End of training

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 5.0,
-    "eval_f1": 0.8082255356174132,
-    "eval_loss": 0.7049386501312256,
-    "eval_runtime": 258.2262,
+    "eval_f1": 0.7925314519666281,
+    "eval_loss": 0.7206650376319885,
+    "eval_runtime": 10.3962,
     "eval_samples": 8605,
-    "eval_samples_per_second": 33.324,
-    "eval_steps_per_second": 1.042,
+    "eval_samples_per_second": 827.704,
+    "eval_steps_per_second": 12.985,
     "total_flos": 4.52871045169152e+16,
-    "train_loss": 0.6680765201611146,
-    "train_runtime": 28188.6693,
+    "train_loss": 0.9877658092399512,
+    "train_runtime": 691.6026,
     "train_samples": 34420,
-    "train_samples_per_second": 6.105,
-    "train_steps_per_second": 0.191
+    "train_samples_per_second": 248.842,
+    "train_steps_per_second": 3.89
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 5.0,
-    "eval_f1": 0.8082255356174132,
-    "eval_loss": 0.7049386501312256,
-    "eval_runtime": 258.2262,
+    "eval_f1": 0.7925314519666281,
+    "eval_loss": 0.7206650376319885,
+    "eval_runtime": 10.3962,
     "eval_samples": 8605,
-    "eval_samples_per_second": 33.324,
-    "eval_steps_per_second": 1.042
+    "eval_samples_per_second": 827.704,
+    "eval_steps_per_second": 12.985
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 5.0,
     "total_flos": 4.52871045169152e+16,
-    "train_loss": 0.6680765201611146,
-    "train_runtime": 28188.6693,
+    "train_loss": 0.9877658092399512,
+    "train_runtime": 691.6026,
     "train_samples": 34420,
-    "train_samples_per_second": 6.105,
-    "train_steps_per_second": 0.191
+    "train_samples_per_second": 248.842,
+    "train_steps_per_second": 3.89
 }

trainer_state.json

@@ -1,105 +1,105 @@
 {
-  "best_metric": null,
-  "best_model_checkpoint": null,
+  "best_metric": 0.7206650376319885,
+  "best_model_checkpoint": "./bert-multilingual-sdg-classification/checkpoint-2690",
   "epoch": 5.0,
-  "eval_steps": 500.0,
-  "global_step": 5380,
+  "eval_steps": 500,
+  "global_step": 2690,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0,
-      "grad_norm": 13.417204856872559,
-      "learning_rate": 1.8008368200836822e-05,
-      "loss": 1.4134,
-      "step": 1076
+      "epoch": 0.929368029739777,
+      "grad_norm": 6.992225646972656,
+      "learning_rate": 8.316666666666668e-06,
+      "loss": 2.1122,
+      "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_f1": 0.7558544457720137,
-      "eval_loss": 0.8534303903579712,
-      "eval_runtime": 268.7511,
-      "eval_samples_per_second": 32.018,
-      "eval_steps_per_second": 1.001,
-      "step": 1076
+      "eval_f1": 0.6814140148396195,
+      "eval_loss": 1.0624698400497437,
+      "eval_runtime": 10.5752,
+      "eval_samples_per_second": 813.695,
+      "eval_steps_per_second": 12.766,
+      "step": 538
     },
     {
-      "epoch": 2.0,
-      "grad_norm": 13.140690803527832,
-      "learning_rate": 1.3506276150627616e-05,
-      "loss": 0.7309,
-      "step": 2152
+      "epoch": 1.858736059479554,
+      "grad_norm": 6.2237749099731445,
+      "learning_rate": 8.090909090909092e-06,
+      "loss": 0.9564,
+      "step": 1000
     },
     {
       "epoch": 2.0,
-      "eval_f1": 0.7933829613883598,
-      "eval_loss": 0.7049058675765991,
-      "eval_runtime": 268.5786,
-      "eval_samples_per_second": 32.039,
-      "eval_steps_per_second": 1.002,
-      "step": 2152
+      "eval_f1": 0.7685561242660267,
+      "eval_loss": 0.8072949051856995,
+      "eval_runtime": 10.5617,
+      "eval_samples_per_second": 814.737,
+      "eval_steps_per_second": 12.782,
+      "step": 1076
     },
     {
-      "epoch": 3.0,
-      "grad_norm": 8.68571949005127,
-      "learning_rate": 9.004184100418411e-06,
-      "loss": 0.5277,
-      "step": 3228
+      "epoch": 2.7881040892193307,
+      "grad_norm": 7.1777119636535645,
+      "learning_rate": 5.698564593301436e-06,
+      "loss": 0.7652,
+      "step": 1500
     },
     {
       "epoch": 3.0,
-      "eval_f1": 0.8081018590927308,
-      "eval_loss": 0.6787558197975159,
-      "eval_runtime": 277.348,
-      "eval_samples_per_second": 31.026,
-      "eval_steps_per_second": 0.97,
-      "step": 3228
+      "eval_f1": 0.7885510737037653,
+      "eval_loss": 0.7433095574378967,
+      "eval_runtime": 10.6937,
+      "eval_samples_per_second": 804.676,
+      "eval_steps_per_second": 12.624,
+      "step": 1614
     },
     {
-      "epoch": 4.0,
-      "grad_norm": 7.1122260093688965,
-      "learning_rate": 4.5020920502092055e-06,
-      "loss": 0.3819,
-      "step": 4304
+      "epoch": 3.717472118959108,
+      "grad_norm": 7.244228363037109,
+      "learning_rate": 3.30622009569378e-06,
+      "loss": 0.6619,
+      "step": 2000
     },
     {
       "epoch": 4.0,
-      "eval_f1": 0.8077891874826753,
-      "eval_loss": 0.6842708587646484,
-      "eval_runtime": 271.8952,
-      "eval_samples_per_second": 31.648,
-      "eval_steps_per_second": 0.989,
-      "step": 4304
+      "eval_f1": 0.7918831109586941,
+      "eval_loss": 0.726136326789856,
+      "eval_runtime": 10.5238,
+      "eval_samples_per_second": 817.671,
+      "eval_steps_per_second": 12.828,
+      "step": 2152
     },
     {
-      "epoch": 5.0,
-      "grad_norm": 10.43897819519043,
-      "learning_rate": 0.0,
-      "loss": 0.2865,
-      "step": 5380
+      "epoch": 4.646840148698884,
+      "grad_norm": 5.082164287567139,
+      "learning_rate": 9.186602870813398e-07,
+      "loss": 0.6038,
+      "step": 2500
     },
     {
       "epoch": 5.0,
-      "eval_f1": 0.8082255356174132,
-      "eval_loss": 0.7049386501312256,
-      "eval_runtime": 260.8665,
-      "eval_samples_per_second": 32.986,
-      "eval_steps_per_second": 1.031,
-      "step": 5380
+      "eval_f1": 0.7925314519666281,
+      "eval_loss": 0.7206650376319885,
+      "eval_runtime": 10.5422,
+      "eval_samples_per_second": 816.247,
+      "eval_steps_per_second": 12.806,
+      "step": 2690
     },
     {
       "epoch": 5.0,
-      "step": 5380,
+      "step": 2690,
       "total_flos": 4.52871045169152e+16,
-      "train_loss": 0.6680765201611146,
-      "train_runtime": 28188.6693,
-      "train_samples_per_second": 6.105,
-      "train_steps_per_second": 0.191
+      "train_loss": 0.9877658092399512,
+      "train_runtime": 691.6026,
+      "train_samples_per_second": 248.842,
+      "train_steps_per_second": 3.89
     }
   ],
   "logging_steps": 500,
-  "max_steps": 5380,
+  "max_steps": 2690,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 5,
   "save_steps": 500,
@@ -116,7 +116,7 @@
     }
   },
   "total_flos": 4.52871045169152e+16,
-  "train_batch_size": 32,
+  "train_batch_size": 64,
   "trial_name": null,
   "trial_params": null
 }