Update models files

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 0.4645697044890593,
-    "eval_loss": 2.4816598892211914,
-    "eval_runtime": 13.9852,
-    "eval_samples": 208,
-    "eval_samples_per_second": 14.873,
-    "eval_steps_per_second": 3.003,
-    "perplexity": 11.961102053527732,
-    "train_loss": 2.365706356890222,
-    "train_runtime": 6608.9309,
-    "train_samples": 1741,
-    "train_samples_per_second": 5.269,
-    "train_steps_per_second": 1.056
+    "eval_accuracy": 0.48530345824830734,
+    "eval_loss": 2.389514446258545,
+    "eval_runtime": 14.0452,
+    "eval_samples": 217,
+    "eval_samples_per_second": 15.45,
+    "eval_steps_per_second": 3.133,
+    "perplexity": 10.908196141513345,
+    "train_loss": 2.27839626364512,
+    "train_runtime": 6672.7861,
+    "train_samples": 1821,
+    "train_samples_per_second": 5.458,
+    "train_steps_per_second": 1.094
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 20.0,
-    "eval_accuracy": 0.4645697044890593,
-    "eval_loss": 2.4816598892211914,
-    "eval_runtime": 13.9852,
-    "eval_samples": 208,
-    "eval_samples_per_second": 14.873,
-    "eval_steps_per_second": 3.003,
-    "perplexity": 11.961102053527732
+    "eval_accuracy": 0.48530345824830734,
+    "eval_loss": 2.389514446258545,
+    "eval_runtime": 14.0452,
+    "eval_samples": 217,
+    "eval_samples_per_second": 15.45,
+    "eval_steps_per_second": 3.133,
+    "perplexity": 10.908196141513345
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5fbe42a51327e3e5df35ce7a6c54b6c54037dcab0173a4669f0a835ad42693af
+oid sha256:6570a17556515668606980d04afad5c4884f6be43b321989ef611686f83c2356
 size 510398013

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "train_loss": 2.365706356890222,
-    "train_runtime": 6608.9309,
-    "train_samples": 1741,
-    "train_samples_per_second": 5.269,
-    "train_steps_per_second": 1.056
+    "train_loss": 2.27839626364512,
+    "train_runtime": 6672.7861,
+    "train_samples": 1821,
+    "train_samples_per_second": 5.458,
+    "train_steps_per_second": 1.094
 }

trainer_state.json

@@ -2,102 +2,108 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 20.0,
-  "global_step": 6980,
+  "global_step": 7300,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.43,
-      "learning_rate": 4.641833810888253e-05,
-      "loss": 3.1564,
+      "epoch": 1.37,
+      "learning_rate": 4.657534246575342e-05,
+      "loss": 3.0871,
       "step": 500
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 4.2836676217765046e-05,
-      "loss": 2.6213,
+      "epoch": 2.74,
+      "learning_rate": 4.3150684931506855e-05,
+      "loss": 2.5357,
       "step": 1000
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 3.9255014326647564e-05,
-      "loss": 2.503,
+      "epoch": 4.11,
+      "learning_rate": 3.9726027397260274e-05,
+      "loss": 2.4276,
       "step": 1500
     },
     {
-      "epoch": 5.73,
-      "learning_rate": 3.567335243553009e-05,
-      "loss": 2.4353,
+      "epoch": 5.48,
+      "learning_rate": 3.63013698630137e-05,
+      "loss": 2.3554,
       "step": 2000
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 3.2091690544412614e-05,
-      "loss": 2.3785,
+      "epoch": 6.85,
+      "learning_rate": 3.287671232876712e-05,
+      "loss": 2.3018,
       "step": 2500
     },
     {
-      "epoch": 8.6,
-      "learning_rate": 2.851002865329513e-05,
-      "loss": 2.3302,
+      "epoch": 8.22,
+      "learning_rate": 2.945205479452055e-05,
+      "loss": 2.2534,
       "step": 3000
     },
     {
-      "epoch": 10.03,
-      "learning_rate": 2.492836676217765e-05,
-      "loss": 2.2988,
+      "epoch": 9.59,
+      "learning_rate": 2.6027397260273973e-05,
+      "loss": 2.2174,
       "step": 3500
     },
     {
-      "epoch": 11.46,
-      "learning_rate": 2.1346704871060173e-05,
-      "loss": 2.2569,
+      "epoch": 10.96,
+      "learning_rate": 2.2602739726027396e-05,
+      "loss": 2.1862,
       "step": 4000
     },
     {
-      "epoch": 12.89,
-      "learning_rate": 1.7765042979942695e-05,
-      "loss": 2.235,
+      "epoch": 12.33,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 2.1502,
       "step": 4500
     },
     {
-      "epoch": 14.33,
-      "learning_rate": 1.4183381088825215e-05,
-      "loss": 2.2087,
+      "epoch": 13.7,
+      "learning_rate": 1.5753424657534248e-05,
+      "loss": 2.1359,
       "step": 5000
     },
     {
-      "epoch": 15.76,
-      "learning_rate": 1.0601719197707736e-05,
-      "loss": 2.19,
+      "epoch": 15.07,
+      "learning_rate": 1.2328767123287671e-05,
+      "loss": 2.1134,
       "step": 5500
     },
     {
-      "epoch": 17.19,
-      "learning_rate": 7.020057306590258e-06,
-      "loss": 2.1762,
+      "epoch": 16.44,
+      "learning_rate": 8.904109589041095e-06,
+      "loss": 2.0959,
       "step": 6000
     },
     {
-      "epoch": 18.62,
-      "learning_rate": 3.4383954154727795e-06,
-      "loss": 2.1656,
+      "epoch": 17.81,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 2.0852,
       "step": 6500
     },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.054794520547945e-06,
+      "loss": 2.0761,
+      "step": 7000
+    },
     {
       "epoch": 20.0,
-      "step": 6980,
-      "total_flos": 1.819637710848e+16,
-      "train_loss": 2.365706356890222,
-      "train_runtime": 6608.9309,
-      "train_samples_per_second": 5.269,
-      "train_steps_per_second": 1.056
+      "step": 7300,
+      "total_flos": 1.903251161088e+16,
+      "train_loss": 2.27839626364512,
+      "train_runtime": 6672.7861,
+      "train_samples_per_second": 5.458,
+      "train_steps_per_second": 1.094
     }
   ],
-  "max_steps": 6980,
+  "max_steps": 7300,
   "num_train_epochs": 20,
-  "total_flos": 1.819637710848e+16,
+  "total_flos": 1.903251161088e+16,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:338aec6a31bc28ca0c79c5ca1e08c265007bfaef7330233806b41e142086cb62
+oid sha256:8c49a916d9749c59a961b830f7c5ca0094acf4b9aba10491867872db48bd85ef
 size 3515