swin-tiny-patch4-window7-224-ve-U13-b-80b

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: 0.6122
• Accuracy: 0.7826

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: 5.5e-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: 0.05
• num_epochs: 80

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.92	6	1.3855	0.1304
1.3852	2.0	13	1.3762	0.2826
1.3852	2.92	19	1.3521	0.2826
1.3565	4.0	26	1.2510	0.3478
1.2024	4.92	32	1.1528	0.3478
1.2024	6.0	39	1.0294	0.5
1.0453	6.92	45	0.9608	0.5217
0.8827	8.0	52	0.8801	0.6087
0.8827	8.92	58	0.9884	0.5652
0.7887	10.0	65	0.7927	0.6522
0.6795	10.92	71	0.7237	0.6522
0.6795	12.0	78	0.7250	0.6739
0.5777	12.92	84	0.7140	0.6957
0.496	14.0	91	0.8014	0.6957
0.496	14.92	97	0.8701	0.6739
0.4224	16.0	104	0.9384	0.6522
0.3744	16.92	110	0.7594	0.7174
0.3744	18.0	117	0.6122	0.7826
0.3775	18.92	123	0.8143	0.7174
0.3275	20.0	130	0.9981	0.6522
0.3275	20.92	136	0.8603	0.7174
0.3202	22.0	143	0.8412	0.6957
0.3202	22.92	149	0.8654	0.7174
0.2849	24.0	156	0.9650	0.6957
0.2518	24.92	162	0.8102	0.7609
0.2518	26.0	169	0.7203	0.7826
0.2467	26.92	175	0.9435	0.7391
0.2218	28.0	182	0.8905	0.7391
0.2218	28.92	188	1.0828	0.6957
0.2075	30.0	195	0.8936	0.7174
0.1893	30.92	201	0.8836	0.7826
0.1893	32.0	208	0.9692	0.7174
0.194	32.92	214	1.0390	0.7609
0.1739	34.0	221	0.8695	0.7609
0.1739	34.92	227	1.1836	0.6739
0.1895	36.0	234	1.0131	0.7391
0.1428	36.92	240	0.9618	0.7609
0.1428	38.0	247	0.9950	0.7609
0.1443	38.92	253	0.9113	0.7826
0.1574	40.0	260	0.9213	0.7174
0.1574	40.92	266	0.9437	0.7391
0.1442	42.0	273	0.9226	0.7609
0.1442	42.92	279	0.9430	0.7391
0.1186	44.0	286	0.9759	0.7826
0.1135	44.92	292	0.9651	0.7391
0.1135	46.0	299	0.9536	0.7609
0.1299	46.92	305	0.9118	0.7609
0.134	48.0	312	0.9848	0.7826
0.134	48.92	318	0.8641	0.7609
0.1418	50.0	325	1.0553	0.7609
0.1074	50.92	331	1.2511	0.6957
0.1074	52.0	338	1.0186	0.7391
0.1144	52.92	344	1.0467	0.7174
0.0999	54.0	351	0.9898	0.7391
0.0999	54.92	357	1.1780	0.7391
0.1131	56.0	364	1.0015	0.7609
0.1152	56.92	370	1.0759	0.7609
0.1152	58.0	377	1.1294	0.7174
0.1012	58.92	383	1.0894	0.7391
0.0938	60.0	390	1.0764	0.7391
0.0938	60.92	396	1.1784	0.7174
0.0944	62.0	403	1.1581	0.7174
0.0944	62.92	409	1.0444	0.7391
0.1015	64.0	416	1.0996	0.7391
0.0762	64.92	422	1.1235	0.7609
0.0762	66.0	429	1.0999	0.7391
0.0775	66.92	435	1.0776	0.7391
0.0787	68.0	442	1.0879	0.7391
0.0787	68.92	448	1.0913	0.7391
0.081	70.0	455	1.0558	0.7391
0.0749	70.92	461	1.0401	0.7391
0.0749	72.0	468	1.0539	0.7391
0.0841	72.92	474	1.0663	0.7391
0.0928	73.85	480	1.0712	0.7391

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0