swin-transformer-results

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8055
• Accuracy: 0.6794
• F1: 0.6810
• Precision: 0.6904

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision
1.0686	0.1952	500	1.0585	0.5266	0.5042	0.5355
1.3283	0.3903	1000	1.0015	0.5722	0.5794	0.6006
0.991	0.5855	1500	0.9601	0.5828	0.5865	0.6194
0.7919	0.7806	2000	0.9066	0.6135	0.6191	0.6580
0.9748	0.9758	2500	0.8327	0.6460	0.6443	0.6458
0.7183	1.1710	3000	0.8808	0.6421	0.6419	0.6638
0.769	1.3661	3500	0.8454	0.6526	0.6483	0.6553
0.8558	1.5613	4000	0.8773	0.6482	0.6364	0.6454
0.6713	1.7564	4500	0.8338	0.6561	0.6560	0.6711
0.7476	1.9516	5000	0.8083	0.6632	0.6636	0.6690
0.6896	2.1468	5500	0.8055	0.6794	0.6810	0.6904
0.648	2.3419	6000	0.8252	0.6697	0.6726	0.6822
0.5969	2.5371	6500	0.8179	0.6697	0.6676	0.6661
0.7098	2.7322	7000	0.8139	0.6724	0.6705	0.6698
0.5318	2.9274	7500	0.8033	0.6790	0.6783	0.6793

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cpu
• Datasets 3.0.0
• Tokenizers 0.19.1