swin-tiny-patch4-window7-224-finetuned-st-wsdmhar-stacked

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: 0.1344
• Accuracy: 0.9680

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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4973	1.0	53	1.3214	0.3743
0.7898	2.0	106	0.7212	0.6753
0.5919	3.0	159	0.5983	0.7317
0.5388	4.0	212	0.4451	0.8209
0.475	5.0	265	0.3542	0.8674
0.4174	6.0	318	0.3148	0.8771
0.3487	7.0	371	0.3107	0.8802
0.3385	8.0	424	0.3179	0.8798
0.3324	9.0	477	0.2846	0.8998
0.3347	10.0	530	0.2837	0.8871
0.2952	11.0	583	0.2412	0.9139
0.282	12.0	636	0.3142	0.8767
0.2679	13.0	689	0.2496	0.9005
0.2816	14.0	742	0.2014	0.9239
0.2989	15.0	795	0.2049	0.9218
0.2634	16.0	848	0.2066	0.9232
0.2692	17.0	901	0.1994	0.9284
0.2069	18.0	954	0.1958	0.9304
0.2373	19.0	1007	0.2273	0.9249
0.1992	20.0	1060	0.2094	0.9267
0.1997	21.0	1113	0.1808	0.9387
0.1794	22.0	1166	0.1833	0.9408
0.1736	23.0	1219	0.2456	0.9091
0.2004	24.0	1272	0.1918	0.9294
0.2039	25.0	1325	0.1768	0.9370
0.1829	26.0	1378	0.2090	0.9225
0.1566	27.0	1431	0.1467	0.9456
0.1531	28.0	1484	0.1604	0.9404
0.1553	29.0	1537	0.1612	0.9449
0.1406	30.0	1590	0.1644	0.9494
0.1396	31.0	1643	0.1411	0.9501
0.1049	32.0	1696	0.1616	0.9539
0.1411	33.0	1749	0.1708	0.9446
0.1211	34.0	1802	0.1392	0.9501
0.1113	35.0	1855	0.1369	0.9525
0.1249	36.0	1908	0.1320	0.9535
0.1274	37.0	1961	0.1524	0.9518
0.1191	38.0	2014	0.1438	0.9525
0.0949	39.0	2067	0.1379	0.9573
0.0936	40.0	2120	0.1463	0.9518
0.1008	41.0	2173	0.1681	0.9494
0.0887	42.0	2226	0.1463	0.9566
0.1113	43.0	2279	0.1719	0.9456
0.1087	44.0	2332	0.1343	0.9604
0.097	45.0	2385	0.1431	0.9576
0.1061	46.0	2438	0.1495	0.9580
0.11	47.0	2491	0.1555	0.9549
0.0806	48.0	2544	0.1493	0.9549
0.0979	49.0	2597	0.2320	0.9373
0.0751	50.0	2650	0.1516	0.9573
0.0845	51.0	2703	0.1277	0.9614
0.079	52.0	2756	0.1373	0.9601
0.0818	53.0	2809	0.1569	0.9539
0.0845	54.0	2862	0.1422	0.9604
0.0796	55.0	2915	0.1400	0.9621
0.0975	56.0	2968	0.1375	0.9573
0.0607	57.0	3021	0.1504	0.9580
0.0632	58.0	3074	0.1364	0.9607
0.0542	59.0	3127	0.1278	0.9669
0.0807	60.0	3180	0.1507	0.9518
0.0673	61.0	3233	0.1302	0.9645
0.0773	62.0	3286	0.1388	0.9638
0.0739	63.0	3339	0.1533	0.9573
0.0718	64.0	3392	0.1325	0.9594
0.0719	65.0	3445	0.1304	0.9625
0.0487	66.0	3498	0.1250	0.9645
0.0718	67.0	3551	0.1512	0.9573
0.0851	68.0	3604	0.1299	0.9607
0.0658	69.0	3657	0.1424	0.9625
0.0605	70.0	3710	0.1391	0.9625
0.0732	71.0	3763	0.1320	0.9642
0.0613	72.0	3816	0.1461	0.9607
0.056	73.0	3869	0.1328	0.9635
0.0661	74.0	3922	0.1319	0.9628
0.0581	75.0	3975	0.1337	0.9666
0.0698	76.0	4028	0.1383	0.9645
0.0544	77.0	4081	0.1324	0.9656
0.059	78.0	4134	0.1380	0.9645
0.0554	79.0	4187	0.1435	0.9638
0.0497	80.0	4240	0.1310	0.9649
0.0463	81.0	4293	0.1384	0.9604
0.0622	82.0	4346	0.1363	0.9628
0.0534	83.0	4399	0.1428	0.9635
0.0434	84.0	4452	0.1374	0.9656
0.0591	85.0	4505	0.1332	0.9663
0.0488	86.0	4558	0.1271	0.9697
0.0418	87.0	4611	0.1286	0.9669
0.0505	88.0	4664	0.1372	0.9676
0.0486	89.0	4717	0.1372	0.9676
0.0561	90.0	4770	0.1348	0.9680
0.0498	91.0	4823	0.1340	0.9669
0.0432	92.0	4876	0.1351	0.9621
0.0322	93.0	4929	0.1380	0.9659
0.0389	94.0	4982	0.1370	0.9656
0.0408	95.0	5035	0.1343	0.9683
0.0367	96.0	5088	0.1347	0.9680
0.0337	97.0	5141	0.1366	0.9669
0.0338	98.0	5194	0.1355	0.9669
0.0284	99.0	5247	0.1340	0.9676
0.0501	100.0	5300	0.1344	0.9680

Framework versions

• Transformers 4.43.2
• Pytorch 2.3.1+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1