Training in progress, step 1000

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b9a9d6a1c42d5caeb35bb23cea8ad3b05c91afc20890b7445b862c0ef1531a32
+oid sha256:ca0fbd724bbb6c79202fb96be6b52227cc41312e21279cdbe7c2e3bb0e43cd25
 size 268290900

run-0/checkpoint-1000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cbd257837648e7e0d9f2a9b35ba1325ee0da4bed8cd54d9b5ecd144679e1b873
+oid sha256:ca0fbd724bbb6c79202fb96be6b52227cc41312e21279cdbe7c2e3bb0e43cd25
 size 268290900

run-0/checkpoint-1000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d102376c8d8c6362e33751e0a9f9653a24d89471601c10ef9b051b649ec2a2e6
+oid sha256:d2a26d4f8625be501d27a1f3d6af931ce5538aa6d8364586b7bbb333c8aefd65
 size 536643898

run-0/checkpoint-1000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:60ef01273aaa599804d51a93c6c0c61874ebd50ab9e135a21aab5f7cd0e6487a
+oid sha256:ed9da1cbe846856af516cf037dd225e7f48e2737ef84c42a9bee0753b8c140ef
 size 1064

run-0/checkpoint-1000/trainer_state.json

@@ -10,50 +10,50 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5687096774193549,
-      "eval_loss": 0.1982271522283554,
-      "eval_runtime": 5.0636,
-      "eval_samples_per_second": 612.218,
-      "eval_steps_per_second": 12.837,
+      "eval_accuracy": 0.5793548387096774,
+      "eval_loss": 0.20042110979557037,
+      "eval_runtime": 5.4295,
+      "eval_samples_per_second": 570.955,
+      "eval_steps_per_second": 11.972,
       "step": 318
     },
     {
       "epoch": 1.5723270440251573,
-      "grad_norm": 0.5293694734573364,
-      "learning_rate": 1.4758909853249476e-05,
-      "loss": 0.3141,
+      "grad_norm": 0.5317145586013794,
+      "learning_rate": 1.550763701707098e-05,
+      "loss": 0.3197,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.8093548387096774,
-      "eval_loss": 0.10087499022483826,
-      "eval_runtime": 5.1871,
-      "eval_samples_per_second": 597.632,
-      "eval_steps_per_second": 12.531,
+      "eval_accuracy": 0.8148387096774193,
+      "eval_loss": 0.09952177852392197,
+      "eval_runtime": 5.6253,
+      "eval_samples_per_second": 551.079,
+      "eval_steps_per_second": 11.555,
       "step": 636
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.8661290322580645,
-      "eval_loss": 0.07097209990024567,
-      "eval_runtime": 5.3199,
-      "eval_samples_per_second": 582.716,
-      "eval_steps_per_second": 12.218,
+      "eval_accuracy": 0.8722580645161291,
+      "eval_loss": 0.06819746643304825,
+      "eval_runtime": 5.4054,
+      "eval_samples_per_second": 573.499,
+      "eval_steps_per_second": 12.025,
       "step": 954
     },
     {
       "epoch": 3.1446540880503147,
-      "grad_norm": 0.45407164096832275,
-      "learning_rate": 9.517819706498952e-06,
-      "loss": 0.1164,
+      "grad_norm": 0.45249998569488525,
+      "learning_rate": 1.101527403414196e-05,
+      "loss": 0.1151,
       "step": 1000
     }
   ],
   "logging_steps": 500,
-  "max_steps": 1908,
+  "max_steps": 2226,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 6,
+  "num_train_epochs": 7,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -71,8 +71,8 @@
   "train_batch_size": 48,
   "trial_name": null,
   "trial_params": {
-    "alpha": 0.2594446514054458,
-    "num_train_epochs": 6,
-    "temperature": 16
+    "alpha": 0.7983650996847113,
+    "num_train_epochs": 7,
+    "temperature": 12
   }
 }

run-0/checkpoint-1000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c4ad0cb2e8b6d04d02b40ca9aad4a1943ba29817f3bedd96489e24a1465b3c55
+oid sha256:7f486bed3ce5ea2c70ac8808f2e502101b7fef6b0db95ba252f4b5adc54a410c
 size 5176

runs/Jul21_11-33-52_b781708a4b03/events.out.tfevents.1721562176.b781708a4b03.764.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f24ed02c538027daa78b766195139da130de5ed3b2e4785abc65973d8278d7f
-size 12905
+oid sha256:895a4893e90dc7f3c99fd07c54f16ee79bd3c4023bf04562f294a80ed85fcbf9
+size 13762