vit-base-patch16-224-classifier

This model is a fine-tuned version of google/vit-base-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.5720
• Accuracy: 0.7314

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.2
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.646	1.0	537	0.6400	0.6420
0.5941	2.0	1074	0.5874	0.6974
0.5259	3.0	1611	0.5849	0.7142
0.5459	4.0	2148	0.5645	0.7197
0.5086	5.0	2685	0.5554	0.7230
0.5397	6.0	3222	0.5540	0.7295
0.5646	7.0	3759	0.5491	0.7272
0.4564	8.0	4296	0.5771	0.7235
0.4951	9.0	4833	0.5518	0.7267
0.5074	10.0	5370	0.5556	0.7300
0.5512	11.0	5907	0.5739	0.7165
0.5003	12.0	6444	0.5648	0.7235
0.4442	13.0	6981	0.5581	0.7230
0.4787	14.0	7518	0.5556	0.7402
0.4944	15.0	8055	0.5589	0.7342
0.4678	16.0	8592	0.5567	0.7379
0.5569	17.0	9129	0.5601	0.7314
0.4164	18.0	9666	0.5619	0.7365
0.4406	19.0	10203	0.5711	0.7309
0.453	20.0	10740	0.5720	0.7314

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2