19136
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.1217391304347822e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 2.2080556360475225e-06,
+      "loss": 0.0002,
       "step": 19137
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.1043478260869564e-06,
-      "loss": 0.0002,
+      "epoch": 4.04,
+      "learning_rate": 2.1906693711967543e-06,
+      "loss": 0.0003,
       "step": 19138
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.0869565217391305e-06,
-      "loss": 0.0002,
+      "epoch": 4.04,
+      "learning_rate": 2.173283106345986e-06,
+      "loss": 0.0001,
       "step": 19139
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.0695652173913042e-06,
+      "epoch": 4.04,
+      "learning_rate": 2.1558968414952185e-06,
       "loss": 0.0001,
       "step": 19140
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.052173913043478e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.138510576644451e-06,
+      "loss": 0.0001,
       "step": 19141
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.034782608695652e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 2.1211243117936827e-06,
+      "loss": 0.0002,
       "step": 19142
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.0173913043478258e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.103738046942915e-06,
+      "loss": 0.0002,
       "step": 19143
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.086351782092147e-06,
+      "loss": 0.0001,
       "step": 19144
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.9826086956521736e-06,
+      "epoch": 4.04,
+      "learning_rate": 2.068965517241379e-06,
       "loss": 0.0,
       "step": 19145
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.9652173913043478e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.051579252390611e-06,
+      "loss": 0.0001,
       "step": 19146
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.9478260869565215e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.0341929875398433e-06,
+      "loss": 0.0001,
       "step": 19147
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.9304347826086956e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 2.0168067226890756e-06,
+      "loss": 0.0002,
       "step": 19148
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.9130434782608693e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.9994204578383075e-06,
       "loss": 0.0001,
       "step": 19149
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.8956521739130432e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.98203419298754e-06,
+      "loss": 0.0001,
       "step": 19150
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.8782608695652174e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.9646479281367717e-06,
+      "loss": 0.0002,
       "step": 19151
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.8608695652173913e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.947261663286004e-06,
       "loss": 0.0001,
       "step": 19152
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.843478260869565e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.929875398435236e-06,
+      "loss": 0.0001,
       "step": 19153
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.826086956521739e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.912489133584468e-06,
+      "loss": 0.0001,
       "step": 19154
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.8086956521739129e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.8951028687337003e-06,
+      "loss": 0.0002,
       "step": 19155
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.7913043478260868e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.8777166038829323e-06,
       "loss": 0.0,
       "step": 19156
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.773913043478261e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.8603303390321644e-06,
+      "loss": 0.0,
       "step": 19157
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.7565217391304346e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.8429440741813965e-06,
+      "loss": 0.0002,
       "step": 19158
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.7391304347826085e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.8255578093306286e-06,
+      "loss": 0.0001,
       "step": 19159
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.7217391304347825e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.8081715444798607e-06,
+      "loss": 0.0001,
       "step": 19160
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.7043478260869564e-06,
-      "loss": 0.0002,
+      "epoch": 4.04,
+      "learning_rate": 1.7907852796290928e-06,
+      "loss": 0.0001,
       "step": 19161
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.68695652173913e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.773399014778325e-06,
       "loss": 0.0001,
       "step": 19162
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.6695652173913042e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.7560127499275572e-06,
+      "loss": 0.0,
       "step": 19163
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.6521739130434782e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.7386264850767893e-06,
+      "loss": 0.0,
       "step": 19164
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.634782608695652e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.7212402202260213e-06,
       "loss": 0.0,
       "step": 19165
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.617391304347826e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.7038539553752534e-06,
       "loss": 0.0001,
       "step": 19166
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.5999999999999997e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.6864676905244855e-06,
+      "loss": 0.0003,
       "step": 19167
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.5826086956521736e-06,
+      "epoch": 4.04,
+      "learning_rate": 1.6690814256737176e-06,
       "loss": 0.0,
       "step": 19168
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.5652173913043478e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.65169516082295e-06,
+      "loss": 0.0,
       "step": 19169
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.5478260869565217e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.634308895972182e-06,
+      "loss": 0.0002,
       "step": 19170
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.5304347826086956e-06,
-      "loss": 0.0001,
+      "epoch": 4.04,
+      "learning_rate": 1.616922631121414e-06,
+      "loss": 0.0,
       "step": 19171
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.5130434782608693e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.599536366270646e-06,
+      "loss": 0.0001,
       "step": 19172
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.4956521739130433e-06,
-      "loss": 0.0,
+      "epoch": 4.04,
+      "learning_rate": 1.582150101419878e-06,
+      "loss": 0.0001,
       "step": 19173
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.4782608695652172e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.5647638365691101e-06,
+      "loss": 0.0001,
       "step": 19174
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.460869565217391e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.5473775717183422e-06,
+      "loss": 0.0002,
       "step": 19175
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.4434782608695652e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.5299913068675745e-06,
+      "loss": 0.0,
       "step": 19176
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.4260869565217392e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.5126050420168066e-06,
+      "loss": 0.0,
       "step": 19177
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.4086956521739129e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.4952187771660387e-06,
+      "loss": 0.0002,
       "step": 19178
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.3913043478260868e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.4778325123152708e-06,
       "loss": 0.0001,
       "step": 19179
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.3739130434782607e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.4604462474645029e-06,
       "loss": 0.0,
       "step": 19180
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.3565217391304346e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.443059982613735e-06,
       "loss": 0.0001,
       "step": 19181
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.3391304347826088e-06,
-      "loss": 0.0013,
+      "epoch": 4.05,
+      "learning_rate": 1.425673717762967e-06,
+      "loss": 0.0,
       "step": 19182
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.3217391304347825e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.4082874529121994e-06,
+      "loss": 0.0003,
       "step": 19183
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.3043478260869564e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.3909011880614315e-06,
+      "loss": 0.0001,
       "step": 19184
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.2869565217391303e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.3735149232106635e-06,
+      "loss": 0.0,
       "step": 19185
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.2695652173913043e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.3561286583598956e-06,
       "loss": 0.0,
       "step": 19186
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.252173913043478e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.3387423935091277e-06,
       "loss": 0.0,
       "step": 19187
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.234782608695652e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.3213561286583598e-06,
+      "loss": 0.0,
       "step": 19188
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.217391304347826e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.3039698638075919e-06,
       "loss": 0.0001,
       "step": 19189
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.2e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.2865835989568238e-06,
+      "loss": 0.0,
       "step": 19190
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.1826086956521739e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.2691973341060563e-06,
+      "loss": 0.0001,
       "step": 19191
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.1652173913043476e-06,
-      "loss": 0.0783,
+      "epoch": 4.05,
+      "learning_rate": 1.2518110692552884e-06,
+      "loss": 0.0001,
       "step": 19192
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.1478260869565217e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.2344248044045202e-06,
+      "loss": 0.0,
       "step": 19193
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.1304347826086956e-06,
+      "epoch": 4.05,
+      "learning_rate": 1.2170385395537523e-06,
       "loss": 0.0001,
       "step": 19194
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.1130434782608693e-06,
-      "loss": 0.0002,
+      "epoch": 4.05,
+      "learning_rate": 1.1996522747029844e-06,
+      "loss": 0.0001,
       "step": 19195
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.0956521739130435e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.1822660098522165e-06,
+      "loss": 0.0,
       "step": 19196
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.0782608695652172e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.1648797450014488e-06,
+      "loss": 0.0001,
       "step": 19197
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.0608695652173911e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.147493480150681e-06,
+      "loss": 0.0002,
       "step": 19198
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.0434782608695653e-06,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.130107215299913e-06,
+      "loss": 0.0002,
       "step": 19199
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.026086956521739e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.112720950449145e-06,
+      "loss": 0.0001,
       "step": 19200
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.0086956521739129e-06,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 1.0953346855983772e-06,
+      "loss": 0.0001,
       "step": 19201
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 9.913043478260868e-07,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.0779484207476093e-06,
+      "loss": 0.0,
       "step": 19202
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 9.739130434782607e-07,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.0605621558968413e-06,
+      "loss": 0.0,
       "step": 19203
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 9.565217391304347e-07,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.0431758910460734e-06,
+      "loss": 0.0,
       "step": 19204
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 9.391304347826087e-07,
+      "epoch": 4.05,
+      "learning_rate": 1.0257896261953055e-06,
       "loss": 0.0,
       "step": 19205
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 9.217391304347825e-07,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 1.0084033613445378e-06,
+      "loss": 0.0,
       "step": 19206
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 9.043478260869564e-07,
-      "loss": 0.2202,
+      "epoch": 4.05,
+      "learning_rate": 9.9101709649377e-07,
+      "loss": 0.0,
       "step": 19207
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 8.869565217391305e-07,
-      "loss": 0.0,
+      "epoch": 4.05,
+      "learning_rate": 9.73630831643002e-07,
+      "loss": 0.0002,
       "step": 19208
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 8.695652173913043e-07,
-      "loss": 0.0001,
+      "epoch": 4.05,
+      "learning_rate": 9.56244566792234e-07,
+      "loss": 0.0,
       "step": 19209
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 8.521739130434782e-07,
+      "epoch": 4.05,
+      "learning_rate": 9.388583019414662e-07,
       "loss": 0.0001,
       "step": 19210
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 8.347826086956521e-07,
+      "epoch": 4.05,
+      "learning_rate": 9.214720370906983e-07,
       "loss": 0.0001,
       "step": 19211
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 8.17391304347826e-07,
+      "epoch": 4.06,
+      "learning_rate": 9.040857722399303e-07,
       "loss": 0.0001,
       "step": 19212
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 7.999999999999999e-07,
+      "epoch": 4.06,
+      "learning_rate": 8.866995073891625e-07,
       "loss": 0.0001,
       "step": 19213
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 7.826086956521739e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 8.693132425383946e-07,
+      "loss": 0.0002,
       "step": 19214
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 7.652173913043478e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 8.519269776876267e-07,
+      "loss": 0.0001,
       "step": 19215
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 7.478260869565216e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 8.345407128368588e-07,
+      "loss": 0.0,
       "step": 19216
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 7.304347826086956e-07,
+      "epoch": 4.06,
+      "learning_rate": 8.17154447986091e-07,
       "loss": 0.0001,
       "step": 19217
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 7.130434782608696e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 7.99768183135323e-07,
+      "loss": 0.0,
       "step": 19218
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.956521739130434e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 7.823819182845551e-07,
+      "loss": 0.0,
       "step": 19219
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.782608695652173e-07,
+      "epoch": 4.06,
+      "learning_rate": 7.649956534337873e-07,
       "loss": 0.0,
       "step": 19220
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.608695652173912e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 7.476093885830194e-07,
+      "loss": 0.0001,
       "step": 19221
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.434782608695652e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 7.302231237322514e-07,
+      "loss": 0.0,
       "step": 19222
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.26086956521739e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 7.128368588814835e-07,
+      "loss": 0.0,
       "step": 19223
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 6.08695652173913e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 6.954505940307157e-07,
+      "loss": 0.0,
       "step": 19224
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 5.913043478260869e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 6.780643291799478e-07,
+      "loss": 0.0002,
       "step": 19225
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 5.739130434782609e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 6.606780643291799e-07,
+      "loss": 0.0,
       "step": 19226
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 5.565217391304347e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 6.432917994784119e-07,
+      "loss": 0.0,
       "step": 19227
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 5.391304347826086e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 6.259055346276442e-07,
+      "loss": 0.0001,
       "step": 19228
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 5.217391304347826e-07,
+      "epoch": 4.06,
+      "learning_rate": 6.085192697768762e-07,
       "loss": 0.0,
       "step": 19229
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 5.043478260869564e-07,
+      "epoch": 4.06,
+      "learning_rate": 5.911330049261083e-07,
       "loss": 0.0001,
       "step": 19230
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 4.869565217391304e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 5.737467400753404e-07,
+      "loss": 0.0,
       "step": 19231
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 4.6956521739130434e-07,
-      "loss": 0.0002,
+      "epoch": 4.06,
+      "learning_rate": 5.563604752245725e-07,
+      "loss": 0.0003,
       "step": 19232
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 4.521739130434782e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 5.389742103738046e-07,
+      "loss": 0.0001,
       "step": 19233
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 4.3478260869565214e-07,
+      "epoch": 4.06,
+      "learning_rate": 5.215879455230367e-07,
       "loss": 0.0001,
       "step": 19234
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 4.1739130434782606e-07,
+      "epoch": 4.06,
+      "learning_rate": 5.042016806722689e-07,
       "loss": 0.0001,
       "step": 19235
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 3.9999999999999993e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 4.86815415821501e-07,
+      "loss": 0.0001,
       "step": 19236
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 3.826086956521739e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 4.694291509707331e-07,
+      "loss": 0.0,
       "step": 19237
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 3.652173913043478e-07,
+      "epoch": 4.06,
+      "learning_rate": 4.520428861199652e-07,
       "loss": 0.0001,
       "step": 19238
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 3.478260869565217e-07,
+      "epoch": 4.06,
+      "learning_rate": 4.346566212691973e-07,
       "loss": 0.0001,
       "step": 19239
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 3.304347826086956e-07,
+      "epoch": 4.06,
+      "learning_rate": 4.172703564184294e-07,
       "loss": 0.0,
       "step": 19240
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 3.130434782608695e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 3.998840915676615e-07,
+      "loss": 0.0001,
       "step": 19241
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.9565217391304347e-07,
+      "epoch": 4.06,
+      "learning_rate": 3.8249782671689363e-07,
       "loss": 0.0,
       "step": 19242
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.7826086956521734e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 3.651115618661257e-07,
+      "loss": 0.0001,
       "step": 19243
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.608695652173913e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 3.4772529701535786e-07,
+      "loss": 0.0,
       "step": 19244
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.434782608695652e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 3.3033903216458995e-07,
+      "loss": 0.0,
       "step": 19245
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.260869565217391e-07,
-      "loss": 0.0,
+      "epoch": 4.06,
+      "learning_rate": 3.129527673138221e-07,
+      "loss": 0.0002,
       "step": 19246
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.0869565217391303e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 2.9556650246305413e-07,
+      "loss": 0.0,
       "step": 19247
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 1.9130434782608695e-07,
+      "epoch": 4.06,
+      "learning_rate": 2.7818023761228627e-07,
       "loss": 0.0001,
       "step": 19248
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 1.7391304347826085e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 2.6079397276151836e-07,
+      "loss": 0.0002,
       "step": 19249
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 1.5652173913043475e-07,
-      "loss": 0.0001,
+      "epoch": 4.06,
+      "learning_rate": 2.434077079107505e-07,
+      "loss": 0.0015,
       "step": 19250
     },
     {
-      "epoch": 5.0,
-      "step": 19250,
-      "total_flos": 4.686304027304379e+20,
-      "train_loss": 0.1620459199016551,
-      "train_runtime": 43631.9674,
-      "train_samples_per_second": 28.237,
-      "train_steps_per_second": 0.441
+      "epoch": 4.07,
+      "learning_rate": 2.260214430599826e-07,
+      "loss": 0.0,
+      "step": 19251
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.086351782092147e-07,
+      "loss": 0.0001,
+      "step": 19252
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.9124891335844682e-07,
+      "loss": 0.0,
+      "step": 19253
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7386264850767893e-07,
+      "loss": 0.0002,
+      "step": 19254
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.5647638365691105e-07,
+      "loss": 0.0001,
+      "step": 19255
+    },
+    {
+      "epoch": 4.07,
+      "step": 19255,
+      "total_flos": 4.709738606122584e+20,
+      "train_loss": 5.854813561474184e-06,
+      "train_runtime": 574.8329,
+      "train_samples_per_second": 2143.301,
+      "train_steps_per_second": 33.497
     }
   ],
-  "max_steps": 19250,
+  "max_steps": 19255,
   "num_train_epochs": 5,
-  "total_flos": 4.686304027304379e+20,
+  "total_flos": 4.709738606122584e+20,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad89e76b0670cd2e61d285d1c083b63650b30ff0824b129c7770656484dc7c02
+oid sha256:46a8db1eb3abf552e505c4d382ce37db4dc5d44734d8cd766198dc27b5c70d61
 size 3247