End of training

README.md

@@ -32,7 +32,7 @@ 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: 0.3493
+- Loss: 0.3492
 - Accuracy: 0.8571
 
 ## Model description

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 2.4,
-    "eval_accuracy": 0.7727272727272727,
-    "eval_loss": 0.5162484049797058,
-    "eval_runtime": 15.2929,
-    "eval_samples_per_second": 10.07,
-    "eval_steps_per_second": 0.131,
-    "total_flos": 3.668743974312346e+16,
-    "train_loss": 0.557659904162089,
-    "train_runtime": 121.886,
-    "train_samples_per_second": 15.014,
-    "train_steps_per_second": 0.025
+    "epoch": 3.0,
+    "eval_accuracy": 0.8571428571428571,
+    "eval_loss": 0.34923285245895386,
+    "eval_runtime": 5.7432,
+    "eval_samples_per_second": 26.814,
+    "eval_steps_per_second": 0.871,
+    "total_flos": 4.548645984411648e+16,
+    "train_loss": 0.4136195182800293,
+    "train_runtime": 92.017,
+    "train_samples_per_second": 19.888,
+    "train_steps_per_second": 0.163
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.4,
-    "eval_accuracy": 0.7727272727272727,
-    "eval_loss": 0.5162484049797058,
-    "eval_runtime": 15.2929,
-    "eval_samples_per_second": 10.07,
-    "eval_steps_per_second": 0.131
+    "epoch": 3.0,
+    "eval_accuracy": 0.8571428571428571,
+    "eval_loss": 0.34923285245895386,
+    "eval_runtime": 5.7432,
+    "eval_samples_per_second": 26.814,
+    "eval_steps_per_second": 0.871
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.4,
-    "total_flos": 3.668743974312346e+16,
-    "train_loss": 0.557659904162089,
-    "train_runtime": 121.886,
-    "train_samples_per_second": 15.014,
-    "train_steps_per_second": 0.025
+    "epoch": 3.0,
+    "total_flos": 4.548645984411648e+16,
+    "train_loss": 0.4136195182800293,
+    "train_runtime": 92.017,
+    "train_samples_per_second": 19.888,
+    "train_steps_per_second": 0.163
 }

trainer_state.json

@@ -1,57 +1,64 @@
 {
-  "best_metric": 0.7727272727272727,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-3",
-  "epoch": 2.4,
+  "best_metric": 0.8571428571428571,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-10",
+  "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 3,
+  "global_step": 15,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.8,
-      "eval_accuracy": 0.6493506493506493,
-      "eval_loss": 0.6294699311256409,
-      "eval_runtime": 15.6961,
-      "eval_samples_per_second": 9.811,
-      "eval_steps_per_second": 0.127,
-      "step": 1
+      "epoch": 1.0,
+      "eval_accuracy": 0.7857142857142857,
+      "eval_loss": 0.48992156982421875,
+      "eval_runtime": 5.6974,
+      "eval_samples_per_second": 27.03,
+      "eval_steps_per_second": 0.878,
+      "step": 5
     },
     {
-      "epoch": 1.6,
-      "eval_accuracy": 0.7597402597402597,
-      "eval_loss": 0.5519005060195923,
-      "eval_runtime": 15.2666,
-      "eval_samples_per_second": 10.087,
-      "eval_steps_per_second": 0.131,
-      "step": 2
+      "epoch": 2.0,
+      "grad_norm": 7.286997318267822,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.4907,
+      "step": 10
     },
     {
-      "epoch": 2.4,
-      "eval_accuracy": 0.7727272727272727,
-      "eval_loss": 0.5162484049797058,
-      "eval_runtime": 15.9531,
-      "eval_samples_per_second": 9.653,
-      "eval_steps_per_second": 0.125,
-      "step": 3
+      "epoch": 2.0,
+      "eval_accuracy": 0.8571428571428571,
+      "eval_loss": 0.34923285245895386,
+      "eval_runtime": 5.5078,
+      "eval_samples_per_second": 27.96,
+      "eval_steps_per_second": 0.908,
+      "step": 10
     },
     {
-      "epoch": 2.4,
-      "step": 3,
-      "total_flos": 3.668743974312346e+16,
-      "train_loss": 0.557659904162089,
-      "train_runtime": 121.886,
-      "train_samples_per_second": 15.014,
-      "train_steps_per_second": 0.025
+      "epoch": 3.0,
+      "eval_accuracy": 0.8571428571428571,
+      "eval_loss": 0.34931477904319763,
+      "eval_runtime": 5.5518,
+      "eval_samples_per_second": 27.739,
+      "eval_steps_per_second": 0.901,
+      "step": 15
+    },
+    {
+      "epoch": 3.0,
+      "step": 15,
+      "total_flos": 4.548645984411648e+16,
+      "train_loss": 0.4136195182800293,
+      "train_runtime": 92.017,
+      "train_samples_per_second": 19.888,
+      "train_steps_per_second": 0.163
     }
   ],
   "logging_steps": 10,
-  "max_steps": 3,
+  "max_steps": 15,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,
-  "total_flos": 3.668743974312346e+16,
-  "train_batch_size": 128,
+  "total_flos": 4.548645984411648e+16,
+  "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null
 }