End of training

all_results.json

@@ -1,33 +1,41 @@
 {
     "epoch": 10.0,
-    "eval_PREDICATE_f1": 0.9465648854961832,
-    "eval_PREDICATE_number": 129,
-    "eval_PREDICATE_precision": 0.9323308270676691,
-    "eval_PREDICATE_recall": 0.9612403100775194,
-    "eval_loss": 0.2416924387216568,
-    "eval_overall_accuracy": 0.9583558680367766,
-    "eval_overall_f1": 0.8395061728395061,
-    "eval_overall_precision": 0.8429752066115702,
-    "eval_overall_recall": 0.8360655737704918,
-    "eval_runtime": 0.5304,
-    "eval_samples": 96,
-    "eval_samples_per_second": 180.983,
-    "eval_steps_per_second": 22.623,
+    "eval_PREDICATE_f1": 0.9649805447470817,
+    "eval_PREDICATE_number": 126,
+    "eval_PREDICATE_precision": 0.9465648854961832,
+    "eval_PREDICATE_recall": 0.9841269841269841,
+    "eval_loss": 0.18749289214611053,
+    "eval_overall_accuracy": 0.9576365663322185,
+    "eval_overall_f1": 0.8493150684931506,
+    "eval_overall_precision": 0.8188679245283019,
+    "eval_overall_recall": 0.8821138211382114,
+    "eval_runtime": 1.816,
+    "eval_samples": 99,
+    "eval_samples_per_second": 54.515,
+    "eval_steps_per_second": 7.159,
     "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-    "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
+    "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
     "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
     "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6917293233082706,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7666666666666667,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6301369863013698,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7555555555555555,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6938775510204082,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8292682926829268,
-    "train_loss": 0.10899043403289936,
-    "train_runtime": 173.8981,
-    "train_samples": 859,
-    "train_samples_per_second": 49.397,
-    "train_steps_per_second": 3.105
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7123287671232876,
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7027027027027027,
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7222222222222222,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8817204301075268,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8367346938775511,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9318181818181818,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+    "train_loss": 0.13130194339568593,
+    "train_runtime": 174.9188,
+    "train_samples": 884,
+    "train_samples_per_second": 50.538,
+    "train_steps_per_second": 3.201
 }

config.json

@@ -16,7 +16,10 @@
     "2": "B-\u041a\u0410\u0423\u0417\u0410\u0422\u0418\u0412",
     "3": "B-\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420",
     "4": "B-\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420",
-    "5": "O"
+    "5": "B-\u0434\u0440\u0443\u0433\u043e\u0435",
+    "6": "B-\u043a\u0430\u0443\u0437\u0430\u0442\u043e\u0440",
+    "7": "B-\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440",
+    "8": "O"
   },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
@@ -26,7 +29,10 @@
     "B-\u041a\u0410\u0423\u0417\u0410\u0422\u0418\u0412": 2,
     "B-\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420": 3,
     "B-\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420": 4,
-    "O": 5
+    "B-\u0434\u0440\u0443\u0433\u043e\u0435": 5,
+    "B-\u043a\u0430\u0443\u0437\u0430\u0442\u043e\u0440": 6,
+    "B-\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440": 7,
+    "O": 8
   },
   "layer_norm_eps": 1e-12,
   "max_position_embeddings": 512,

eval_results.json

@@ -1,28 +1,36 @@
 {
     "epoch": 10.0,
-    "eval_PREDICATE_f1": 0.9465648854961832,
-    "eval_PREDICATE_number": 129,
-    "eval_PREDICATE_precision": 0.9323308270676691,
-    "eval_PREDICATE_recall": 0.9612403100775194,
-    "eval_loss": 0.2416924387216568,
-    "eval_overall_accuracy": 0.9583558680367766,
-    "eval_overall_f1": 0.8395061728395061,
-    "eval_overall_precision": 0.8429752066115702,
-    "eval_overall_recall": 0.8360655737704918,
-    "eval_runtime": 0.5304,
-    "eval_samples": 96,
-    "eval_samples_per_second": 180.983,
-    "eval_steps_per_second": 22.623,
+    "eval_PREDICATE_f1": 0.9649805447470817,
+    "eval_PREDICATE_number": 126,
+    "eval_PREDICATE_precision": 0.9465648854961832,
+    "eval_PREDICATE_recall": 0.9841269841269841,
+    "eval_loss": 0.18749289214611053,
+    "eval_overall_accuracy": 0.9576365663322185,
+    "eval_overall_f1": 0.8493150684931506,
+    "eval_overall_precision": 0.8188679245283019,
+    "eval_overall_recall": 0.8821138211382114,
+    "eval_runtime": 1.816,
+    "eval_samples": 99,
+    "eval_samples_per_second": 54.515,
+    "eval_steps_per_second": 7.159,
     "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-    "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
+    "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
     "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
     "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6917293233082706,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7666666666666667,
-    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6301369863013698,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7555555555555555,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6938775510204082,
-    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8292682926829268
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7123287671232876,
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7027027027027027,
+    "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7222222222222222,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8817204301075268,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8367346938775511,
+    "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9318181818181818,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+    "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+    "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab14cf1798a2cb5934d60be7d84ff133f9ea70fd7878d397a0bd97db208d990d
-size 710968817
+oid sha256:7af8e323efcb030ab0858fb32ffefc6630dfeb0bcd200f59f45e8017ab2e3616
+size 710978033

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 10.0,
-    "train_loss": 0.10899043403289936,
-    "train_runtime": 173.8981,
-    "train_samples": 859,
-    "train_samples_per_second": 49.397,
-    "train_steps_per_second": 3.105
+    "train_loss": 0.13130194339568593,
+    "train_runtime": 174.9188,
+    "train_samples": 884,
+    "train_samples_per_second": 50.538,
+    "train_steps_per_second": 3.201
 }

trainer_state.json

@@ -2,958 +2,1050 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 10.0,
-  "global_step": 540,
+  "global_step": 560,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.02,
-      "learning_rate": 1.5151515151515152e-06,
-      "loss": 1.7524,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 2.233,
       "step": 1
     },
     {
       "epoch": 0.09,
-      "learning_rate": 7.5757575757575764e-06,
-      "loss": 1.7511,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 2.1869,
       "step": 5
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.5151515151515153e-05,
-      "loss": 1.5404,
+      "epoch": 0.18,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 1.9259,
       "step": 10
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 2.272727272727273e-05,
-      "loss": 1.118,
+      "epoch": 0.27,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 1.4271,
       "step": 15
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.0303030303030306e-05,
-      "loss": 0.6504,
+      "epoch": 0.36,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.7441,
       "step": 20
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 3.787878787878788e-05,
-      "loss": 0.8655,
+      "epoch": 0.45,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 0.9447,
       "step": 25
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.545454545454546e-05,
-      "loss": 0.4171,
+      "epoch": 0.54,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.531,
       "step": 30
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 4.999808023410233e-05,
-      "loss": 0.3723,
+      "epoch": 0.62,
+      "learning_rate": 4.9999554100506465e-05,
+      "loss": 0.428,
       "step": 35
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.997648625366471e-05,
-      "loss": 0.3472,
+      "epoch": 0.71,
+      "learning_rate": 4.998394928831034e-05,
+      "loss": 0.4043,
       "step": 40
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.993091938082206e-05,
-      "loss": 0.1754,
+      "epoch": 0.8,
+      "learning_rate": 4.9946065404995485e-05,
+      "loss": 0.2392,
       "step": 45
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 4.986142335149978e-05,
-      "loss": 0.2462,
+      "epoch": 0.89,
+      "learning_rate": 4.988593623296038e-05,
+      "loss": 0.3225,
       "step": 50
     },
     {
-      "epoch": 1.0,
-      "eval_PREDICATE_f1": 0.9416058394160584,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.8896551724137931,
-      "eval_PREDICATE_recall": 1.0,
-      "eval_loss": 0.15542413294315338,
-      "eval_overall_accuracy": 0.9486208761492698,
-      "eval_overall_f1": 0.8,
-      "eval_overall_precision": 0.8135593220338984,
-      "eval_overall_recall": 0.7868852459016393,
-      "eval_runtime": 0.5007,
-      "eval_samples_per_second": 191.714,
-      "eval_steps_per_second": 23.964,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6115702479338844,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7708333333333334,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.5068493150684932,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.6190476190476191,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6046511627906976,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.6341463414634146,
-      "step": 54
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 4.976806486926926e-05,
-      "loss": 0.216,
+      "epoch": 0.98,
+      "learning_rate": 4.980361539151416e-05,
+      "loss": 0.289,
       "step": 55
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 4.9650933541324506e-05,
-      "loss": 0.1322,
+      "epoch": 1.0,
+      "eval_PREDICATE_f1": 0.9498069498069498,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.924812030075188,
+      "eval_PREDICATE_recall": 0.9761904761904762,
+      "eval_loss": 0.16499629616737366,
+      "eval_overall_accuracy": 0.947603121516165,
+      "eval_overall_f1": 0.7853881278538812,
+      "eval_overall_precision": 0.8958333333333334,
+      "eval_overall_recall": 0.6991869918699187,
+      "eval_runtime": 0.4866,
+      "eval_samples_per_second": 203.444,
+      "eval_steps_per_second": 26.715,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.42424242424242425,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7777777777777778,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.2916666666666667,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7368421052631579,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.875,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.6363636363636364,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 56
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.969917628906234e-05,
+      "loss": 0.1803,
       "step": 60
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 4.951014179247555e-05,
-      "loss": 0.1814,
+      "epoch": 1.16,
+      "learning_rate": 4.9572712057646e-05,
+      "loss": 0.2196,
       "step": 65
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 4.934582475724101e-05,
-      "loss": 0.1226,
+      "epoch": 1.25,
+      "learning_rate": 4.9424335469892594e-05,
+      "loss": 0.1454,
       "step": 70
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.9158140150143484e-05,
-      "loss": 0.1595,
+      "epoch": 1.34,
+      "learning_rate": 4.925417883845266e-05,
+      "loss": 0.1431,
       "step": 75
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 4.8947268114332274e-05,
-      "loss": 0.172,
+      "epoch": 1.43,
+      "learning_rate": 4.9062393898011904e-05,
+      "loss": 0.2021,
       "step": 80
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 4.8713411048678635e-05,
-      "loss": 0.1202,
+      "epoch": 1.52,
+      "learning_rate": 4.884915166998405e-05,
+      "loss": 0.1312,
       "step": 85
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 4.8456793413509634e-05,
-      "loss": 0.1816,
+      "epoch": 1.61,
+      "learning_rate": 4.8614642310004974e-05,
+      "loss": 0.1555,
       "step": 90
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 4.817766151516693e-05,
-      "loss": 0.164,
+      "epoch": 1.7,
+      "learning_rate": 4.835907493836421e-05,
+      "loss": 0.2674,
       "step": 95
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 4.787628326959747e-05,
-      "loss": 0.1345,
+      "epoch": 1.79,
+      "learning_rate": 4.808267745352502e-05,
+      "loss": 0.1073,
       "step": 100
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.755294794520276e-05,
-      "loss": 0.1863,
+      "epoch": 1.88,
+      "learning_rate": 4.778569632889919e-05,
+      "loss": 0.1676,
       "step": 105
     },
     {
-      "epoch": 2.0,
-      "eval_PREDICATE_f1": 0.9446494464944649,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9014084507042254,
-      "eval_PREDICATE_recall": 0.9922480620155039,
-      "eval_loss": 0.1267632097005844,
-      "eval_overall_accuracy": 0.9556517036235803,
-      "eval_overall_f1": 0.8220338983050847,
-      "eval_overall_precision": 0.8508771929824561,
-      "eval_overall_recall": 0.7950819672131147,
-      "eval_runtime": 0.5235,
-      "eval_samples_per_second": 183.39,
-      "eval_steps_per_second": 22.924,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6440677966101694,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.8444444444444444,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.5205479452054794,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.6829268292682927,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6829268292682927,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.6829268292682927,
-      "step": 108
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 4.72079658851938e-05,
-      "loss": 0.0882,
+      "epoch": 1.96,
+      "learning_rate": 4.746839639305808e-05,
+      "loss": 0.1543,
       "step": 110
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 4.684166820971779e-05,
-      "loss": 0.0917,
+      "epoch": 2.0,
+      "eval_PREDICATE_f1": 0.9689922480620156,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.946969696969697,
+      "eval_PREDICATE_recall": 0.9920634920634921,
+      "eval_loss": 0.1350417286157608,
+      "eval_overall_accuracy": 0.9587513935340022,
+      "eval_overall_f1": 0.852589641434263,
+      "eval_overall_precision": 0.8359375,
+      "eval_overall_recall": 0.8699186991869918,
+      "eval_runtime": 0.4934,
+      "eval_samples_per_second": 200.654,
+      "eval_steps_per_second": 26.348,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7019867549668874,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.6708860759493671,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7361111111111112,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8089887640449439,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8181818181818182,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 112
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.713106059357554e-05,
+      "loss": 0.0796,
       "step": 115
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 4.6454406498043105e-05,
-      "loss": 0.1179,
+      "epoch": 2.14,
+      "learning_rate": 4.6773989744713596e-05,
+      "loss": 0.112,
       "step": 120
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 4.604655245110684e-05,
-      "loss": 0.0599,
+      "epoch": 2.23,
+      "learning_rate": 4.639750225917573e-05,
+      "loss": 0.1316,
       "step": 125
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 4.561849753474951e-05,
-      "loss": 0.0923,
+      "epoch": 2.32,
+      "learning_rate": 4.600193386416697e-05,
+      "loss": 0.0973,
       "step": 130
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 4.517065260397887e-05,
-      "loss": 0.0903,
+      "epoch": 2.41,
+      "learning_rate": 4.558763730201413e-05,
+      "loss": 0.1587,
       "step": 135
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 4.470344750862368e-05,
-      "loss": 0.0941,
+      "epoch": 2.5,
+      "learning_rate": 4.515498201561296e-05,
+      "loss": 0.0902,
       "step": 140
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 4.421733068075595e-05,
+      "epoch": 2.59,
+      "learning_rate": 4.470435381898287e-05,
       "loss": 0.0881,
       "step": 145
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 4.371276870427753e-05,
-      "loss": 0.0653,
+      "epoch": 2.68,
+      "learning_rate": 4.423615455322293e-05,
+      "loss": 0.1606,
       "step": 150
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 4.319024586708428e-05,
-      "loss": 0.1036,
+      "epoch": 2.77,
+      "learning_rate": 4.375080172817591e-05,
+      "loss": 0.0843,
       "step": 155
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 4.265026369623761e-05,
-      "loss": 0.0668,
+      "epoch": 2.86,
+      "learning_rate": 4.324872815012005e-05,
+      "loss": 0.0939,
       "step": 160
     },
     {
-      "epoch": 3.0,
-      "eval_PREDICATE_f1": 0.9584905660377359,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9338235294117647,
-      "eval_PREDICATE_recall": 0.9844961240310077,
-      "eval_loss": 0.1287979632616043,
-      "eval_overall_accuracy": 0.9599783666846944,
-      "eval_overall_f1": 0.8458333333333333,
-      "eval_overall_precision": 0.8601694915254238,
-      "eval_overall_recall": 0.8319672131147541,
-      "eval_runtime": 0.4971,
-      "eval_samples_per_second": 193.12,
-      "eval_steps_per_second": 24.14,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6929133858267718,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.8148148148148148,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6027397260273972,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7356321839080461,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6956521739130435,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.7804878048780488,
-      "step": 162
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 4.209334047658956e-05,
-      "loss": 0.0458,
+      "epoch": 2.95,
+      "learning_rate": 4.273038153582034e-05,
+      "loss": 0.1525,
       "step": 165
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 4.152001075332341e-05,
-      "loss": 0.044,
+      "epoch": 3.0,
+      "eval_PREDICATE_f1": 0.9727626459143969,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.9541984732824428,
+      "eval_PREDICATE_recall": 0.9920634920634921,
+      "eval_loss": 0.14091716706752777,
+      "eval_overall_accuracy": 0.9548494983277592,
+      "eval_overall_f1": 0.8396946564885496,
+      "eval_overall_precision": 0.7913669064748201,
+      "eval_overall_recall": 0.8943089430894309,
+      "eval_runtime": 1.7665,
+      "eval_samples_per_second": 56.044,
+      "eval_steps_per_second": 7.359,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6707317073170731,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.5978260869565217,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7638888888888888,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8421052631578948,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.7843137254901961,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9090909090909091,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 168
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.21962241132837e-05,
+      "loss": 0.0602,
       "step": 170
     },
     {
-      "epoch": 3.24,
-      "learning_rate": 4.09308248188874e-05,
-      "loss": 0.0489,
+      "epoch": 3.12,
+      "learning_rate": 4.164673220957387e-05,
+      "loss": 0.0662,
       "step": 175
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 4.0326348184813826e-05,
-      "loss": 0.056,
+      "epoch": 3.21,
+      "learning_rate": 4.108239582605374e-05,
+      "loss": 0.0804,
       "step": 180
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 3.970716103893065e-05,
-      "loss": 0.05,
+      "epoch": 3.3,
+      "learning_rate": 4.050371820143384e-05,
+      "loss": 0.0492,
       "step": 185
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 3.907385768848656e-05,
-      "loss": 0.0403,
+      "epoch": 3.39,
+      "learning_rate": 3.991121536301653e-05,
+      "loss": 0.0752,
       "step": 190
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 3.842704598972384e-05,
-      "loss": 0.0573,
+      "epoch": 3.48,
+      "learning_rate": 3.9305415666536236e-05,
+      "loss": 0.0595,
       "step": 195
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 3.776734676444678e-05,
-      "loss": 0.0624,
+      "epoch": 3.57,
+      "learning_rate": 3.868685932500595e-05,
+      "loss": 0.0829,
       "step": 200
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 3.709539320414544e-05,
-      "loss": 0.0465,
+      "epoch": 3.66,
+      "learning_rate": 3.80560979269902e-05,
+      "loss": 0.0923,
       "step": 205
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 3.641183026224675e-05,
-      "loss": 0.0677,
+      "epoch": 3.75,
+      "learning_rate": 3.7413693944733995e-05,
+      "loss": 0.033,
       "step": 210
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 3.5717314035076355e-05,
-      "loss": 0.039,
+      "epoch": 3.84,
+      "learning_rate": 3.676022023258643e-05,
+      "loss": 0.0674,
       "step": 215
     },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.609625951616623e-05,
+      "loss": 0.088,
+      "step": 220
+    },
     {
       "epoch": 4.0,
-      "eval_PREDICATE_f1": 0.9407407407407407,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.900709219858156,
-      "eval_PREDICATE_recall": 0.9844961240310077,
-      "eval_loss": 0.16953788697719574,
-      "eval_overall_accuracy": 0.9556517036235803,
-      "eval_overall_f1": 0.8309572301425663,
-      "eval_overall_precision": 0.8259109311740891,
-      "eval_overall_recall": 0.8360655737704918,
-      "eval_runtime": 0.5165,
-      "eval_samples_per_second": 185.878,
-      "eval_steps_per_second": 23.235,
+      "eval_PREDICATE_f1": 0.9689922480620156,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.946969696969697,
+      "eval_PREDICATE_recall": 0.9920634920634921,
+      "eval_loss": 0.13493739068508148,
+      "eval_overall_accuracy": 0.959866220735786,
+      "eval_overall_f1": 0.8548707753479126,
+      "eval_overall_precision": 0.8365758754863813,
+      "eval_overall_recall": 0.8739837398373984,
+      "eval_runtime": 0.5164,
+      "eval_samples_per_second": 191.71,
+      "eval_steps_per_second": 25.174,
       "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
       "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
       "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6499999999999999,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.8297872340425532,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.5342465753424658,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.76,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6440677966101694,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.926829268292683,
-      "step": 216
-    },
-    {
-      "epoch": 4.07,
-      "learning_rate": 3.501251113212521e-05,
-      "loss": 0.0332,
-      "step": 220
-    },
-    {
-      "epoch": 4.17,
-      "learning_rate": 3.429809803622551e-05,
-      "loss": 0.0549,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7123287671232876,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7027027027027027,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7222222222222222,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8444444444444444,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8260869565217391,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8636363636363636,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 224
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.5422403872724686e-05,
+      "loss": 0.0347,
       "step": 225
     },
     {
-      "epoch": 4.26,
-      "learning_rate": 3.357476045424997e-05,
-      "loss": 0.0214,
+      "epoch": 4.11,
+      "learning_rate": 3.473925420316946e-05,
+      "loss": 0.042,
       "step": 230
     },
     {
-      "epoch": 4.35,
-      "learning_rate": 3.2843192658957774e-05,
-      "loss": 0.0352,
+      "epoch": 4.2,
+      "learning_rate": 3.404741969621996e-05,
+      "loss": 0.037,
       "step": 235
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 3.210409682261866e-05,
-      "loss": 0.0521,
+      "epoch": 4.29,
+      "learning_rate": 3.334751728517223e-05,
+      "loss": 0.0259,
       "step": 240
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 3.135818234305511e-05,
-      "loss": 0.0267,
+      "epoch": 4.38,
+      "learning_rate": 3.264017109775771e-05,
+      "loss": 0.0471,
       "step": 245
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 3.060616516274921e-05,
-      "loss": 0.0454,
+      "epoch": 4.46,
+      "learning_rate": 3.1926011899586485e-05,
+      "loss": 0.0315,
       "step": 250
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 2.984876708166782e-05,
-      "loss": 0.0234,
+      "epoch": 4.55,
+      "learning_rate": 3.120567653167127e-05,
+      "loss": 0.0374,
       "step": 255
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 2.908671506446566e-05,
-      "loss": 0.034,
+      "epoch": 4.64,
+      "learning_rate": 3.0479807342533722e-05,
+      "loss": 0.0492,
       "step": 260
     },
     {
-      "epoch": 4.91,
-      "learning_rate": 2.8320740542731212e-05,
-      "loss": 0.0198,
+      "epoch": 4.73,
+      "learning_rate": 2.9749051615399637e-05,
+      "loss": 0.0336,
       "step": 265
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.7551578712945208e-05,
-      "loss": 0.0187,
+      "epoch": 4.82,
+      "learning_rate": 2.9014060990993496e-05,
+      "loss": 0.0518,
       "step": 270
     },
     {
-      "epoch": 5.0,
-      "eval_PREDICATE_f1": 0.9465648854961832,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9323308270676691,
-      "eval_PREDICATE_recall": 0.9612403100775194,
-      "eval_loss": 0.1955076903104782,
-      "eval_overall_accuracy": 0.955110870740941,
-      "eval_overall_f1": 0.8195329087048832,
-      "eval_overall_precision": 0.8502202643171806,
-      "eval_overall_recall": 0.7909836065573771,
-      "eval_runtime": 0.5123,
-      "eval_samples_per_second": 187.387,
-      "eval_steps_per_second": 23.423,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6511627906976744,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.75,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.5753424657534246,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.6835443037974684,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.7105263157894737,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.6585365853658537,
-      "step": 270
+      "epoch": 4.91,
+      "learning_rate": 2.827549088644748e-05,
+      "loss": 0.0492,
+      "step": 275
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 2.6779967830825454e-05,
-      "loss": 0.0178,
-      "step": 275
+      "epoch": 5.0,
+      "learning_rate": 2.7533999910842766e-05,
+      "loss": 0.0454,
+      "step": 280
     },
     {
-      "epoch": 5.19,
-      "learning_rate": 2.600664850273538e-05,
-      "loss": 0.0231,
+      "epoch": 5.0,
+      "eval_PREDICATE_f1": 0.9765625,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.9615384615384616,
+      "eval_PREDICATE_recall": 0.9920634920634921,
+      "eval_loss": 0.15792229771614075,
+      "eval_overall_accuracy": 0.959866220735786,
+      "eval_overall_f1": 0.8537074148296593,
+      "eval_overall_precision": 0.841897233201581,
+      "eval_overall_recall": 0.8658536585365854,
+      "eval_runtime": 0.5032,
+      "eval_samples_per_second": 196.725,
+      "eval_steps_per_second": 25.833,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7218045112781954,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7868852459016393,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6666666666666666,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8333333333333333,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.7692307692307693,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9090909090909091,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
       "step": 280
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 2.5232362974836394e-05,
-      "loss": 0.0149,
+      "epoch": 5.09,
+      "learning_rate": 2.6790249277904582e-05,
+      "loss": 0.0142,
       "step": 285
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 2.4457854420666278e-05,
-      "loss": 0.0237,
+      "epoch": 5.18,
+      "learning_rate": 2.6044902216374496e-05,
+      "loss": 0.0234,
       "step": 290
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 2.3683866227827603e-05,
-      "loss": 0.0134,
+      "epoch": 5.27,
+      "learning_rate": 2.5298623378585884e-05,
+      "loss": 0.0142,
       "step": 295
     },
     {
-      "epoch": 5.56,
-      "learning_rate": 2.2911141284470466e-05,
-      "loss": 0.0154,
+      "epoch": 5.36,
+      "learning_rate": 2.4552078247770007e-05,
+      "loss": 0.0325,
       "step": 300
     },
     {
-      "epoch": 5.65,
-      "learning_rate": 2.214042126625486e-05,
-      "loss": 0.0361,
+      "epoch": 5.45,
+      "learning_rate": 2.3805932544621002e-05,
+      "loss": 0.0394,
       "step": 305
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 2.137244592447658e-05,
-      "loss": 0.0088,
+      "epoch": 5.54,
+      "learning_rate": 2.3060851633649245e-05,
+      "loss": 0.016,
       "step": 310
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 2.0607952376040353e-05,
-      "loss": 0.042,
+      "epoch": 5.62,
+      "learning_rate": 2.2317499929852273e-05,
+      "loss": 0.0354,
       "step": 315
     },
     {
-      "epoch": 5.93,
-      "learning_rate": 1.9847674395961406e-05,
-      "loss": 0.0216,
+      "epoch": 5.71,
+      "learning_rate": 2.1576540306232418e-05,
+      "loss": 0.0132,
       "step": 320
     },
     {
-      "epoch": 6.0,
-      "eval_PREDICATE_f1": 0.9501915708812261,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9393939393939394,
-      "eval_PREDICATE_recall": 0.9612403100775194,
-      "eval_loss": 0.2082541584968567,
-      "eval_overall_accuracy": 0.9561925365062196,
-      "eval_overall_f1": 0.8252631578947369,
-      "eval_overall_precision": 0.8484848484848485,
-      "eval_overall_recall": 0.8032786885245902,
-      "eval_runtime": 0.4958,
-      "eval_samples_per_second": 193.643,
-      "eval_steps_per_second": 24.205,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6717557251908397,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7586206896551724,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6027397260273972,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.6829268292682927,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6829268292682927,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.6829268292682927,
-      "step": 324
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 1.909234171307466e-05,
-      "loss": 0.0091,
+      "epoch": 5.8,
+      "learning_rate": 2.0838633502689545e-05,
+      "loss": 0.0382,
       "step": 325
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 1.8342679309627543e-05,
-      "loss": 0.0164,
+      "epoch": 5.89,
+      "learning_rate": 2.0104437536815883e-05,
+      "loss": 0.0343,
       "step": 330
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 1.7599406725428617e-05,
-      "loss": 0.01,
+      "epoch": 5.98,
+      "learning_rate": 1.9374607117118485e-05,
+      "loss": 0.0297,
       "step": 335
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 1.6863237367220058e-05,
-      "loss": 0.0103,
+      "epoch": 6.0,
+      "eval_PREDICATE_f1": 0.968503937007874,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.9609375,
+      "eval_PREDICATE_recall": 0.9761904761904762,
+      "eval_loss": 0.16979572176933289,
+      "eval_overall_accuracy": 0.9587513935340022,
+      "eval_overall_f1": 0.8509803921568627,
+      "eval_overall_precision": 0.821969696969697,
+      "eval_overall_recall": 0.8821138211382114,
+      "eval_runtime": 0.5095,
+      "eval_samples_per_second": 194.318,
+      "eval_steps_per_second": 25.516,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7083333333333334,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7083333333333334,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7083333333333334,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8453608247422681,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.7735849056603774,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9318181818181818,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.3333333333333333,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.2222222222222222,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.6666666666666666,
+      "step": 336
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.864979305919248e-05,
+      "loss": 0.0104,
       "step": 340
     },
     {
-      "epoch": 6.39,
-      "learning_rate": 1.613487782393661e-05,
-      "loss": 0.0151,
+      "epoch": 6.16,
+      "learning_rate": 1.7930641705365836e-05,
+      "loss": 0.0247,
       "step": 345
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 1.5415027188508573e-05,
-      "loss": 0.0039,
+      "epoch": 6.25,
+      "learning_rate": 1.721779434833299e-05,
+      "loss": 0.0066,
       "step": 350
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 1.4704376386859448e-05,
-      "loss": 0.0211,
+      "epoch": 6.34,
+      "learning_rate": 1.651188665929151e-05,
+      "loss": 0.0246,
       "step": 355
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 1.4003607514742529e-05,
-      "loss": 0.0257,
+      "epoch": 6.43,
+      "learning_rate": 1.5813548121091622e-05,
+      "loss": 0.017,
       "step": 360
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 1.3313393183052747e-05,
-      "loss": 0.0072,
+      "epoch": 6.52,
+      "learning_rate": 1.512340146690402e-05,
+      "loss": 0.0117,
       "step": 365
     },
     {
-      "epoch": 6.85,
-      "learning_rate": 1.2634395872242433e-05,
-      "loss": 0.0171,
+      "epoch": 6.61,
+      "learning_rate": 1.4442062124906763e-05,
+      "loss": 0.0273,
       "step": 370
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 1.1967267296460208e-05,
-      "loss": 0.0176,
+      "epoch": 6.7,
+      "learning_rate": 1.3770137669486171e-05,
+      "loss": 0.0214,
       "step": 375
     },
     {
-      "epoch": 7.0,
-      "eval_PREDICATE_f1": 0.9465648854961832,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9323308270676691,
-      "eval_PREDICATE_recall": 0.9612403100775194,
-      "eval_loss": 0.220278799533844,
-      "eval_overall_accuracy": 0.9578150351541374,
-      "eval_overall_f1": 0.8356997971602433,
-      "eval_overall_precision": 0.8273092369477911,
-      "eval_overall_recall": 0.8442622950819673,
-      "eval_runtime": 0.4918,
-      "eval_samples_per_second": 195.192,
-      "eval_steps_per_second": 24.399,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6906474820143884,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7272727272727273,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6575342465753424,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7472527472527474,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.68,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8292682926829268,
-      "step": 378
-    },
-    {
-      "epoch": 7.04,
-      "learning_rate": 1.131264777802387e-05,
-      "loss": 0.008,
+      "epoch": 6.79,
+      "learning_rate": 1.3108227279441243e-05,
+      "loss": 0.0132,
       "step": 380
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 1.0671165632827097e-05,
-      "loss": 0.0118,
+      "epoch": 6.88,
+      "learning_rate": 1.2456921203674715e-05,
+      "loss": 0.0276,
       "step": 385
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 1.0043436567270312e-05,
-      "loss": 0.0111,
+      "epoch": 6.96,
+      "learning_rate": 1.1816800234847303e-05,
+      "loss": 0.0161,
       "step": 390
     },
     {
-      "epoch": 7.31,
-      "learning_rate": 9.43006308729432e-06,
-      "loss": 0.0085,
+      "epoch": 7.0,
+      "eval_PREDICATE_f1": 0.9725490196078431,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.9612403100775194,
+      "eval_PREDICATE_recall": 0.9841269841269841,
+      "eval_loss": 0.1633814424276352,
+      "eval_overall_accuracy": 0.9637681159420289,
+      "eval_overall_f1": 0.8674698795180722,
+      "eval_overall_precision": 0.8571428571428571,
+      "eval_overall_recall": 0.8780487804878049,
+      "eval_runtime": 0.5173,
+      "eval_samples_per_second": 191.382,
+      "eval_steps_per_second": 25.131,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7338129496402879,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7611940298507462,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7083333333333334,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8695652173913043,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8333333333333334,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9090909090909091,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.2222222222222222,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.16666666666666666,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.3333333333333333,
+      "step": 392
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1188435191464303e-05,
+      "loss": 0.0116,
       "step": 395
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 8.831633920083968e-06,
-      "loss": 0.0163,
+      "epoch": 7.14,
+      "learning_rate": 1.0572386408856552e-05,
+      "loss": 0.0228,
       "step": 400
     },
     {
-      "epoch": 7.5,
-      "learning_rate": 8.248723448996942e-06,
-      "loss": 0.0045,
+      "epoch": 7.23,
+      "learning_rate": 9.969203239509592e-06,
+      "loss": 0.0138,
       "step": 405
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 7.681891162260015e-06,
-      "loss": 0.0077,
+      "epoch": 7.32,
+      "learning_rate": 9.379423563186652e-06,
+      "loss": 0.009,
       "step": 410
     },
     {
-      "epoch": 7.69,
-      "learning_rate": 7.1316811159619305e-06,
-      "loss": 0.0078,
+      "epoch": 7.41,
+      "learning_rate": 8.803573307282209e-06,
+      "loss": 0.0096,
       "step": 415
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 6.598621411858319e-06,
-      "loss": 0.0147,
+      "epoch": 7.5,
+      "learning_rate": 8.242165977833974e-06,
+      "loss": 0.0071,
       "step": 420
     },
     {
-      "epoch": 7.87,
-      "learning_rate": 6.083223690489901e-06,
-      "loss": 0.0145,
+      "epoch": 7.59,
+      "learning_rate": 7.695702201611338e-06,
+      "loss": 0.0164,
       "step": 425
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 5.585982640100415e-06,
-      "loss": 0.0037,
+      "epoch": 7.68,
+      "learning_rate": 7.1646692796888456e-06,
+      "loss": 0.0073,
       "step": 430
     },
     {
-      "epoch": 8.0,
-      "eval_PREDICATE_f1": 0.9465648854961832,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9323308270676691,
-      "eval_PREDICATE_recall": 0.9612403100775194,
-      "eval_loss": 0.2312764674425125,
-      "eval_overall_accuracy": 0.9567333693888589,
-      "eval_overall_f1": 0.8329896907216495,
-      "eval_overall_precision": 0.8381742738589212,
-      "eval_overall_recall": 0.8278688524590164,
-      "eval_runtime": 0.5097,
-      "eval_samples_per_second": 188.335,
-      "eval_steps_per_second": 23.542,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6865671641791045,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7540983606557377,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6301369863013698,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7272727272727273,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6808510638297872,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.7804878048780488,
-      "step": 432
-    },
-    {
-      "epoch": 8.06,
-      "learning_rate": 5.107375521825791e-06,
-      "loss": 0.0066,
+      "epoch": 7.77,
+      "learning_rate": 6.649540752902544e-06,
+      "loss": 0.0065,
       "step": 435
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 4.647861711610024e-06,
-      "loss": 0.0083,
+      "epoch": 7.86,
+      "learning_rate": 6.150775979576906e-06,
+      "loss": 0.0291,
       "step": 440
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 4.207882259287707e-06,
-      "loss": 0.0091,
+      "epoch": 7.95,
+      "learning_rate": 5.668819725898747e-06,
+      "loss": 0.0128,
       "step": 445
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 3.7878594652562575e-06,
-      "loss": 0.009,
+      "epoch": 8.0,
+      "eval_PREDICATE_f1": 0.9689922480620156,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.946969696969697,
+      "eval_PREDICATE_recall": 0.9920634920634921,
+      "eval_loss": 0.18599779903888702,
+      "eval_overall_accuracy": 0.9587513935340022,
+      "eval_overall_f1": 0.853228962818004,
+      "eval_overall_precision": 0.8226415094339623,
+      "eval_overall_recall": 0.8861788617886179,
+      "eval_runtime": 0.4889,
+      "eval_samples_per_second": 202.488,
+      "eval_steps_per_second": 26.589,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7222222222222222,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7222222222222222,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7222222222222222,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8817204301075268,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8367346938775511,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9318181818181818,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 448
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 5.204101769303474e-06,
+      "loss": 0.0195,
       "step": 450
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 3.388196475144198e-06,
-      "loss": 0.0047,
+      "epoch": 8.12,
+      "learning_rate": 4.757036515227381e-06,
+      "loss": 0.018,
       "step": 455
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 3.0092768928645375e-06,
-      "loss": 0.0048,
+      "epoch": 8.21,
+      "learning_rate": 4.328022627567657e-06,
+      "loss": 0.0124,
       "step": 460
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 2.651464412424667e-06,
-      "loss": 0.0114,
+      "epoch": 8.3,
+      "learning_rate": 3.91744267317968e-06,
+      "loss": 0.0071,
       "step": 465
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 2.315102468846142e-06,
-      "loss": 0.0058,
+      "epoch": 8.39,
+      "learning_rate": 3.5256627807286085e-06,
+      "loss": 0.0189,
       "step": 470
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 2.0005139085293945e-06,
-      "loss": 0.0049,
+      "epoch": 8.48,
+      "learning_rate": 3.1530323141995553e-06,
+      "loss": 0.0053,
       "step": 475
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 1.7080006793798176e-06,
-      "loss": 0.0125,
+      "epoch": 8.57,
+      "learning_rate": 2.7998835613573137e-06,
+      "loss": 0.0147,
       "step": 480
     },
     {
-      "epoch": 8.98,
-      "learning_rate": 1.4378435409925867e-06,
-      "loss": 0.0089,
+      "epoch": 8.66,
+      "learning_rate": 2.4665314374336335e-06,
+      "loss": 0.0111,
       "step": 485
     },
     {
-      "epoch": 9.0,
-      "eval_PREDICATE_f1": 0.9465648854961832,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9323308270676691,
-      "eval_PREDICATE_recall": 0.9612403100775194,
-      "eval_loss": 0.24090717732906342,
-      "eval_overall_accuracy": 0.9588967009194159,
-      "eval_overall_f1": 0.8418891170431212,
-      "eval_overall_precision": 0.8436213991769548,
-      "eval_overall_recall": 0.8401639344262295,
-      "eval_runtime": 0.502,
-      "eval_samples_per_second": 191.245,
-      "eval_steps_per_second": 23.906,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7014925373134328,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7704918032786885,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6438356164383562,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7555555555555555,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6938775510204082,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8292682926829268,
-      "step": 486
-    },
-    {
-      "epoch": 9.07,
-      "learning_rate": 1.1903017951744144e-06,
-      "loss": 0.0055,
+      "epoch": 8.75,
+      "learning_rate": 2.153273204306153e-06,
+      "loss": 0.0043,
       "step": 490
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 9.656130370609056e-07,
-      "loss": 0.0089,
+      "epoch": 8.84,
+      "learning_rate": 1.8603882054194915e-06,
+      "loss": 0.0123,
       "step": 495
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 7.639929270683438e-07,
+      "epoch": 8.93,
+      "learning_rate": 1.5881376166848149e-06,
       "loss": 0.0064,
       "step": 500
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 5.856349838988612e-07,
-      "loss": 0.0057,
+      "epoch": 9.0,
+      "eval_PREDICATE_f1": 0.9649805447470817,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.9465648854961832,
+      "eval_PREDICATE_recall": 0.9841269841269841,
+      "eval_loss": 0.18625301122665405,
+      "eval_overall_accuracy": 0.9576365663322185,
+      "eval_overall_f1": 0.8493150684931506,
+      "eval_overall_precision": 0.8188679245283019,
+      "eval_overall_recall": 0.8821138211382114,
+      "eval_runtime": 0.5207,
+      "eval_samples_per_second": 190.14,
+      "eval_steps_per_second": 24.968,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7123287671232876,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7027027027027027,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7222222222222222,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8817204301075268,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8367346938775511,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9318181818181818,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 504
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.3367642135800329e-06,
+      "loss": 0.0049,
       "step": 505
     },
     {
-      "epoch": 9.44,
-      "learning_rate": 4.307103987976041e-07,
-      "loss": 0.002,
+      "epoch": 9.11,
+      "learning_rate": 1.106492154658323e-06,
+      "loss": 0.0062,
       "step": 510
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 2.9936787124022206e-07,
-      "loss": 0.006,
+      "epoch": 9.2,
+      "learning_rate": 8.975267816580201e-07,
+      "loss": 0.0139,
       "step": 515
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 1.917334662083714e-07,
-      "loss": 0.0094,
+      "epoch": 9.29,
+      "learning_rate": 7.100544363921324e-07,
+      "loss": 0.0062,
       "step": 520
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 1.0791049319021085e-07,
-      "loss": 0.0044,
+      "epoch": 9.38,
+      "learning_rate": 5.442422945807462e-07,
+      "loss": 0.026,
       "step": 525
     },
     {
-      "epoch": 9.81,
-      "learning_rate": 4.797940702205572e-08,
-      "loss": 0.0092,
+      "epoch": 9.46,
+      "learning_rate": 4.002382167745428e-07,
+      "loss": 0.0046,
       "step": 530
     },
     {
-      "epoch": 9.91,
-      "learning_rate": 1.1997730666338247e-08,
-      "loss": 0.009,
+      "epoch": 9.55,
+      "learning_rate": 2.781706165022879e-07,
+      "loss": 0.0073,
       "step": 535
     },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7814834575997363e-07,
+      "loss": 0.018,
+      "step": 540
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.0026059794363551e-07,
+      "loss": 0.0058,
+      "step": 545
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.457682831244259e-08,
+      "loss": 0.0058,
+      "step": 550
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.114669205298513e-08,
+      "loss": 0.0108,
+      "step": 555
+    },
     {
       "epoch": 10.0,
       "learning_rate": 0.0,
-      "loss": 0.0043,
-      "step": 540
+      "loss": 0.0155,
+      "step": 560
     },
     {
       "epoch": 10.0,
-      "eval_PREDICATE_f1": 0.9465648854961832,
-      "eval_PREDICATE_number": 129,
-      "eval_PREDICATE_precision": 0.9323308270676691,
-      "eval_PREDICATE_recall": 0.9612403100775194,
-      "eval_loss": 0.2416924387216568,
-      "eval_overall_accuracy": 0.9583558680367766,
-      "eval_overall_f1": 0.8395061728395061,
-      "eval_overall_precision": 0.8429752066115702,
-      "eval_overall_recall": 0.8360655737704918,
-      "eval_runtime": 0.5002,
-      "eval_samples_per_second": 191.922,
-      "eval_steps_per_second": 23.99,
+      "eval_PREDICATE_f1": 0.9649805447470817,
+      "eval_PREDICATE_number": 126,
+      "eval_PREDICATE_precision": 0.9465648854961832,
+      "eval_PREDICATE_recall": 0.9841269841269841,
+      "eval_loss": 0.18749289214611053,
+      "eval_overall_accuracy": 0.9576365663322185,
+      "eval_overall_f1": 0.8493150684931506,
+      "eval_overall_precision": 0.8188679245283019,
+      "eval_overall_recall": 0.8821138211382114,
+      "eval_runtime": 0.4979,
+      "eval_samples_per_second": 198.83,
+      "eval_steps_per_second": 26.109,
       "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_f1": 0.0,
-      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 1,
+      "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_number": 0,
       "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_precision": 0.0,
       "eval_\u0418\u041d\u0421\u0422\u0420\u0423\u041c\u0415\u041d\u0422_recall": 0.0,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.6917293233082706,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 73,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7666666666666667,
-      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.6301369863013698,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.7555555555555555,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 41,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.6938775510204082,
-      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.8292682926829268,
-      "step": 540
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_f1": 0.7123287671232876,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_number": 72,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_precision": 0.7027027027027027,
+      "eval_\u041a\u0410\u0423\u0417\u0410\u0422\u041e\u0420_recall": 0.7222222222222222,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_f1": 0.8817204301075268,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_number": 44,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_precision": 0.8367346938775511,
+      "eval_\u042d\u041a\u0421\u041f\u0415\u0420\u0418\u0415\u041d\u0426\u0415\u0420_recall": 0.9318181818181818,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_f1": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_number": 1,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_precision": 0.0,
+      "eval_\u0434\u0440\u0443\u0433\u043e\u0435_recall": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_f1": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_number": 3,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_precision": 0.0,
+      "eval_\u044d\u043a\u0441\u043f\u0435\u0440\u0438\u0435\u043d\u0446\u0435\u0440_recall": 0.0,
+      "step": 560
     },
     {
       "epoch": 10.0,
-      "step": 540,
-      "total_flos": 127804763383596.0,
-      "train_loss": 0.10899043403289936,
-      "train_runtime": 173.8981,
-      "train_samples_per_second": 49.397,
-      "train_steps_per_second": 3.105
+      "step": 560,
+      "total_flos": 132577707663864.0,
+      "train_loss": 0.13130194339568593,
+      "train_runtime": 174.9188,
+      "train_samples_per_second": 50.538,
+      "train_steps_per_second": 3.201
     }
   ],
-  "max_steps": 540,
+  "max_steps": 560,
   "num_train_epochs": 10,
-  "total_flos": 127804763383596.0,
+  "total_flos": 132577707663864.0,
   "trial_name": null,
   "trial_params": null
 }