Training in progress, epoch 1

logs/events.out.tfevents.1709708789.a2333cf40ebd.21010.44

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1c58d2767d2e64cbb63ac442fcb696ca862cc0101651bdc87be29b1cfc8ea00c
-size 5315
+oid sha256:91a3f291e98472f59481890f345fbc80677b9eb1cd90a4eb6278486dd027d36c
+size 5998

logs/events.out.tfevents.1709709038.a2333cf40ebd.21010.45

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cde5c87561790c6dc7fd48b964ccf0f57e633cec542012f2b3e1114c2be35d3
+size 5315

logs/events.out.tfevents.1709709158.a2333cf40ebd.21010.46

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ceab3b6152d047465d2cefd782336c080b42c5b1210c091357aad2ff6b18ef52
+size 5314

logs/events.out.tfevents.1709709278.a2333cf40ebd.21010.47

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5b548f74c25d94d70b6e0bb74510f76eedc69edaeffea8f010be4fda4d356e3
+size 5315

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bafe3ef95928295f19e6b0af4ac6a462933e37fefd763ae18d20b9085c89c76e
+oid sha256:8f82b95e45bf81d582afa4e71bc9d7a293f1df5f4cc4679ddfe0cd23d18205cf
 size 17549312

run-47/checkpoint-96/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90e0f6ba198283f00aaa1b554b11a2fca511e92ed1166415610ba7cfabd50919
+oid sha256:8f82b95e45bf81d582afa4e71bc9d7a293f1df5f4cc4679ddfe0cd23d18205cf
 size 17549312

run-47/checkpoint-96/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b200cf27c9c9a7a16453eefe2d67ed67a0091e29aae0beacda6a77671224a81a
+oid sha256:d7ff3bf96144296f9acbd3bb4655c82325c4c4fb77ce71194072182467004dbd
 size 35122746

run-47/checkpoint-96/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fba0eb8e0f6d761d1bbeca94393c034dc12c19ace3051b7858250700f563bd35
+oid sha256:02ef8baccde93a073d296ce8d3fcf82edac3bac4ac0bfded24d201e289cdf958
 size 1064

run-47/checkpoint-96/trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.7964774951076321,
+  "best_metric": 0.7671232876712328,
   "best_model_checkpoint": "tiny-bert-sst2-distilled/run-47/checkpoint-96",
   "epoch": 1.0,
   "eval_steps": 500,
@@ -10,36 +10,36 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 10.090946197509766,
-      "learning_rate": 0.0002726931435690813,
-      "loss": 0.5743,
+      "grad_norm": 1.9338581562042236,
+      "learning_rate": 0.0001802406049279757,
+      "loss": 0.5354,
       "step": 96
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7964774951076321,
-      "eval_f1": 0.8188153310104529,
-      "eval_loss": 0.4561440944671631,
-      "eval_precision": 0.7378335949764521,
-      "eval_recall": 0.9197651663405088,
-      "eval_runtime": 30.8585,
-      "eval_samples_per_second": 33.119,
-      "eval_steps_per_second": 1.037,
+      "eval_accuracy": 0.7671232876712328,
+      "eval_f1": 0.8023255813953489,
+      "eval_loss": 0.4617577791213989,
+      "eval_precision": 0.696969696969697,
+      "eval_recall": 0.9452054794520548,
+      "eval_runtime": 29.4455,
+      "eval_samples_per_second": 34.708,
+      "eval_steps_per_second": 1.087,
       "step": 96
     }
   ],
   "logging_steps": 500,
-  "max_steps": 384,
+  "max_steps": 864,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 4,
+  "num_train_epochs": 9,
   "save_steps": 500,
   "total_flos": 235695197280.0,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": {
-    "alpha": 0.999132297568699,
-    "learning_rate": 0.00036359085809210834,
-    "num_train_epochs": 4,
-    "temperature": 14
+    "alpha": 0.8114321139277666,
+    "learning_rate": 0.00020277068054397267,
+    "num_train_epochs": 9,
+    "temperature": 7
   }
 }

run-47/checkpoint-96/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6e52d4ef52e4ae3e7583492431957732ea2f0822ace3819ef8d07c56bc154648
+oid sha256:0cca8f0e3496a02e33b0031c0a941841ea6a66348bacabeaa9cc0ea789d82d49
 size 4920

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:db89b5a0eda960d7e133b3a5df1b1cd6e950e12c3a2c337d2e86e0b613696bf4
+oid sha256:0cca8f0e3496a02e33b0031c0a941841ea6a66348bacabeaa9cc0ea789d82d49
 size 4920