results

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.5240
• Accuracy: 0.4813

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	80	2.0769	0.1562
No log	2.0	160	2.0542	0.2125
No log	3.0	240	1.9931	0.3125
No log	4.0	320	1.8756	0.2938
No log	5.0	400	1.6917	0.3875
No log	6.0	480	1.5471	0.4188
1.7305	7.0	560	1.4615	0.4562
1.7305	8.0	640	1.4356	0.4688
1.7305	9.0	720	1.3676	0.4875
1.7305	10.0	800	1.4125	0.5062
1.7305	11.0	880	1.5065	0.4688
1.7305	12.0	960	1.5047	0.4938
0.3363	13.0	1040	1.5180	0.4875
0.3363	14.0	1120	1.5228	0.4813
0.3363	15.0	1200	1.5240	0.4813

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0