swin-tiny-patch4-window7-224-vit0

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: 0.4836
• Accuracy: 0.8314

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.13	0.97	18	1.0297	0.4330
0.9066	2.0	37	0.8349	0.6590
0.7157	2.97	55	0.8050	0.6743
0.6446	4.0	74	0.6934	0.7165
0.5707	4.97	92	0.6324	0.7433
0.5042	6.0	111	0.6156	0.7356
0.4714	6.97	129	0.6825	0.7241
0.4225	8.0	148	0.5692	0.7625
0.3912	8.97	166	0.6150	0.7586
0.3442	10.0	185	0.4901	0.8008
0.289	10.97	203	0.5580	0.7739
0.2827	12.0	222	0.5308	0.7969
0.2375	12.97	240	0.5274	0.8046
0.2493	14.0	259	0.5433	0.8046
0.2309	14.97	277	0.5355	0.7931
0.1963	16.0	296	0.4836	0.8314
0.2162	16.97	314	0.4973	0.8238
0.2256	18.0	333	0.4918	0.8276
0.2124	18.97	351	0.5071	0.8161
0.1797	19.46	360	0.4985	0.8199

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0