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

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: 0.1062
• Accuracy: 0.9749

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
1.545	1.0	53	1.3735	0.4845
0.8697	2.0	106	0.6895	0.7156
0.5783	3.0	159	0.4415	0.8165
0.4849	4.0	212	0.3737	0.8461
0.4004	5.0	265	0.3442	0.8488
0.3553	6.0	318	0.3271	0.8757
0.3318	7.0	371	0.2491	0.9050
0.3894	8.0	424	0.2636	0.9081
0.3201	9.0	477	0.2368	0.9070
0.2915	10.0	530	0.2390	0.9108
0.2582	11.0	583	0.2044	0.9294
0.2696	12.0	636	0.1948	0.9360
0.2429	13.0	689	0.2282	0.9143
0.257	14.0	742	0.1751	0.9339
0.2042	15.0	795	0.1765	0.9349
0.1952	16.0	848	0.1878	0.9284
0.1949	17.0	901	0.1303	0.9494
0.1786	18.0	954	0.1305	0.9552
0.1593	19.0	1007	0.1249	0.9570
0.1741	20.0	1060	0.1076	0.9601
0.1638	21.0	1113	0.1220	0.9580
0.1261	22.0	1166	0.1344	0.9532
0.1599	23.0	1219	0.1293	0.9535
0.1137	24.0	1272	0.1106	0.9621
0.1257	25.0	1325	0.1205	0.9573
0.1067	26.0	1378	0.1541	0.9535
0.1297	27.0	1431	0.1128	0.9604
0.1076	28.0	1484	0.1092	0.9594
0.0917	29.0	1537	0.1011	0.9614
0.0905	30.0	1590	0.1109	0.9604
0.0948	31.0	1643	0.1046	0.9638
0.0984	32.0	1696	0.1026	0.9669
0.0921	33.0	1749	0.1034	0.9642
0.0762	34.0	1802	0.0925	0.9687
0.0818	35.0	1855	0.0966	0.9656
0.0908	36.0	1908	0.0940	0.9687
0.0699	37.0	1961	0.0779	0.9742
0.0972	38.0	2014	0.1104	0.9687
0.0756	39.0	2067	0.0838	0.9742
0.0878	40.0	2120	0.1119	0.9673
0.0819	41.0	2173	0.1164	0.9618
0.0815	42.0	2226	0.1099	0.9666
0.0618	43.0	2279	0.1003	0.9680
0.0709	44.0	2332	0.0934	0.9721
0.0697	45.0	2385	0.0869	0.9731
0.0551	46.0	2438	0.1086	0.9694
0.049	47.0	2491	0.1036	0.9687
0.0646	48.0	2544	0.0854	0.9735
0.0704	49.0	2597	0.0959	0.9714
0.0578	50.0	2650	0.1034	0.9707
0.0579	51.0	2703	0.0965	0.9700
0.051	52.0	2756	0.0962	0.9721
0.0477	53.0	2809	0.1218	0.9690
0.0769	54.0	2862	0.1027	0.9714
0.0493	55.0	2915	0.1175	0.9725
0.0535	56.0	2968	0.1140	0.9690
0.0359	57.0	3021	0.0990	0.9725
0.0388	58.0	3074	0.0965	0.9700
0.0455	59.0	3127	0.1119	0.9700
0.0584	60.0	3180	0.0989	0.9735
0.0555	61.0	3233	0.1130	0.9680
0.0567	62.0	3286	0.1045	0.9721
0.0543	63.0	3339	0.1168	0.9707
0.0562	64.0	3392	0.1196	0.9649
0.0472	65.0	3445	0.1034	0.9725
0.0387	66.0	3498	0.1125	0.9728
0.0485	67.0	3551	0.1057	0.9738
0.0395	68.0	3604	0.1252	0.9725
0.0266	69.0	3657	0.1023	0.9742
0.0409	70.0	3710	0.1095	0.9738
0.0349	71.0	3763	0.1101	0.9752
0.0205	72.0	3816	0.1127	0.9725
0.0336	73.0	3869	0.1131	0.9735
0.0305	74.0	3922	0.0987	0.9749
0.0298	75.0	3975	0.1051	0.9742
0.0304	76.0	4028	0.1049	0.9728
0.051	77.0	4081	0.1134	0.9711
0.045	78.0	4134	0.1334	0.9707
0.0345	79.0	4187	0.1233	0.9707
0.0328	80.0	4240	0.1106	0.9728
0.0391	81.0	4293	0.1073	0.9735
0.0383	82.0	4346	0.1189	0.9707
0.0299	83.0	4399	0.1131	0.9756
0.0195	84.0	4452	0.1267	0.9714
0.0181	85.0	4505	0.1200	0.9700
0.0266	86.0	4558	0.1086	0.9752
0.0322	87.0	4611	0.1149	0.9735
0.0325	88.0	4664	0.1130	0.9738
0.0303	89.0	4717	0.1105	0.9749
0.0275	90.0	4770	0.1078	0.9752
0.0281	91.0	4823	0.1077	0.9742
0.0231	92.0	4876	0.1060	0.9752
0.022	93.0	4929	0.1077	0.9749
0.0219	94.0	4982	0.1080	0.9749
0.0184	95.0	5035	0.1061	0.9756
0.0198	96.0	5088	0.1047	0.9749
0.0355	97.0	5141	0.1084	0.9735
0.0309	98.0	5194	0.1088	0.9735
0.0324	99.0	5247	0.1066	0.9742
0.0216	100.0	5300	0.1062	0.9749

Framework versions

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