End of training

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 2.0,
-    "eval_accuracy": 0.9904761904761905,
-    "eval_loss": 0.017159689217805862,
-    "eval_runtime": 1.7705,
-    "eval_samples_per_second": 118.612,
-    "eval_steps_per_second": 3.954,
-    "total_flos": 9.395826371334144e+16,
-    "train_loss": 0.323590816060702,
-    "train_runtime": 55.2419,
-    "train_samples_per_second": 171.066,
-    "train_steps_per_second": 1.358
+    "epoch": 6.0,
+    "eval_accuracy": 1.0,
+    "eval_loss": 0.03769616782665253,
+    "eval_runtime": 0.946,
+    "eval_samples_per_second": 112.046,
+    "eval_steps_per_second": 4.228,
+    "total_flos": 1.4153791066361856e+17,
+    "train_loss": 0.4622622158792284,
+    "train_runtime": 90.8298,
+    "train_samples_per_second": 208.962,
+    "train_steps_per_second": 1.541
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.0,
-    "eval_accuracy": 0.9904761904761905,
-    "eval_loss": 0.017159689217805862,
-    "eval_runtime": 1.7705,
-    "eval_samples_per_second": 118.612,
-    "eval_steps_per_second": 3.954
+    "epoch": 6.0,
+    "eval_accuracy": 1.0,
+    "eval_loss": 0.03769616782665253,
+    "eval_runtime": 0.946,
+    "eval_samples_per_second": 112.046,
+    "eval_steps_per_second": 4.228
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.0,
-    "total_flos": 9.395826371334144e+16,
-    "train_loss": 0.323590816060702,
-    "train_runtime": 55.2419,
-    "train_samples_per_second": 171.066,
-    "train_steps_per_second": 1.358
+    "epoch": 6.0,
+    "total_flos": 1.4153791066361856e+17,
+    "train_loss": 0.4622622158792284,
+    "train_runtime": 90.8298,
+    "train_samples_per_second": 208.962,
+    "train_steps_per_second": 1.541
 }

trainer_state.json

@@ -1,66 +1,109 @@
 {
-  "best_metric": 0.9904761904761905,
-  "best_model_checkpoint": "/kaggle/working/swin-brain-modality-classification/checkpoint-30",
-  "epoch": 2.0,
+  "best_metric": 1.0,
+  "best_model_checkpoint": "/kaggle/working/swin-brain-modality-classification/checkpoint-45",
+  "epoch": 6.0,
   "eval_steps": 500,
-  "global_step": 30,
+  "global_step": 45,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.6666666666666666,
-      "grad_norm": 5.664389610290527,
-      "learning_rate": 4.850746268656717e-05,
-      "loss": 0.803,
+      "epoch": 0.9333333333333333,
+      "eval_accuracy": 0.6415094339622641,
+      "eval_loss": 1.0901466608047485,
+      "eval_runtime": 1.0666,
+      "eval_samples_per_second": 99.379,
+      "eval_steps_per_second": 3.75,
+      "step": 7
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 5.982262134552002,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 1.2435,
       "step": 10
     },
     {
-      "epoch": 1.0,
-      "eval_accuracy": 0.9619047619047619,
-      "eval_loss": 0.1342908889055252,
-      "eval_runtime": 1.7227,
-      "eval_samples_per_second": 121.902,
-      "eval_steps_per_second": 4.063,
+      "epoch": 2.0,
+      "eval_accuracy": 0.8584905660377359,
+      "eval_loss": 0.42658886313438416,
+      "eval_runtime": 0.9395,
+      "eval_samples_per_second": 112.826,
+      "eval_steps_per_second": 4.258,
       "step": 15
     },
     {
-      "epoch": 1.3333333333333333,
-      "grad_norm": 3.640326976776123,
-      "learning_rate": 4.104477611940299e-05,
-      "loss": 0.1269,
+      "epoch": 2.6666666666666665,
+      "grad_norm": 8.782153129577637,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.5378,
       "step": 20
     },
     {
-      "epoch": 2.0,
-      "grad_norm": 1.9675283432006836,
-      "learning_rate": 3.358208955223881e-05,
-      "loss": 0.0409,
+      "epoch": 2.9333333333333336,
+      "eval_accuracy": 0.9528301886792453,
+      "eval_loss": 0.15236791968345642,
+      "eval_runtime": 0.9334,
+      "eval_samples_per_second": 113.563,
+      "eval_steps_per_second": 4.285,
+      "step": 22
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 5.220347881317139,
+      "learning_rate": 4.3650793650793655e-05,
+      "loss": 0.1747,
       "step": 30
     },
     {
-      "epoch": 2.0,
-      "eval_accuracy": 0.9904761904761905,
-      "eval_loss": 0.017159689217805862,
-      "eval_runtime": 1.722,
-      "eval_samples_per_second": 121.952,
-      "eval_steps_per_second": 4.065,
+      "epoch": 4.0,
+      "eval_accuracy": 0.9716981132075472,
+      "eval_loss": 0.07465644180774689,
+      "eval_runtime": 0.9385,
+      "eval_samples_per_second": 112.947,
+      "eval_steps_per_second": 4.262,
       "step": 30
     },
     {
-      "epoch": 2.0,
-      "step": 30,
-      "total_flos": 9.395826371334144e+16,
-      "train_loss": 0.323590816060702,
-      "train_runtime": 55.2419,
-      "train_samples_per_second": 171.066,
-      "train_steps_per_second": 1.358
+      "epoch": 4.933333333333334,
+      "eval_accuracy": 0.9622641509433962,
+      "eval_loss": 0.08645126223564148,
+      "eval_runtime": 0.9952,
+      "eval_samples_per_second": 106.514,
+      "eval_steps_per_second": 4.019,
+      "step": 37
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 5.531038761138916,
+      "learning_rate": 3.968253968253968e-05,
+      "loss": 0.0905,
+      "step": 40
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.03769616782665253,
+      "eval_runtime": 1.0531,
+      "eval_samples_per_second": 100.657,
+      "eval_steps_per_second": 3.798,
+      "step": 45
+    },
+    {
+      "epoch": 6.0,
+      "step": 45,
+      "total_flos": 1.4153791066361856e+17,
+      "train_loss": 0.4622622158792284,
+      "train_runtime": 90.8298,
+      "train_samples_per_second": 208.962,
+      "train_steps_per_second": 1.541
     }
   ],
   "logging_steps": 10,
-  "max_steps": 75,
+  "max_steps": 140,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 5,
+  "num_train_epochs": 20,
   "save_steps": 500,
   "stateful_callbacks": {
     "CustomEarlyStoppingCallback": {
@@ -83,7 +126,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 9.395826371334144e+16,
+  "total_flos": 1.4153791066361856e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null