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

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: 0.7882
• Accuracy: 0.7391

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	1.3858	0.1304
1.3856	2.0	13	1.3777	0.3696
1.3856	2.92	19	1.3488	0.2391
1.361	4.0	26	1.2503	0.2826
1.2088	4.92	32	1.1317	0.4130
1.2088	6.0	39	1.0244	0.4565
1.0729	6.92	45	1.0413	0.4565
0.9554	8.0	52	0.9286	0.5652
0.9554	8.92	58	0.9103	0.5652
0.8221	10.0	65	0.8519	0.6522
0.732	10.92	71	0.8300	0.5870
0.732	12.0	78	0.8103	0.6304
0.6491	12.92	84	0.9533	0.5870
0.5724	14.0	91	0.7882	0.7391
0.5724	14.92	97	0.8072	0.6957
0.5305	16.0	104	0.7651	0.7391
0.4879	16.92	110	0.7379	0.7174
0.4879	18.0	117	0.7590	0.6739
0.4346	18.92	123	0.9283	0.6739
0.3671	20.0	130	1.0188	0.6304
0.3671	20.92	136	0.8959	0.7391
0.3725	22.0	143	0.9502	0.6957
0.3725	22.92	149	0.9627	0.6522
0.3321	24.0	156	0.9619	0.6957
0.3376	24.92	162	1.0459	0.6739
0.3376	26.0	169	1.0167	0.6522
0.3699	26.92	175	0.9949	0.6304
0.3098	28.0	182	0.9944	0.6739
0.3098	28.92	188	1.0860	0.6304
0.253	30.0	195	1.1721	0.6522
0.2615	30.92	201	1.1626	0.6739
0.2615	32.0	208	1.2464	0.6304
0.242	32.92	214	1.2179	0.6522
0.2173	34.0	221	1.2407	0.6304
0.2173	34.92	227	1.1585	0.6739
0.2305	36.0	234	1.3048	0.6522
0.2114	36.92	240	1.1776	0.6522
0.2114	38.0	247	1.1460	0.6522
0.2243	38.92	253	1.2424	0.6957
0.1822	40.0	260	1.2804	0.6739
0.1822	40.92	266	1.3472	0.6739
0.2065	42.0	273	1.3632	0.6739
0.2065	42.92	279	1.2832	0.6739
0.1942	44.0	286	1.3500	0.6739
0.1699	44.92	292	1.3242	0.6739
0.1699	46.0	299	1.3189	0.6957
0.1764	46.92	305	1.2840	0.6739
0.1771	48.0	312	1.3069	0.6957
0.1771	48.92	318	1.1585	0.6957
0.2095	50.0	325	1.3702	0.6957
0.1404	50.92	331	1.3539	0.6957
0.1404	52.0	338	1.3723	0.6957
0.1449	52.92	344	1.3877	0.6957
0.1348	54.0	351	1.3381	0.6739
0.1348	54.92	357	1.3700	0.6739
0.1683	56.0	364	1.2871	0.6957
0.1577	56.92	370	1.3214	0.6957
0.1577	58.0	377	1.3992	0.6522
0.1474	58.92	383	1.3800	0.6522
0.1267	60.0	390	1.2535	0.6739
0.1267	60.92	396	1.3200	0.6739
0.1171	62.0	403	1.3730	0.6739
0.1171	62.92	409	1.3678	0.6739
0.1461	64.0	416	1.3788	0.6739
0.1124	64.92	422	1.3944	0.6739
0.1124	66.0	429	1.3724	0.6739
0.1168	66.92	435	1.3553	0.6522
0.1243	68.0	442	1.3829	0.6739
0.1243	68.92	448	1.4040	0.6739
0.1375	70.0	455	1.4127	0.6522
0.1017	70.92	461	1.4070	0.6522
0.1017	72.0	468	1.3989	0.6739
0.1346	72.92	474	1.3995	0.6739
0.1382	73.85	480	1.3988	0.6739

Framework versions

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