swin-tiny-patch4-window7-224-finetuned-teeth_dataset

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: 1.1564
• Accuracy: 0.8391

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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.8	3	4.5796	0.0152
No log	1.87	7	4.5200	0.0261
4.5616	2.93	11	4.4705	0.0326
4.5616	4.0	15	4.4127	0.0674
4.5616	4.8	18	4.3493	0.0804
4.44	5.87	22	4.2425	0.1130
4.44	6.93	26	4.1107	0.1370
4.1823	8.0	30	3.9340	0.1609
4.1823	8.8	33	3.7821	0.1935
4.1823	9.87	37	3.5314	0.2783
3.6357	10.93	41	3.2857	0.3043
3.6357	12.0	45	3.1064	0.3696
3.6357	12.8	48	2.9713	0.3826
3.0041	13.87	52	2.7172	0.4870
3.0041	14.93	56	2.5111	0.5435
2.4604	16.0	60	2.3561	0.5696
2.4604	16.8	63	2.2684	0.5717
2.4604	17.87	67	2.0961	0.6348
1.971	18.93	71	1.9555	0.6783
1.971	20.0	75	1.8400	0.6891
1.971	20.8	78	1.7856	0.7239
1.651	21.87	82	1.6797	0.7370
1.651	22.93	86	1.6007	0.7717
1.3665	24.0	90	1.5256	0.7739
1.3665	24.8	93	1.4876	0.7652
1.3665	25.87	97	1.4395	0.7783
1.1954	26.93	101	1.3679	0.7870
1.1954	28.0	105	1.3043	0.8022
1.1954	28.8	108	1.2906	0.8022
0.9886	29.87	112	1.2313	0.8109
0.9886	30.93	116	1.1829	0.8348
0.8803	32.0	120	1.1564	0.8391
0.8803	32.8	123	1.1421	0.8304
0.8803	33.87	127	1.1144	0.8326
0.815	34.93	131	1.1074	0.8304
0.815	36.0	135	1.0919	0.8283
0.815	36.8	138	1.0821	0.8326
0.7619	37.87	142	1.0701	0.8348
0.7619	38.93	146	1.0642	0.8348
0.6991	40.0	150	1.0631	0.8391

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2