End of training

README.md

@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [microsoft/swin-tiny-patch4-window7-224](https://huggingface.co/microsoft/swin-tiny-patch4-window7-224) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3453
-- Accuracy: 0.8895
+- Loss: 0.2871
+- Accuracy: 0.9116
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 11.294117647058824,
-    "eval_accuracy": 0.9226519337016574,
-    "eval_loss": 0.27896857261657715,
-    "eval_runtime": 1.4568,
-    "eval_samples_per_second": 124.248,
-    "eval_steps_per_second": 4.119,
-    "total_flos": 4.5598374310035456e+17,
-    "train_loss": 0.40474244703849155,
-    "train_runtime": 265.8333,
-    "train_samples_per_second": 73.309,
-    "train_steps_per_second": 0.542
+    "epoch": 18.823529411764707,
+    "eval_accuracy": 0.9116022099447514,
+    "eval_loss": 0.28710222244262695,
+    "eval_runtime": 1.5215,
+    "eval_samples_per_second": 118.965,
+    "eval_steps_per_second": 3.944,
+    "total_flos": 7.600391915087462e+17,
+    "train_loss": 0.3423821290334066,
+    "train_runtime": 426.6472,
+    "train_samples_per_second": 76.128,
+    "train_steps_per_second": 0.563
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 11.294117647058824,
-    "eval_accuracy": 0.9226519337016574,
-    "eval_loss": 0.27896857261657715,
-    "eval_runtime": 1.4568,
-    "eval_samples_per_second": 124.248,
-    "eval_steps_per_second": 4.119
+    "epoch": 18.823529411764707,
+    "eval_accuracy": 0.9116022099447514,
+    "eval_loss": 0.28710222244262695,
+    "eval_runtime": 1.5215,
+    "eval_samples_per_second": 118.965,
+    "eval_steps_per_second": 3.944
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 11.294117647058824,
-    "total_flos": 4.5598374310035456e+17,
-    "train_loss": 0.40474244703849155,
-    "train_runtime": 265.8333,
-    "train_samples_per_second": 73.309,
-    "train_steps_per_second": 0.542
+    "epoch": 18.823529411764707,
+    "total_flos": 7.600391915087462e+17,
+    "train_loss": 0.3423821290334066,
+    "train_runtime": 426.6472,
+    "train_samples_per_second": 76.128,
+    "train_steps_per_second": 0.563
 }

trainer_state.json

@@ -1,233 +1,366 @@
 {
-  "best_metric": 0.9226519337016574,
-  "best_model_checkpoint": "/kaggle/working/swin-brain-modality-classification/checkpoint-144",
-  "epoch": 11.294117647058824,
+  "best_metric": 0.9116022099447514,
+  "best_model_checkpoint": "/kaggle/working/swin-brain-modality-classification/checkpoint-51",
+  "epoch": 18.823529411764707,
   "eval_steps": 500,
-  "global_step": 144,
+  "global_step": 240,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.7843137254901961,
-      "grad_norm": 7.8652119636535645,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 1.2317,
+      "grad_norm": 5.9126081466674805,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 1.3069,
       "step": 10
     },
     {
       "epoch": 0.9411764705882353,
-      "eval_accuracy": 0.7679558011049724,
-      "eval_loss": 0.8055561780929565,
-      "eval_runtime": 1.4266,
-      "eval_samples_per_second": 126.873,
-      "eval_steps_per_second": 4.206,
+      "eval_accuracy": 0.6077348066298343,
+      "eval_loss": 0.9999544024467468,
+      "eval_runtime": 1.4322,
+      "eval_samples_per_second": 126.379,
+      "eval_steps_per_second": 4.189,
       "step": 12
     },
     {
       "epoch": 1.5686274509803921,
-      "grad_norm": 10.274601936340332,
-      "learning_rate": 4.8062015503875976e-05,
-      "loss": 0.7321,
+      "grad_norm": 7.789252758026123,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.8924,
       "step": 20
     },
     {
       "epoch": 1.9607843137254903,
       "eval_accuracy": 0.8784530386740331,
-      "eval_loss": 0.34271344542503357,
-      "eval_runtime": 1.441,
-      "eval_samples_per_second": 125.611,
-      "eval_steps_per_second": 4.164,
+      "eval_loss": 0.43337252736091614,
+      "eval_runtime": 1.4151,
+      "eval_samples_per_second": 127.902,
+      "eval_steps_per_second": 4.24,
       "step": 25
     },
     {
       "epoch": 2.3529411764705883,
-      "grad_norm": 7.0333251953125,
-      "learning_rate": 4.418604651162791e-05,
-      "loss": 0.4526,
+      "grad_norm": 6.7294697761535645,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.5365,
       "step": 30
     },
     {
       "epoch": 2.980392156862745,
-      "eval_accuracy": 0.9116022099447514,
-      "eval_loss": 0.2998741567134857,
-      "eval_runtime": 1.4374,
-      "eval_samples_per_second": 125.92,
-      "eval_steps_per_second": 4.174,
+      "eval_accuracy": 0.9005524861878453,
+      "eval_loss": 0.3143160939216614,
+      "eval_runtime": 1.4092,
+      "eval_samples_per_second": 128.446,
+      "eval_steps_per_second": 4.258,
       "step": 38
     },
     {
       "epoch": 3.1372549019607843,
-      "grad_norm": 6.83636474609375,
-      "learning_rate": 4.0310077519379843e-05,
-      "loss": 0.3845,
+      "grad_norm": 8.53775691986084,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.4119,
       "step": 40
     },
     {
       "epoch": 3.9215686274509802,
-      "grad_norm": 12.36241340637207,
-      "learning_rate": 3.6434108527131784e-05,
-      "loss": 0.3984,
+      "grad_norm": 8.774258613586426,
+      "learning_rate": 4.3981481481481486e-05,
+      "loss": 0.3814,
       "step": 50
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8950276243093923,
-      "eval_loss": 0.30480191111564636,
-      "eval_runtime": 1.4688,
-      "eval_samples_per_second": 123.229,
-      "eval_steps_per_second": 4.085,
+      "eval_accuracy": 0.9116022099447514,
+      "eval_loss": 0.28710222244262695,
+      "eval_runtime": 1.4257,
+      "eval_samples_per_second": 126.96,
+      "eval_steps_per_second": 4.209,
       "step": 51
     },
     {
       "epoch": 4.705882352941177,
-      "grad_norm": 4.690925121307373,
-      "learning_rate": 3.2558139534883724e-05,
-      "loss": 0.3362,
+      "grad_norm": 5.285433769226074,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.3336,
       "step": 60
     },
     {
       "epoch": 4.9411764705882355,
-      "eval_accuracy": 0.9060773480662984,
-      "eval_loss": 0.2832357585430145,
-      "eval_runtime": 1.4103,
-      "eval_samples_per_second": 128.343,
-      "eval_steps_per_second": 4.254,
+      "eval_accuracy": 0.9116022099447514,
+      "eval_loss": 0.2963091731071472,
+      "eval_runtime": 1.415,
+      "eval_samples_per_second": 127.916,
+      "eval_steps_per_second": 4.24,
       "step": 63
     },
     {
       "epoch": 5.490196078431373,
-      "grad_norm": 4.939608097076416,
-      "learning_rate": 2.868217054263566e-05,
-      "loss": 0.3356,
+      "grad_norm": 7.04965353012085,
+      "learning_rate": 3.935185185185186e-05,
+      "loss": 0.353,
       "step": 70
     },
     {
       "epoch": 5.96078431372549,
-      "eval_accuracy": 0.8950276243093923,
-      "eval_loss": 0.29102668166160583,
-      "eval_runtime": 1.4198,
-      "eval_samples_per_second": 127.482,
-      "eval_steps_per_second": 4.226,
+      "eval_accuracy": 0.8729281767955801,
+      "eval_loss": 0.31954672932624817,
+      "eval_runtime": 1.4556,
+      "eval_samples_per_second": 124.344,
+      "eval_steps_per_second": 4.122,
       "step": 76
     },
     {
       "epoch": 6.2745098039215685,
-      "grad_norm": 5.255800724029541,
-      "learning_rate": 2.48062015503876e-05,
-      "loss": 0.2958,
+      "grad_norm": 5.833162307739258,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.3069,
       "step": 80
     },
     {
       "epoch": 6.980392156862745,
       "eval_accuracy": 0.9116022099447514,
-      "eval_loss": 0.2802250385284424,
-      "eval_runtime": 1.405,
-      "eval_samples_per_second": 128.824,
-      "eval_steps_per_second": 4.27,
+      "eval_loss": 0.29521241784095764,
+      "eval_runtime": 1.4158,
+      "eval_samples_per_second": 127.846,
+      "eval_steps_per_second": 4.238,
       "step": 89
     },
     {
       "epoch": 7.0588235294117645,
-      "grad_norm": 4.129535675048828,
-      "learning_rate": 2.0930232558139536e-05,
-      "loss": 0.2735,
+      "grad_norm": 5.050061225891113,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.2789,
       "step": 90
     },
     {
       "epoch": 7.8431372549019605,
-      "grad_norm": 3.834648370742798,
-      "learning_rate": 1.7054263565891473e-05,
-      "loss": 0.2728,
+      "grad_norm": 4.222379207611084,
+      "learning_rate": 3.240740740740741e-05,
+      "loss": 0.293,
       "step": 100
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.9116022099447514,
-      "eval_loss": 0.2806554436683655,
-      "eval_runtime": 1.4695,
-      "eval_samples_per_second": 123.174,
-      "eval_steps_per_second": 4.083,
+      "eval_accuracy": 0.8895027624309392,
+      "eval_loss": 0.3174145817756653,
+      "eval_runtime": 1.4186,
+      "eval_samples_per_second": 127.591,
+      "eval_steps_per_second": 4.23,
       "step": 102
     },
     {
       "epoch": 8.627450980392156,
-      "grad_norm": 5.502172470092773,
-      "learning_rate": 1.3178294573643413e-05,
-      "loss": 0.2605,
+      "grad_norm": 7.039156436920166,
+      "learning_rate": 3.0092592592592593e-05,
+      "loss": 0.2667,
       "step": 110
     },
     {
       "epoch": 8.941176470588236,
-      "eval_accuracy": 0.9171270718232044,
-      "eval_loss": 0.2814862132072449,
-      "eval_runtime": 1.4181,
-      "eval_samples_per_second": 127.639,
-      "eval_steps_per_second": 4.231,
+      "eval_accuracy": 0.8950276243093923,
+      "eval_loss": 0.3225868344306946,
+      "eval_runtime": 1.4137,
+      "eval_samples_per_second": 128.03,
+      "eval_steps_per_second": 4.244,
       "step": 114
     },
     {
       "epoch": 9.411764705882353,
-      "grad_norm": 3.6628360748291016,
-      "learning_rate": 9.302325581395349e-06,
-      "loss": 0.2407,
+      "grad_norm": 5.598822593688965,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.2424,
       "step": 120
     },
     {
       "epoch": 9.96078431372549,
-      "eval_accuracy": 0.9171270718232044,
-      "eval_loss": 0.2806278169155121,
-      "eval_runtime": 1.4234,
-      "eval_samples_per_second": 127.162,
-      "eval_steps_per_second": 4.215,
+      "eval_accuracy": 0.8895027624309392,
+      "eval_loss": 0.3213161826133728,
+      "eval_runtime": 1.4313,
+      "eval_samples_per_second": 126.455,
+      "eval_steps_per_second": 4.192,
       "step": 127
     },
     {
       "epoch": 10.196078431372548,
-      "grad_norm": 3.5698726177215576,
-      "learning_rate": 5.426356589147287e-06,
-      "loss": 0.2528,
+      "grad_norm": 4.785697937011719,
+      "learning_rate": 2.5462962962962965e-05,
+      "loss": 0.2544,
       "step": 130
     },
     {
       "epoch": 10.980392156862745,
-      "grad_norm": 4.9788408279418945,
-      "learning_rate": 1.550387596899225e-06,
-      "loss": 0.2549,
+      "grad_norm": 5.349719047546387,
+      "learning_rate": 2.314814814814815e-05,
+      "loss": 0.2605,
       "step": 140
     },
     {
       "epoch": 10.980392156862745,
-      "eval_accuracy": 0.9171270718232044,
-      "eval_loss": 0.2795223295688629,
-      "eval_runtime": 1.466,
-      "eval_samples_per_second": 123.468,
-      "eval_steps_per_second": 4.093,
+      "eval_accuracy": 0.8895027624309392,
+      "eval_loss": 0.31716108322143555,
+      "eval_runtime": 1.4269,
+      "eval_samples_per_second": 126.846,
+      "eval_steps_per_second": 4.205,
       "step": 140
     },
     {
-      "epoch": 11.294117647058824,
-      "eval_accuracy": 0.9226519337016574,
-      "eval_loss": 0.27896857261657715,
-      "eval_runtime": 1.4683,
-      "eval_samples_per_second": 123.273,
-      "eval_steps_per_second": 4.086,
-      "step": 144
+      "epoch": 11.764705882352942,
+      "grad_norm": 6.121713161468506,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.232,
+      "step": 150
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.8895027624309392,
+      "eval_loss": 0.33846884965896606,
+      "eval_runtime": 1.406,
+      "eval_samples_per_second": 128.737,
+      "eval_steps_per_second": 4.268,
+      "step": 153
+    },
+    {
+      "epoch": 12.549019607843137,
+      "grad_norm": 7.647618770599365,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.242,
+      "step": 160
+    },
+    {
+      "epoch": 12.941176470588236,
+      "eval_accuracy": 0.8950276243093923,
+      "eval_loss": 0.32744264602661133,
+      "eval_runtime": 1.4273,
+      "eval_samples_per_second": 126.813,
+      "eval_steps_per_second": 4.204,
+      "step": 165
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 6.248785972595215,
+      "learning_rate": 1.6203703703703704e-05,
+      "loss": 0.215,
+      "step": 170
+    },
+    {
+      "epoch": 13.96078431372549,
+      "eval_accuracy": 0.8950276243093923,
+      "eval_loss": 0.33850720524787903,
+      "eval_runtime": 1.4433,
+      "eval_samples_per_second": 125.407,
+      "eval_steps_per_second": 4.157,
+      "step": 178
+    },
+    {
+      "epoch": 14.117647058823529,
+      "grad_norm": 5.345800876617432,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2123,
+      "step": 180
+    },
+    {
+      "epoch": 14.901960784313726,
+      "grad_norm": 5.421293258666992,
+      "learning_rate": 1.1574074074074075e-05,
+      "loss": 0.2131,
+      "step": 190
+    },
+    {
+      "epoch": 14.980392156862745,
+      "eval_accuracy": 0.8950276243093923,
+      "eval_loss": 0.34223416447639465,
+      "eval_runtime": 1.409,
+      "eval_samples_per_second": 128.462,
+      "eval_steps_per_second": 4.258,
+      "step": 191
+    },
+    {
+      "epoch": 15.686274509803921,
+      "grad_norm": 4.188720703125,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.201,
+      "step": 200
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.8784530386740331,
+      "eval_loss": 0.341948539018631,
+      "eval_runtime": 1.4191,
+      "eval_samples_per_second": 127.549,
+      "eval_steps_per_second": 4.228,
+      "step": 204
+    },
+    {
+      "epoch": 16.470588235294116,
+      "grad_norm": 4.887516498565674,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.1976,
+      "step": 210
+    },
+    {
+      "epoch": 16.941176470588236,
+      "eval_accuracy": 0.9005524861878453,
+      "eval_loss": 0.3447644114494324,
+      "eval_runtime": 1.4043,
+      "eval_samples_per_second": 128.89,
+      "eval_steps_per_second": 4.273,
+      "step": 216
+    },
+    {
+      "epoch": 17.254901960784313,
+      "grad_norm": 5.946260452270508,
+      "learning_rate": 4.6296296296296296e-06,
+      "loss": 0.1886,
+      "step": 220
+    },
+    {
+      "epoch": 17.96078431372549,
+      "eval_accuracy": 0.8895027624309392,
+      "eval_loss": 0.3459985554218292,
+      "eval_runtime": 1.4593,
+      "eval_samples_per_second": 124.035,
+      "eval_steps_per_second": 4.112,
+      "step": 229
+    },
+    {
+      "epoch": 18.03921568627451,
+      "grad_norm": 6.021714210510254,
+      "learning_rate": 2.3148148148148148e-06,
+      "loss": 0.2,
+      "step": 230
+    },
+    {
+      "epoch": 18.823529411764707,
+      "grad_norm": 3.8110241889953613,
+      "learning_rate": 0.0,
+      "loss": 0.1972,
+      "step": 240
+    },
+    {
+      "epoch": 18.823529411764707,
+      "eval_accuracy": 0.8895027624309392,
+      "eval_loss": 0.34530630707740784,
+      "eval_runtime": 1.454,
+      "eval_samples_per_second": 124.486,
+      "eval_steps_per_second": 4.127,
+      "step": 240
     },
     {
-      "epoch": 11.294117647058824,
-      "step": 144,
-      "total_flos": 4.5598374310035456e+17,
-      "train_loss": 0.40474244703849155,
-      "train_runtime": 265.8333,
-      "train_samples_per_second": 73.309,
-      "train_steps_per_second": 0.542
+      "epoch": 18.823529411764707,
+      "step": 240,
+      "total_flos": 7.600391915087462e+17,
+      "train_loss": 0.3423821290334066,
+      "train_runtime": 426.6472,
+      "train_samples_per_second": 76.128,
+      "train_steps_per_second": 0.563
     }
   ],
   "logging_steps": 10,
-  "max_steps": 144,
+  "max_steps": 240,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 12,
+  "num_train_epochs": 20,
   "save_steps": 500,
   "stateful_callbacks": {
     "CustomEarlyStoppingCallback": {
@@ -250,7 +383,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 4.5598374310035456e+17,
+  "total_flos": 7.600391915087462e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null