Training in progress, step 4000, checkpoint

checkpoint-4000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6be46af600c0440c160ea139d30d07d741b3b4c28a4580f19d5fe4f252c5fad6
+oid sha256:2f712d6488e15866056a900f1108042582a23b6bab9309dda7a46494cc161b2c
 size 932401285

checkpoint-4000/pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69ecdb7a6d40191249e2062f95fd32e02300630c0b45bf8f38f85b83428bf76f
+oid sha256:1077ce1061a21df7ceca3bda929b302e9cd32d389609b8b72a6448c508501013
 size 466205145

checkpoint-4000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:78438bd24a328c256737c9033c9f92513574c229e2961ab877845fca7a0535f8
+oid sha256:369366a7fb22212d18fbe4bc3f1995e7a54c9ee9d85a0d19677f63ee8583f2b6
 size 14575

checkpoint-4000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dad04d422f3682299741a5b7a868993950396d0e3c3302efb96bf1db636c95df
+oid sha256:3af5a7494727cfb7aac640eecaced1d4117147018201912518f71fb553002d0b
 size 627

checkpoint-4000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 1.134506344795227,
+  "best_metric": 1.1194497346878052,
   "best_model_checkpoint": "/kaggle/output/checkpoint-4000",
-  "epoch": 64.0,
+  "epoch": 0.16297262059973924,
   "eval_steps": 1000,
   "global_step": 4000,
   "is_hyper_param_search": false,
@@ -9,75 +9,75 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.02,
+      "epoch": 0.0,
       "learning_rate": 2.7777777777777777e-11,
-      "loss": 1.2395,
+      "loss": 1.029,
       "step": 1
     },
     {
-      "epoch": 16.0,
+      "epoch": 0.04,
       "learning_rate": 2.7638888888888893e-08,
-      "loss": 1.1422,
+      "loss": 1.181,
       "step": 1000
     },
     {
-      "epoch": 16.0,
-      "eval_accuracy": 0.33073852295409184,
-      "eval_loss": 1.1557353734970093,
-      "eval_runtime": 54.1805,
-      "eval_samples_per_second": 92.469,
-      "eval_steps_per_second": 11.572,
+      "epoch": 0.04,
+      "eval_accuracy": 0.3273453093812375,
+      "eval_loss": 1.1529844999313354,
+      "eval_runtime": 54.2837,
+      "eval_samples_per_second": 92.293,
+      "eval_steps_per_second": 11.55,
       "step": 1000
     },
     {
-      "epoch": 32.0,
+      "epoch": 0.08,
       "learning_rate": 5.541666666666667e-08,
-      "loss": 1.1401,
+      "loss": 1.1527,
       "step": 2000
     },
     {
-      "epoch": 32.0,
-      "eval_accuracy": 0.3311377245508982,
-      "eval_loss": 1.1507551670074463,
-      "eval_runtime": 53.98,
-      "eval_samples_per_second": 92.812,
-      "eval_steps_per_second": 11.615,
+      "epoch": 0.08,
+      "eval_accuracy": 0.33013972055888224,
+      "eval_loss": 1.1351025104522705,
+      "eval_runtime": 54.0518,
+      "eval_samples_per_second": 92.689,
+      "eval_steps_per_second": 11.6,
       "step": 2000
     },
     {
-      "epoch": 48.0,
+      "epoch": 0.12,
       "learning_rate": 8.316666666666666e-08,
-      "loss": 1.1287,
+      "loss": 1.142,
       "step": 3000
     },
     {
-      "epoch": 48.0,
-      "eval_accuracy": 0.3321357285429142,
-      "eval_loss": 1.1441665887832642,
-      "eval_runtime": 53.9873,
-      "eval_samples_per_second": 92.8,
-      "eval_steps_per_second": 11.614,
+      "epoch": 0.12,
+      "eval_accuracy": 0.3317365269461078,
+      "eval_loss": 1.127414345741272,
+      "eval_runtime": 54.0871,
+      "eval_samples_per_second": 92.628,
+      "eval_steps_per_second": 11.592,
       "step": 3000
     },
     {
-      "epoch": 64.0,
-      "learning_rate": 1.1094444444444445e-07,
-      "loss": 1.117,
+      "epoch": 0.16,
+      "learning_rate": 1.1091666666666668e-07,
+      "loss": 1.1371,
       "step": 4000
     },
     {
-      "epoch": 64.0,
-      "eval_accuracy": 0.3287425149700599,
-      "eval_loss": 1.134506344795227,
-      "eval_runtime": 54.0067,
-      "eval_samples_per_second": 92.766,
-      "eval_steps_per_second": 11.61,
+      "epoch": 0.16,
+      "eval_accuracy": 0.331936127744511,
+      "eval_loss": 1.1194497346878052,
+      "eval_runtime": 54.3907,
+      "eval_samples_per_second": 92.111,
+      "eval_steps_per_second": 11.528,
       "step": 4000
     }
   ],
   "logging_steps": 1000,
   "max_steps": 10000000,
-  "num_train_epochs": 161291,
+  "num_train_epochs": 408,
   "save_steps": 1000,
   "total_flos": 8361420521472000.0,
   "trial_name": null,

checkpoint-4000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cb2cd1b7e648a386ccb79e47a451842e11001981c4363e993b678e0818db5cab
+oid sha256:5520446b18da09f4a43173b75839457981a344c0e709fd4bbcb249e3f60c859e
 size 4091