End of training

README.md

@@ -33,7 +33,7 @@ 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 the imagefolder dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.1988
+- Loss: 0.2100
 - Accuracy: 0.9105
 
 ## Model description

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 3.0,
-    "eval_accuracy": 0.9036144578313253,
-    "eval_loss": 0.25001978874206543,
-    "eval_runtime": 11.7154,
-    "eval_samples_per_second": 49.593,
-    "eval_steps_per_second": 1.622,
-    "total_flos": 3.892774168288051e+17,
-    "train_loss": 0.44857889656128924,
-    "train_runtime": 357.7744,
-    "train_samples_per_second": 43.771,
-    "train_steps_per_second": 0.344
+    "epoch": 5.0,
+    "eval_accuracy": 0.9104991394148021,
+    "eval_loss": 0.21004654467105865,
+    "eval_runtime": 11.6442,
+    "eval_samples_per_second": 49.896,
+    "eval_steps_per_second": 1.632,
+    "total_flos": 6.487956947146752e+17,
+    "train_loss": 0.20639470670281387,
+    "train_runtime": 554.206,
+    "train_samples_per_second": 47.094,
+    "train_steps_per_second": 0.37
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.0,
-    "eval_accuracy": 0.9036144578313253,
-    "eval_loss": 0.25001978874206543,
-    "eval_runtime": 11.7154,
-    "eval_samples_per_second": 49.593,
-    "eval_steps_per_second": 1.622
+    "epoch": 5.0,
+    "eval_accuracy": 0.9104991394148021,
+    "eval_loss": 0.21004654467105865,
+    "eval_runtime": 11.6442,
+    "eval_samples_per_second": 49.896,
+    "eval_steps_per_second": 1.632
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 3.0,
-    "total_flos": 3.892774168288051e+17,
-    "train_loss": 0.44857889656128924,
-    "train_runtime": 357.7744,
-    "train_samples_per_second": 43.771,
-    "train_steps_per_second": 0.344
+    "epoch": 5.0,
+    "total_flos": 6.487956947146752e+17,
+    "train_loss": 0.20639470670281387,
+    "train_runtime": 554.206,
+    "train_samples_per_second": 47.094,
+    "train_steps_per_second": 0.37
 }

trainer_state.json

@@ -1,138 +1,212 @@
 {
-  "best_metric": 0.9036144578313253,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-poc\\checkpoint-82",
-  "epoch": 3.0,
+  "best_metric": 0.9104991394148021,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-poc\\checkpoint-164",
+  "epoch": 5.0,
   "eval_steps": 500,
-  "global_step": 123,
+  "global_step": 205,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.24390243902439024,
-      "grad_norm": 39.657066345214844,
-      "learning_rate": 3.846153846153846e-05,
-      "loss": 1.4333,
+      "grad_norm": 18.683757781982422,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.2672,
       "step": 10
     },
     {
       "epoch": 0.4878048780487805,
-      "grad_norm": 23.66535758972168,
-      "learning_rate": 4.681818181818182e-05,
-      "loss": 0.8054,
+      "grad_norm": 17.49846649169922,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.2349,
       "step": 20
     },
     {
       "epoch": 0.7317073170731707,
-      "grad_norm": 14.953439712524414,
-      "learning_rate": 4.2272727272727275e-05,
-      "loss": 0.4905,
+      "grad_norm": 11.330144882202148,
+      "learning_rate": 4.7554347826086956e-05,
+      "loss": 0.2134,
       "step": 30
     },
     {
       "epoch": 0.975609756097561,
-      "grad_norm": 21.484201431274414,
-      "learning_rate": 3.7727272727272725e-05,
-      "loss": 0.4233,
+      "grad_norm": 24.904529571533203,
+      "learning_rate": 4.483695652173913e-05,
+      "loss": 0.2261,
       "step": 40
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.8674698795180723,
-      "eval_loss": 0.3217690885066986,
-      "eval_runtime": 13.4934,
-      "eval_samples_per_second": 43.058,
-      "eval_steps_per_second": 1.408,
+      "eval_accuracy": 0.8984509466437177,
+      "eval_loss": 0.23582199215888977,
+      "eval_runtime": 11.0074,
+      "eval_samples_per_second": 52.783,
+      "eval_steps_per_second": 1.726,
       "step": 41
     },
     {
       "epoch": 1.2195121951219512,
-      "grad_norm": 10.432565689086914,
-      "learning_rate": 3.318181818181819e-05,
-      "loss": 0.3435,
+      "grad_norm": 8.899707794189453,
+      "learning_rate": 4.2119565217391304e-05,
+      "loss": 0.1967,
       "step": 50
     },
     {
       "epoch": 1.4634146341463414,
-      "grad_norm": 22.239377975463867,
-      "learning_rate": 2.863636363636364e-05,
-      "loss": 0.3269,
+      "grad_norm": 18.724058151245117,
+      "learning_rate": 3.940217391304348e-05,
+      "loss": 0.1736,
       "step": 60
     },
     {
       "epoch": 1.7073170731707317,
-      "grad_norm": 14.019749641418457,
-      "learning_rate": 2.4090909090909093e-05,
-      "loss": 0.2853,
+      "grad_norm": 15.106910705566406,
+      "learning_rate": 3.668478260869566e-05,
+      "loss": 0.1903,
       "step": 70
     },
     {
       "epoch": 1.951219512195122,
-      "grad_norm": 12.378067016601562,
-      "learning_rate": 1.9545454545454546e-05,
-      "loss": 0.3203,
+      "grad_norm": 14.325936317443848,
+      "learning_rate": 3.3967391304347826e-05,
+      "loss": 0.2418,
       "step": 80
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.9036144578313253,
-      "eval_loss": 0.25001978874206543,
-      "eval_runtime": 11.6049,
-      "eval_samples_per_second": 50.065,
-      "eval_steps_per_second": 1.637,
+      "eval_accuracy": 0.9087779690189329,
+      "eval_loss": 0.21669870615005493,
+      "eval_runtime": 11.4581,
+      "eval_samples_per_second": 50.707,
+      "eval_steps_per_second": 1.658,
       "step": 82
     },
     {
       "epoch": 2.1951219512195124,
-      "grad_norm": 11.0855131149292,
-      "learning_rate": 1.5e-05,
-      "loss": 0.2641,
+      "grad_norm": 11.764991760253906,
+      "learning_rate": 3.125e-05,
+      "loss": 0.2116,
       "step": 90
     },
     {
       "epoch": 2.4390243902439024,
-      "grad_norm": 8.276205062866211,
-      "learning_rate": 1.0454545454545455e-05,
-      "loss": 0.2553,
+      "grad_norm": 8.702162742614746,
+      "learning_rate": 2.8532608695652175e-05,
+      "loss": 0.2111,
       "step": 100
     },
     {
       "epoch": 2.682926829268293,
-      "grad_norm": 12.671364784240723,
-      "learning_rate": 5.909090909090909e-06,
-      "loss": 0.2546,
+      "grad_norm": 15.251940727233887,
+      "learning_rate": 2.5815217391304346e-05,
+      "loss": 0.2149,
       "step": 110
     },
     {
       "epoch": 2.926829268292683,
-      "grad_norm": 14.88228988647461,
-      "learning_rate": 1.3636363636363636e-06,
-      "loss": 0.2418,
+      "grad_norm": 19.997068405151367,
+      "learning_rate": 2.3097826086956523e-05,
+      "loss": 0.2044,
       "step": 120
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.8967297762478486,
-      "eval_loss": 0.2369231879711151,
-      "eval_runtime": 12.394,
-      "eval_samples_per_second": 46.877,
-      "eval_steps_per_second": 1.533,
+      "eval_accuracy": 0.9036144578313253,
+      "eval_loss": 0.20651938021183014,
+      "eval_runtime": 11.4071,
+      "eval_samples_per_second": 50.933,
+      "eval_steps_per_second": 1.666,
       "step": 123
     },
     {
-      "epoch": 3.0,
-      "step": 123,
-      "total_flos": 3.892774168288051e+17,
-      "train_loss": 0.44857889656128924,
-      "train_runtime": 357.7744,
-      "train_samples_per_second": 43.771,
-      "train_steps_per_second": 0.344
+      "epoch": 3.1707317073170733,
+      "grad_norm": 19.748979568481445,
+      "learning_rate": 2.0380434782608694e-05,
+      "loss": 0.206,
+      "step": 130
+    },
+    {
+      "epoch": 3.4146341463414633,
+      "grad_norm": 18.73997688293457,
+      "learning_rate": 1.766304347826087e-05,
+      "loss": 0.1849,
+      "step": 140
+    },
+    {
+      "epoch": 3.658536585365854,
+      "grad_norm": 12.744668006896973,
+      "learning_rate": 1.4945652173913044e-05,
+      "loss": 0.1857,
+      "step": 150
+    },
+    {
+      "epoch": 3.902439024390244,
+      "grad_norm": 14.340224266052246,
+      "learning_rate": 1.2228260869565218e-05,
+      "loss": 0.1995,
+      "step": 160
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9104991394148021,
+      "eval_loss": 0.21004654467105865,
+      "eval_runtime": 11.0045,
+      "eval_samples_per_second": 52.797,
+      "eval_steps_per_second": 1.727,
+      "step": 164
+    },
+    {
+      "epoch": 4.146341463414634,
+      "grad_norm": 12.231698036193848,
+      "learning_rate": 9.510869565217392e-06,
+      "loss": 0.1822,
+      "step": 170
+    },
+    {
+      "epoch": 4.390243902439025,
+      "grad_norm": 18.93744468688965,
+      "learning_rate": 6.7934782608695655e-06,
+      "loss": 0.1864,
+      "step": 180
+    },
+    {
+      "epoch": 4.634146341463414,
+      "grad_norm": 15.003849983215332,
+      "learning_rate": 4.07608695652174e-06,
+      "loss": 0.2083,
+      "step": 190
+    },
+    {
+      "epoch": 4.878048780487805,
+      "grad_norm": 24.660541534423828,
+      "learning_rate": 1.3586956521739131e-06,
+      "loss": 0.1972,
+      "step": 200
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9104991394148021,
+      "eval_loss": 0.19883577525615692,
+      "eval_runtime": 12.4083,
+      "eval_samples_per_second": 46.823,
+      "eval_steps_per_second": 1.531,
+      "step": 205
+    },
+    {
+      "epoch": 5.0,
+      "step": 205,
+      "total_flos": 6.487956947146752e+17,
+      "train_loss": 0.20639470670281387,
+      "train_runtime": 554.206,
+      "train_samples_per_second": 47.094,
+      "train_steps_per_second": 0.37
     }
   ],
   "logging_steps": 10,
-  "max_steps": 123,
+  "max_steps": 205,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 3,
+  "num_train_epochs": 5,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -146,7 +220,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.892774168288051e+17,
+  "total_flos": 6.487956947146752e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null