swinv2-tiny-patch4-window8-256-dmae-humeda-DAV72

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

• Loss: 0.7140
• Accuracy: 0.8743

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.0851	1.0	15	1.0597	0.4857
0.9595	2.0	30	0.9427	0.64
0.8642	3.0	45	0.6913	0.7086
0.5836	4.0	60	0.5780	0.7257
0.5395	5.0	75	0.4822	0.7829
0.4215	6.0	90	0.4077	0.8229
0.4329	7.0	105	0.4352	0.8114
0.3695	8.0	120	0.3244	0.8743
0.3314	9.0	135	0.3186	0.8914
0.3176	10.0	150	0.3788	0.8514
0.3368	11.0	165	0.3458	0.8629
0.2558	12.0	180	0.4196	0.8457
0.2579	13.0	195	0.3485	0.8743
0.2413	14.0	210	0.4509	0.8629
0.2374	15.0	225	0.3904	0.8743
0.2214	16.0	240	0.3461	0.8514
0.2189	17.0	255	0.5986	0.8229
0.2458	18.0	270	0.3360	0.8914
0.2431	19.0	285	0.3475	0.8857
0.2136	20.0	300	0.3242	0.88
0.1871	21.0	315	0.4103	0.8857
0.1996	22.0	330	0.3606	0.9029
0.1367	23.0	345	0.4657	0.8629
0.1963	24.0	360	0.4267	0.8743
0.1519	25.0	375	0.4322	0.8686
0.1365	26.0	390	0.4214	0.88
0.1158	27.0	405	0.4472	0.8743
0.1621	28.0	420	0.4020	0.8743
0.1271	29.0	435	0.4054	0.8857
0.136	30.0	450	0.4286	0.9143
0.1386	31.0	465	0.5015	0.8857
0.1153	32.0	480	0.6675	0.8629
0.1139	33.0	495	0.5458	0.8971
0.144	34.0	510	0.5303	0.88
0.1542	35.0	525	0.5164	0.8914
0.1208	36.0	540	0.5690	0.88
0.1034	37.0	555	0.7427	0.8571
0.0889	38.0	570	0.9084	0.8286
0.1355	39.0	585	0.5977	0.8743
0.0895	40.0	600	0.5400	0.8914
0.1072	41.0	615	0.6018	0.8743
0.1356	42.0	630	0.5493	0.8743
0.0953	43.0	645	0.5350	0.8914
0.0781	44.0	660	0.5269	0.88
0.0854	45.0	675	0.5428	0.88
0.0983	46.0	690	0.4897	0.8857
0.0944	47.0	705	0.5177	0.8971
0.1152	48.0	720	0.6401	0.8629
0.0608	49.0	735	0.7380	0.8629
0.0898	50.0	750	0.4922	0.8971
0.0923	51.0	765	0.5427	0.8971
0.0743	52.0	780	0.9941	0.84
0.0753	53.0	795	0.5342	0.8857
0.0751	54.0	810	0.6452	0.88
0.1222	55.0	825	0.6297	0.8743
0.0786	56.0	840	0.6592	0.8629
0.134	57.0	855	0.6541	0.8686
0.092	58.0	870	0.6523	0.8571
0.1036	59.0	885	0.5562	0.8971
0.0825	60.0	900	0.6117	0.8743
0.0923	61.0	915	0.5778	0.8686
0.0909	62.0	930	0.5974	0.8686
0.0536	63.0	945	0.7557	0.8514
0.0572	64.0	960	0.6255	0.8857
0.0824	65.0	975	0.6768	0.8686
0.0773	66.0	990	0.5942	0.9029
0.0495	67.0	1005	0.7902	0.8571
0.0649	68.0	1020	0.6097	0.8914
0.0852	69.0	1035	0.6614	0.8914
0.0634	70.0	1050	0.6604	0.8914
0.0774	71.0	1065	0.7848	0.8514
0.0803	72.0	1080	0.6424	0.8914
0.0645	73.0	1095	0.7508	0.8857
0.0483	74.0	1110	0.7523	0.8629
0.0586	75.0	1125	0.8278	0.8629
0.1	76.0	1140	0.7503	0.8686
0.0434	77.0	1155	0.7820	0.8743
0.0792	78.0	1170	0.7016	0.88
0.055	79.0	1185	0.8635	0.8571
0.0666	80.0	1200	0.7208	0.8686
0.0563	81.0	1215	0.7606	0.8686
0.0535	82.0	1230	0.7329	0.88
0.0499	83.0	1245	0.7253	0.88
0.0418	84.0	1260	0.7429	0.8686
0.0736	85.0	1275	0.7621	0.8743
0.0593	86.0	1290	0.7970	0.8571
0.0658	87.0	1305	0.7211	0.8686
0.0531	88.0	1320	0.7420	0.8686
0.0604	89.0	1335	0.7151	0.8743
0.0661	90.0	1350	0.6881	0.8857
0.058	91.0	1365	0.7139	0.8686
0.0436	92.0	1380	0.7260	0.8686
0.0733	93.0	1395	0.7150	0.8743
0.0501	93.3390	1400	0.7140	0.8743

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 2.19.0
• Tokenizers 0.21.1