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: 1.4603
• Accuracy: 0.4562

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-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.2036	1.0	10	1.5604	0.425
1.007	2.0	20	1.4904	0.4313
0.8453	3.0	30	1.4603	0.4625
0.6992	4.0	40	1.4409	0.4562
0.5703	5.0	50	1.4206	0.4688
0.4656	6.0	60	1.4128	0.4625
0.3896	7.0	70	1.4181	0.4625
0.3274	8.0	80	1.4162	0.475
0.2839	9.0	90	1.4199	0.45
0.2512	10.0	100	1.4261	0.4625
0.2278	11.0	110	1.4347	0.4625
0.2106	12.0	120	1.4319	0.4688
0.1973	13.0	130	1.4400	0.4688
0.1874	14.0	140	1.4453	0.4625
0.1797	15.0	150	1.4488	0.4562
0.1738	16.0	160	1.4542	0.4562
0.1693	17.0	170	1.4592	0.4625
0.166	18.0	180	1.4602	0.4562
0.1638	19.0	190	1.4607	0.4562
0.1627	20.0	200	1.4603	0.4562

Framework versions

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