Training in progress, step 200, checkpoint

checkpoint-200/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6bafe9677d96eb63bae43314ebe31a30361b1347d590fab78e657d5093cda7e5
+oid sha256:e261a8240f9ba007a28a67f46f5463f2d5850e2f5b3c08fc10ba3787cf10911e
 size 1474893317

checkpoint-200/pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7b06790aa3e1d971e160c719ebdd5ee22a691c28aff1977c8a9e4b150aaeac7a
+oid sha256:e6baa0ed3c0013d58345b0e3c81e7ffae409f63038ae73b3a5056a229c83bff9
 size 737457141

checkpoint-200/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f3284e82c30f90f720fbb01dfc9a4aa926d82378c51e59b01cb9aa4aa352e1b
+oid sha256:62ce52bbbc36bf5ba47e56ac99982edf5dc9b0952621007ccab5c2816e59d7f3
 size 14575

checkpoint-200/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f2099172cf643e0621a49c4fd812f032200315a0b3fd63d5198b626196954959
+oid sha256:738b3803967108b90cf940ac65ab8d2d54fdd8d26b5509e734c1ef7e40612c3c
 size 627

checkpoint-200/trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.5658872723579407,
-  "best_model_checkpoint": "/kaggle/output/checkpoint-200",
+  "best_metric": 0.5696164965629578,
+  "best_model_checkpoint": "/kaggle/output/checkpoint-100",
   "epoch": 0.4132231404958678,
   "eval_steps": 50,
   "global_step": 200,
@@ -10,52 +10,76 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.6979,
+      "learning_rate": 0.0,
+      "loss": 0.7246,
       "step": 1
     },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 0.5982,
+      "step": 50
+    },
     {
       "epoch": 0.1,
       "eval_accuracy": 0.75,
-      "eval_loss": 0.6723740696907043,
-      "eval_runtime": 28.903,
-      "eval_samples_per_second": 40.965,
-      "eval_steps_per_second": 20.482,
+      "eval_loss": 0.6296697854995728,
+      "eval_runtime": 30.5612,
+      "eval_samples_per_second": 38.742,
+      "eval_steps_per_second": 19.371,
       "step": 50
     },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00015,
+      "loss": 0.5505,
+      "step": 100
+    },
     {
       "epoch": 0.21,
       "eval_accuracy": 0.75,
-      "eval_loss": 0.6272028088569641,
-      "eval_runtime": 28.756,
-      "eval_samples_per_second": 41.174,
-      "eval_steps_per_second": 20.587,
+      "eval_loss": 0.5696164965629578,
+      "eval_runtime": 30.5816,
+      "eval_samples_per_second": 38.716,
+      "eval_steps_per_second": 19.358,
       "step": 100
     },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 0.5838,
+      "step": 150
+    },
     {
       "epoch": 0.31,
       "eval_accuracy": 0.75,
-      "eval_loss": 0.5726658701896667,
-      "eval_runtime": 28.7791,
-      "eval_samples_per_second": 41.141,
-      "eval_steps_per_second": 20.57,
+      "eval_loss": 0.562861442565918,
+      "eval_runtime": 31.0771,
+      "eval_samples_per_second": 38.099,
+      "eval_steps_per_second": 19.049,
       "step": 150
     },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 0.5925,
+      "step": 200
+    },
     {
       "epoch": 0.41,
       "eval_accuracy": 0.75,
-      "eval_loss": 0.5658872723579407,
-      "eval_runtime": 28.8005,
-      "eval_samples_per_second": 41.11,
-      "eval_steps_per_second": 20.555,
+      "eval_loss": 0.5931239724159241,
+      "eval_runtime": 30.6023,
+      "eval_samples_per_second": 38.69,
+      "eval_steps_per_second": 19.345,
       "step": 200
     }
   ],
-  "logging_steps": 1000,
+  "logging_steps": 50,
   "max_steps": 1000,
   "num_train_epochs": 3,
   "save_steps": 100,
-  "total_flos": 648800331133968.0,
+  "total_flos": 701423485924368.0,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-200/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbe5e88ca49040823bdb83dc035b8a4470c58ae105b0482f624b7d48f5e08cbe
+oid sha256:cd0372335b8d69875fb40a5865f6914bab9ff539640d01b439e8f707b1e5a43d
 size 4027