Training in progress, epoch 3

logs/events.out.tfevents.1719316225.49cbd00d2005.382.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6cd44486db105a3795a27bd7e66817354b0b0c46c44ae0bd71e8c1ac02b9c2e4
-size 5961
+oid sha256:8f35900daa174a6a5fef76c0a33cef30b6c69a9bdc6be0978c13e94e7dc7999d
+size 6495

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9880a4d6b160b708dee866b97b67f4767461cdec03dbdef71638961456f15266
+oid sha256:088089bcc78f6a495d727bfd4f2722ab86ef44dde434e0dd36c7765a2949601f
 size 17549312

run-0/checkpoint-1581/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:053f4e74e37bd46e97455119628ae6d07b968538e4db43c7805f3b05ff409a39
+oid sha256:088089bcc78f6a495d727bfd4f2722ab86ef44dde434e0dd36c7765a2949601f
 size 17549312

run-0/checkpoint-1581/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa80d5ade6ff2f3070e08ba4f385965ae39d536ef54fd0fc63a64eb79be2dd74
+oid sha256:d2eab7a50471167ccdec8f251e96d92f16211722a1ba3ce445def961439cd852
 size 35123898

run-0/checkpoint-1581/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b63263b846c1da873d89d97103d702d45aa231d1b709bf63a6038446c1306d74
+oid sha256:6b7f8a1250df84a05c873d7a11e46287a5e3ea18e01e3d4c7581abd4866e3ed4
 size 1064

run-0/checkpoint-1581/trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.786697247706422,
-  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-0/checkpoint-1581",
+  "best_metric": 0.8176605504587156,
+  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-0/checkpoint-527",
   "epoch": 3.0,
   "eval_steps": 500,
   "global_step": 1581,
@@ -10,57 +10,57 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 12.377176284790039,
-      "learning_rate": 1.5980854192157344e-05,
-      "loss": 3.0218,
+      "grad_norm": 15.690781593322754,
+      "learning_rate": 0.0004329892843734803,
+      "loss": 1.29,
       "step": 527
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7672018348623854,
-      "eval_loss": 2.191861391067505,
-      "eval_runtime": 2.5906,
-      "eval_samples_per_second": 336.603,
-      "eval_steps_per_second": 2.702,
+      "eval_accuracy": 0.8176605504587156,
+      "eval_loss": 1.5847134590148926,
+      "eval_runtime": 2.4131,
+      "eval_samples_per_second": 361.36,
+      "eval_steps_per_second": 2.901,
       "step": 527
     },
     {
       "epoch": 2.0,
-      "grad_norm": Infinity,
-      "learning_rate": 1.2796813034327135e-05,
-      "loss": 2.0797,
+      "grad_norm": 3.8686888217926025,
+      "learning_rate": 0.0002886595229156535,
+      "loss": 0.5557,
       "step": 1054
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.7809633027522935,
-      "eval_loss": 1.8230279684066772,
-      "eval_runtime": 2.5719,
-      "eval_samples_per_second": 339.043,
-      "eval_steps_per_second": 2.722,
+      "eval_accuracy": 0.8142201834862385,
+      "eval_loss": 1.6788822412490845,
+      "eval_runtime": 2.4261,
+      "eval_samples_per_second": 359.428,
+      "eval_steps_per_second": 2.885,
       "step": 1054
     },
     {
       "epoch": 3.0,
-      "grad_norm": 40.96967315673828,
-      "learning_rate": 9.600642195895665e-06,
-      "loss": 1.697,
+      "grad_norm": 6.820591926574707,
+      "learning_rate": 0.00014432976145782676,
+      "loss": 0.3571,
       "step": 1581
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.786697247706422,
-      "eval_loss": 1.7212010622024536,
-      "eval_runtime": 2.549,
-      "eval_samples_per_second": 342.099,
-      "eval_steps_per_second": 2.746,
+      "eval_accuracy": 0.8142201834862385,
+      "eval_loss": 1.6854803562164307,
+      "eval_runtime": 2.4219,
+      "eval_samples_per_second": 360.053,
+      "eval_steps_per_second": 2.89,
       "step": 1581
     }
   ],
   "logging_steps": 500,
-  "max_steps": 3162,
+  "max_steps": 2108,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 6,
+  "num_train_epochs": 4,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -78,9 +78,9 @@
   "train_batch_size": 128,
   "trial_name": null,
   "trial_params": {
-    "alpha": 0.275210688972374,
-    "learning_rate": 1.9177025030588814e-05,
-    "num_train_epochs": 6,
+    "alpha": 0.19981548442581198,
+    "learning_rate": 0.000577319045831307,
+    "num_train_epochs": 4,
     "temperature": 6
   }
 }

run-0/checkpoint-1581/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:15e7b4feae857373f91378a3d0efc15d0ec396bfba71e74c965086843aa6acf4
+oid sha256:9b553ab9101fed6c5a9e75fca4cb81df81514d2d3f8ff94edc0c546949c119f9
 size 5176