Training in progress, step 1000

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3106101d76ba51ae6ac38bd5fbbb5381ba03142f77f63a4407420d2d5bd11364
+oid sha256:0492e7682ec1250a0c1b9ae09319ee939d2f6b518ac9ce9c7d535d268986e268
 size 268290900

run-1/checkpoint-1000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a085d735cd8cb688348e3d616bb12ef7bafe15d11b3ddb22f501564224348220
+oid sha256:0492e7682ec1250a0c1b9ae09319ee939d2f6b518ac9ce9c7d535d268986e268
 size 268290900

run-1/checkpoint-1000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:de2939019b76d56435d04fa44a1daaf95a1a67c19ae9f063ef8ba801e49059be
+oid sha256:1d998c093fbd7db3a10c06ebc303e2d3cf1fc1ce9d01ba291b44daf140ed629c
 size 536643898

run-1/checkpoint-1000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f182501c34e4ea3ebc7617d27edab7e1367582b147e518cd90295ec7f2eaa0f
+oid sha256:c794bc4c67ef18245dd516031ce405ab557e4d551d225d8dd1e1abc0f2be8e33
 size 1064

run-1/checkpoint-1000/trainer_state.json

@@ -10,50 +10,50 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.6690322580645162,
-      "eval_loss": 0.42948082089424133,
-      "eval_runtime": 5.393,
-      "eval_samples_per_second": 574.82,
-      "eval_steps_per_second": 12.053,
+      "eval_accuracy": 0.6529032258064517,
+      "eval_loss": 0.28555235266685486,
+      "eval_runtime": 5.4121,
+      "eval_samples_per_second": 572.786,
+      "eval_steps_per_second": 12.01,
       "step": 318
     },
     {
       "epoch": 1.5723270440251573,
-      "grad_norm": 0.8646999001502991,
-      "learning_rate": 1.371069182389937e-05,
-      "loss": 0.6591,
+      "grad_norm": 0.6621501445770264,
+      "learning_rate": 1.685534591194969e-05,
+      "loss": 0.4579,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.8229032258064516,
-      "eval_loss": 0.16900987923145294,
-      "eval_runtime": 5.2956,
-      "eval_samples_per_second": 585.388,
-      "eval_steps_per_second": 12.274,
+      "eval_accuracy": 0.8451612903225807,
+      "eval_loss": 0.1139635443687439,
+      "eval_runtime": 5.9245,
+      "eval_samples_per_second": 523.247,
+      "eval_steps_per_second": 10.971,
       "step": 636
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.8729032258064516,
-      "eval_loss": 0.0963619202375412,
-      "eval_runtime": 5.3142,
-      "eval_samples_per_second": 583.338,
-      "eval_steps_per_second": 12.231,
+      "eval_accuracy": 0.8961290322580645,
+      "eval_loss": 0.06579811871051788,
+      "eval_runtime": 5.4507,
+      "eval_samples_per_second": 568.736,
+      "eval_steps_per_second": 11.925,
       "step": 954
     },
     {
       "epoch": 3.1446540880503147,
-      "grad_norm": 0.7748064398765564,
-      "learning_rate": 7.421383647798742e-06,
-      "loss": 0.1958,
+      "grad_norm": 0.5677153468132019,
+      "learning_rate": 1.371069182389937e-05,
+      "loss": 0.1346,
       "step": 1000
     }
   ],
   "logging_steps": 500,
-  "max_steps": 1590,
+  "max_steps": 3180,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 5,
+  "num_train_epochs": 10,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -71,8 +71,8 @@
   "train_batch_size": 48,
   "trial_name": null,
   "trial_params": {
-    "alpha": 0.4868927644085208,
-    "num_train_epochs": 5,
-    "temperature": 2
+    "alpha": 0.31688267813599835,
+    "num_train_epochs": 10,
+    "temperature": 3
   }
 }

run-1/checkpoint-1000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f061e43c65f095a686a86278f34a31824db2598d218eedeec610b734a0e65f9
+oid sha256:6a074013ebca87018fe538250c2f45bfdeb260149a400a0644508c302966c2ef
 size 5240

runs/Oct12_06-40-39_b76c1be2ae55/events.out.tfevents.1728716534.b76c1be2ae55.1423.2

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b71460b0071f46213a0848524a678d109de6f766c17d002a989004287e7f0698
-size 13299
+oid sha256:2635f27678ee5945de34e08fe81a8b40cbf90c8a1788fafa3994fb3c8638c7d3
+size 13833