swinv2-tiny-patch4-window8-256-dmae-humeda-DAV77

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: 0.3654
• Accuracy: 0.8914

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
1.1317	0.9524	15	1.0850	0.3657
1.0134	1.9524	30	1.0048	0.56
0.9444	2.9524	45	0.7731	0.7371
0.6457	3.9524	60	0.5598	0.7486
0.5593	4.9524	75	0.4836	0.7771
0.4781	5.9524	90	0.4486	0.7829
0.4769	6.9524	105	0.4365	0.8114
0.4366	7.9524	120	0.5434	0.7657
0.4248	8.9524	135	0.3744	0.8457
0.3298	9.9524	150	0.3618	0.8343
0.342	10.9524	165	0.3661	0.8629
0.3033	11.9524	180	0.3753	0.8343
0.3106	12.9524	195	0.4607	0.8229
0.264	13.9524	210	0.3623	0.8457
0.2407	14.9524	225	0.3982	0.8343
0.2758	15.9524	240	0.3694	0.8514
0.2719	16.9524	255	0.5112	0.8171
0.2311	17.9524	270	0.3977	0.8571
0.246	18.9524	285	0.4087	0.8629
0.2193	19.9524	300	0.4239	0.8343
0.2419	20.9524	315	0.3980	0.8514
0.2084	21.9524	330	0.4278	0.8686
0.1973	22.9524	345	0.3654	0.8914
0.1807	23.9524	360	0.4050	0.8629
0.1693	24.9524	375	0.5299	0.8229
0.1594	25.9524	390	0.4832	0.8514
0.1876	26.9524	405	0.5069	0.84
0.1514	27.9524	420	0.5056	0.8571
0.1818	28.9524	435	0.5403	0.8286
0.1682	29.9524	450	0.5058	0.84
0.1681	30.9524	465	0.5187	0.8114
0.1394	31.9524	480	0.5843	0.8629
0.1659	32.9524	495	0.4707	0.8629
0.1753	33.9524	510	0.5603	0.8229
0.1884	34.9524	525	0.5372	0.8343
0.1399	35.9524	540	0.5559	0.8629
0.1603	36.9524	555	0.6177	0.8629
0.1353	37.9524	570	0.5262	0.8457
0.0874	38.9524	585	0.4945	0.8629
0.1054	39.9524	600	0.6391	0.8629
0.1156	40.9524	615	0.6080	0.8514
0.1247	41.9524	630	0.6483	0.8114
0.1396	42.9524	645	0.5377	0.8457
0.117	43.9524	660	0.5460	0.8629
0.1403	44.9524	675	0.6856	0.84
0.1089	45.9524	690	0.6401	0.8514
0.1022	46.9524	705	0.6795	0.8514
0.09	47.9524	720	0.6025	0.8457
0.0948	48.9524	735	0.6489	0.8514
0.1177	49.9524	750	0.6105	0.8571
0.0797	50.9524	765	0.7485	0.8229
0.0872	51.9524	780	0.6390	0.84
0.1038	52.9524	795	0.6190	0.8743
0.1361	53.9524	810	0.6417	0.8457
0.1205	54.9524	825	0.6161	0.84
0.1026	55.9524	840	0.5836	0.8514
0.1059	56.9524	855	0.6865	0.8571
0.0999	57.9524	870	0.7455	0.8629
0.1075	58.9524	885	0.7018	0.8343
0.0952	59.9524	900	0.6851	0.8286
0.0796	60.9524	915	0.6301	0.8514
0.0952	61.9524	930	0.6734	0.8343
0.1041	62.9524	945	0.6475	0.8514
0.0961	63.9524	960	0.7369	0.8229
0.0897	64.9524	975	0.7261	0.84
0.0591	65.9524	990	0.7303	0.8286
0.106	66.9524	1005	0.6512	0.84
0.0817	67.9524	1020	0.6835	0.8229
0.0653	68.9524	1035	0.7211	0.8514
0.0801	69.9524	1050	0.7762	0.8343
0.0754	70.9524	1065	0.7669	0.8571
0.067	71.9524	1080	0.8578	0.8457
0.0896	72.9524	1095	0.8271	0.84
0.0622	73.9524	1110	0.7458	0.8286
0.0741	74.9524	1125	0.7236	0.84
0.0687	75.9524	1140	0.7986	0.84
0.0877	76.9524	1155	0.7999	0.8286
0.1034	77.9524	1170	0.7840	0.8286
0.0716	78.9524	1185	0.7871	0.8343
0.0659	79.9524	1200	0.7860	0.8571
0.0844	80.9524	1215	0.8366	0.8514
0.0858	81.9524	1230	0.8152	0.8629
0.0531	82.9524	1245	0.7717	0.8286
0.075	83.9524	1260	0.8578	0.8171
0.059	84.9524	1275	0.8240	0.8229
0.0896	85.9524	1290	0.8907	0.8343
0.0741	86.9524	1305	0.8814	0.84
0.0697	87.9524	1320	0.9080	0.8286
0.0552	88.9524	1335	0.8345	0.8343
0.0576	89.9524	1350	0.8746	0.8229
0.0729	90.9524	1365	0.8196	0.8343
0.0782	91.9524	1380	0.8073	0.8343
0.0584	92.9524	1395	0.8011	0.8286
0.0471	93.9524	1410	0.8076	0.8286
0.0544	94.9524	1425	0.8390	0.8229
0.0576	95.9524	1440	0.8575	0.8286
0.0608	96.9524	1455	0.8392	0.8229
0.064	97.9524	1470	0.8266	0.8286
0.0742	98.9524	1485	0.8311	0.8286
0.0471	99.9524	1500	0.8333	0.8286

Framework versions

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