swinv2-tiny-patch4-window8-256-ve-U13-b-80c

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.7710
• 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: 4e-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: 80

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.92	6	1.6333	0.1087
1.4981	2.0	13	1.6225	0.1087
1.4981	2.92	19	1.5921	0.1087
1.4704	4.0	26	1.5001	0.1087
1.4116	4.92	32	1.4078	0.1087
1.4116	6.0	39	1.2960	0.3478
1.3094	6.92	45	1.2926	0.3043
1.2014	8.0	52	1.1532	0.5435
1.2014	8.92	58	1.1059	0.4783
1.0577	10.0	65	0.9510	0.6304
0.9185	10.92	71	0.9695	0.4783
0.9185	12.0	78	0.8792	0.6087
0.8369	12.92	84	0.8616	0.6957
0.7406	14.0	91	0.7816	0.6957
0.7406	14.92	97	0.7638	0.7609
0.6929	16.0	104	0.7710	0.7826
0.6192	16.92	110	0.7471	0.6957
0.6192	18.0	117	0.7265	0.7391
0.5936	18.92	123	0.7841	0.7609
0.5125	20.0	130	0.9320	0.6739
0.5125	20.92	136	0.7512	0.7609
0.4905	22.0	143	0.7466	0.6957
0.4905	22.92	149	0.8030	0.6957
0.4315	24.0	156	0.8184	0.7391
0.4272	24.92	162	0.8196	0.6957
0.4272	26.0	169	0.8712	0.6957
0.4261	26.92	175	0.7834	0.6957
0.4217	28.0	182	0.8394	0.6739
0.4217	28.92	188	0.9941	0.6739
0.3502	30.0	195	0.8909	0.7174
0.368	30.92	201	0.9995	0.7174
0.368	32.0	208	0.9418	0.6739
0.3473	32.92	214	0.8595	0.6739
0.3079	34.0	221	0.9562	0.6957
0.3079	34.92	227	0.8992	0.6739
0.3226	36.0	234	0.9908	0.6739
0.2603	36.92	240	0.9469	0.6957
0.2603	38.0	247	0.9942	0.6739
0.3028	38.92	253	1.0084	0.6739
0.2576	40.0	260	0.9908	0.6957
0.2576	40.92	266	1.0661	0.6957
0.2713	42.0	273	1.1347	0.6522
0.2713	42.92	279	1.1054	0.6739
0.2578	44.0	286	1.1089	0.6957
0.2367	44.92	292	1.1452	0.6739
0.2367	46.0	299	1.0272	0.6957
0.2301	46.92	305	1.1043	0.6739
0.2191	48.0	312	1.0815	0.6739
0.2191	48.92	318	0.9934	0.6957
0.2635	50.0	325	1.0866	0.6957
0.1874	50.92	331	1.0507	0.7174
0.1874	52.0	338	1.1002	0.7174
0.2057	52.92	344	1.0400	0.6739
0.1808	54.0	351	1.1092	0.7174
0.1808	54.92	357	1.1550	0.7174
0.2107	56.0	364	1.0579	0.6957
0.2149	56.92	370	1.0936	0.6957
0.2149	58.0	377	1.1692	0.6957
0.1865	58.92	383	1.1357	0.7174
0.1832	60.0	390	1.1549	0.6739
0.1832	60.92	396	1.1631	0.6957
0.1732	62.0	403	1.1312	0.6957
0.1732	62.92	409	1.1210	0.6957
0.1856	64.0	416	1.1835	0.6739
0.1503	64.92	422	1.1892	0.7174
0.1503	66.0	429	1.1865	0.6739
0.1713	66.92	435	1.1608	0.6739
0.1804	68.0	442	1.1699	0.6739
0.1804	68.92	448	1.1694	0.7174
0.1761	70.0	455	1.1744	0.6957
0.1619	70.92	461	1.1783	0.6957
0.1619	72.0	468	1.1797	0.6957
0.1649	72.92	474	1.1788	0.6957
0.1843	73.85	480	1.1780	0.6957

Framework versions

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