Boya3_SGD_1e3_20Epoch_Swin-large_fold1

This model is a fine-tuned version of microsoft/swin-large-patch4-window12-384-in22k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.8852
• Accuracy: 0.4024

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: 0.001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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
2.6532	1.0	632	2.5742	0.2512
2.3757	2.0	1264	2.3681	0.2865
2.2527	3.0	1896	2.2467	0.3060
2.2221	4.0	2528	2.1715	0.3179
2.1297	5.0	3160	2.1100	0.3230
2.068	6.0	3792	2.0715	0.3456
1.9695	7.0	4424	2.0381	0.3444
2.1086	8.0	5056	2.0071	0.3635
2.093	9.0	5688	1.9854	0.3651
2.05	10.0	6320	1.9645	0.3710
2.0434	11.0	6952	1.9480	0.3786
2.0666	12.0	7584	1.9363	0.3817
1.846	13.0	8216	1.9201	0.3889
1.9809	14.0	8848	1.9124	0.3897
1.844	15.0	9480	1.9027	0.3948
1.9048	16.0	10112	1.8971	0.3948
2.0342	17.0	10744	1.8912	0.4
1.822	18.0	11376	1.8876	0.4008
1.8676	19.0	12008	1.8858	0.4024
1.9147	20.0	12640	1.8852	0.4024

Framework versions

• Transformers 4.32.1
• Pytorch 2.1.1+cu121
• Datasets 2.21.0
• Tokenizers 0.13.2