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

This model is a fine-tuned version of nielsr/swin-tiny-patch4-window7-224-finetuned-eurosat on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.1444
• Accuracy: 0.9683

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
1.0624	1.0	53	0.8879	0.6092
0.6893	2.0	106	0.6601	0.7090
0.6152	3.0	159	0.5114	0.7855
0.5456	4.0	212	0.3819	0.8423
0.4673	5.0	265	0.3267	0.8719
0.4166	6.0	318	0.2804	0.9039
0.3757	7.0	371	0.2881	0.8994
0.3798	8.0	424	0.2635	0.9032
0.3303	9.0	477	0.2703	0.9074
0.3346	10.0	530	0.2565	0.9005
0.2971	11.0	583	0.2182	0.9311
0.2992	12.0	636	0.2240	0.9256
0.2637	13.0	689	0.2131	0.9239
0.2653	14.0	742	0.1801	0.9397
0.2472	15.0	795	0.1807	0.9377
0.2263	16.0	848	0.1612	0.9466
0.1786	17.0	901	0.1735	0.9418
0.2103	18.0	954	0.1786	0.9463
0.1725	19.0	1007	0.1631	0.9473
0.1787	20.0	1060	0.1439	0.9532
0.1924	21.0	1113	0.1388	0.9504
0.1662	22.0	1166	0.1470	0.9508
0.1724	23.0	1219	0.1538	0.9497
0.1633	24.0	1272	0.1731	0.9384
0.174	25.0	1325	0.1555	0.9539
0.1657	26.0	1378	0.1542	0.9494
0.1513	27.0	1431	0.1526	0.9508
0.126	28.0	1484	0.1560	0.9511
0.1508	29.0	1537	0.1607	0.9480
0.1368	30.0	1590	0.1729	0.9435
0.1166	31.0	1643	0.1555	0.9532
0.1076	32.0	1696	0.1400	0.9580
0.1189	33.0	1749	0.1419	0.9590
0.1512	34.0	1802	0.1364	0.9580
0.1323	35.0	1855	0.1497	0.9539
0.1031	36.0	1908	0.1437	0.9580
0.1215	37.0	1961	0.1460	0.9559
0.1069	38.0	2014	0.1362	0.9601
0.129	39.0	2067	0.1490	0.9590
0.1202	40.0	2120	0.1616	0.9545
0.1011	41.0	2173	0.1518	0.9570
0.1092	42.0	2226	0.1308	0.9618
0.1163	43.0	2279	0.1458	0.9590
0.1074	44.0	2332	0.1414	0.9549
0.0814	45.0	2385	0.1509	0.9580
0.0985	46.0	2438	0.1287	0.9628
0.0863	47.0	2491	0.1277	0.9625
0.0932	48.0	2544	0.1453	0.9559
0.0863	49.0	2597	0.1520	0.9566
0.0887	50.0	2650	0.1279	0.9656
0.0744	51.0	2703	0.1552	0.9566
0.0928	52.0	2756	0.1465	0.9621
0.0776	53.0	2809	0.1575	0.9583
0.088	54.0	2862	0.1614	0.9563
0.0909	55.0	2915	0.1312	0.9638
0.089	56.0	2968	0.1357	0.9652
0.0587	57.0	3021	0.1510	0.9614
0.0931	58.0	3074	0.1466	0.9580
0.0878	59.0	3127	0.1499	0.9590
0.0725	60.0	3180	0.1524	0.9597
0.0543	61.0	3233	0.1543	0.9583
0.0773	62.0	3286	0.1513	0.9635
0.0626	63.0	3339	0.1511	0.9601
0.0649	64.0	3392	0.1467	0.9594
0.0705	65.0	3445	0.1443	0.9590
0.0737	66.0	3498	0.1361	0.9607
0.0518	67.0	3551	0.1441	0.9594
0.0502	68.0	3604	0.1535	0.9590
0.0701	69.0	3657	0.1362	0.9663
0.0826	70.0	3710	0.1492	0.9611
0.0715	71.0	3763	0.1615	0.9625
0.0635	72.0	3816	0.1488	0.9642
0.0522	73.0	3869	0.1456	0.9621
0.0485	74.0	3922	0.1386	0.9645
0.0629	75.0	3975	0.1463	0.9632
0.0568	76.0	4028	0.1472	0.9621
0.0556	77.0	4081	0.1440	0.9659
0.0547	78.0	4134	0.1421	0.9635
0.0527	79.0	4187	0.1444	0.9683
0.054	80.0	4240	0.1464	0.9628
0.0641	81.0	4293	0.1491	0.9635
0.0546	82.0	4346	0.1529	0.9611
0.059	83.0	4399	0.1462	0.9652
0.0485	84.0	4452	0.1567	0.9632
0.0388	85.0	4505	0.1548	0.9621
0.0421	86.0	4558	0.1484	0.9621
0.0375	87.0	4611	0.1681	0.9597
0.0376	88.0	4664	0.1513	0.9632
0.0514	89.0	4717	0.1485	0.9642
0.0598	90.0	4770	0.1541	0.9638
0.0431	91.0	4823	0.1474	0.9628
0.0432	92.0	4876	0.1498	0.9645
0.0391	93.0	4929	0.1506	0.9645
0.0408	94.0	4982	0.1462	0.9642
0.0335	95.0	5035	0.1509	0.9652
0.0447	96.0	5088	0.1508	0.9635
0.0477	97.0	5141	0.1510	0.9635
0.0504	98.0	5194	0.1510	0.9642
0.0406	99.0	5247	0.1479	0.9649
0.0343	100.0	5300	0.1480	0.9645

Framework versions

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