Boya1_SGD_1e3_20Epoch_Swin-large-224_fold1

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

• Loss: 2.1316
• Accuracy: 0.3288

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
2.6134	1.0	924	2.5724	0.1941
2.5066	2.0	1848	2.4917	0.1947
2.5572	3.0	2772	2.4417	0.1971
2.3516	4.0	3696	2.4048	0.2080
2.4257	5.0	4620	2.3743	0.2332
2.3129	6.0	5544	2.3404	0.2411
2.2992	7.0	6468	2.3155	0.2636
2.266	8.0	7392	2.2914	0.2683
2.2389	9.0	8316	2.2655	0.2778
2.2861	10.0	9240	2.2421	0.2873
2.2192	11.0	10164	2.2250	0.2905
2.2057	12.0	11088	2.2046	0.2984
2.0519	13.0	12012	2.1880	0.3046
2.1151	14.0	12936	2.1736	0.3144
2.1116	15.0	13860	2.1599	0.3204
2.0726	16.0	14784	2.1517	0.3234
2.1017	17.0	15708	2.1408	0.3272
2.051	18.0	16632	2.1358	0.3285
2.0423	19.0	17556	2.1327	0.3296
2.0517	20.0	18480	2.1316	0.3288

Framework versions

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