Model save

README.md

@@ -32,7 +32,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the image_folder dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.2124
+- Loss: 1.2039
 - Accuracy: 0.6667
 
 ## Model description
@@ -52,24 +52,24 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 5e-05
-- train_batch_size: 64
-- eval_batch_size: 64
+- learning_rate: 2e-05
+- train_batch_size: 32
+- eval_batch_size: 32
 - seed: 42
 - gradient_accumulation_steps: 4
-- total_train_batch_size: 256
+- total_train_batch_size: 128
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_ratio: 0.1
+- lr_scheduler_warmup_ratio: 0.01
 - num_epochs: 3
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|
-| No log        | 0.92  | 3    | 1.2355          | 0.6092   |
-| No log        | 1.85  | 6    | 1.2133          | 0.6552   |
-| No log        | 2.77  | 9    | 1.2124          | 0.6667   |
+| No log        | 0.96  | 6    | 1.2279          | 0.6322   |
+| 0.4158        | 1.92  | 12   | 1.2115          | 0.6437   |
+| 0.4158        | 2.88  | 18   | 1.2039          | 0.6667   |
 
 
 ### Framework versions