swin-tiny-patch4-window7-224-finetuned-ViT

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2048
• Accuracy: 0.9371

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• 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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1549	1.0	41	0.8420	0.7654
0.5664	2.0	82	0.4272	0.8525
0.4053	3.0	123	0.3613	0.8767
0.3093	4.0	164	0.3808	0.8658
0.3	5.0	205	0.2971	0.8863
0.2445	6.0	246	0.3009	0.8851
0.2508	7.0	287	0.2582	0.9105
0.1998	8.0	328	0.2356	0.9202
0.1722	9.0	369	0.2187	0.9287
0.1176	10.0	410	0.2048	0.9371
0.1073	11.0	451	0.2066	0.9274
0.1075	12.0	492	0.2239	0.9262
0.1109	13.0	533	0.2336	0.9202
0.095	14.0	574	0.2187	0.9347
0.0824	15.0	615	0.2178	0.9287
0.0757	16.0	656	0.1958	0.9359
0.0643	17.0	697	0.2183	0.9335

Framework versions

• Transformers 4.51.1
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0