vit_fold_4_v3

This model is a fine-tuned version of google/vit-base-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.0697
• Accuracy: 0.9712
• F1 Score: 0.9704
• Recall: 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: 1e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 150
• num_epochs: 100
• label_smoothing_factor: 0.15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Score	Recall
2.8349	1.0	20	2.7592	0.3654	0.3301	0.3382
2.6126	2.0	40	2.6007	0.4615	0.4046	0.4215
2.3036	3.0	60	2.3497	0.5801	0.5389	0.5412
1.9396	4.0	80	2.0293	0.7179	0.7116	0.6921
1.6020	5.0	100	1.6924	0.8397	0.8420	0.8279
1.3336	6.0	120	1.4336	0.8942	0.8944	0.8876
1.2089	7.0	140	1.3085	0.9199	0.9193	0.9120
1.1917	8.0	160	1.2627	0.9327	0.9320	0.9275
1.1060	9.0	180	1.2247	0.9327	0.9318	0.9303
1.0978	10.0	200	1.2057	0.9391	0.9389	0.9361
1.0722	11.0	220	1.1787	0.9487	0.9480	0.9448
1.0497	12.0	240	1.1722	0.9455	0.9449	0.9410
1.0300	13.0	260	1.1581	0.9455	0.9439	0.9402
1.0434	14.0	280	1.1715	0.9423	0.9412	0.9354
1.0186	15.0	300	1.1747	0.9455	0.9443	0.9379
1.0203	16.0	320	1.1336	0.9519	0.9504	0.9462
1.0076	17.0	340	1.1313	0.9519	0.9508	0.9462
1.0057	18.0	360	1.1145	0.9647	0.9637	0.9625
0.9954	19.0	380	1.1122	0.9583	0.9572	0.9556
0.9902	20.0	400	1.1076	0.9679	0.9671	0.9643
0.9843	21.0	420	1.0975	0.9679	0.9671	0.9643
0.9962	22.0	440	1.1017	0.9679	0.9671	0.9643
0.9876	23.0	460	1.1234	0.9583	0.9576	0.9531
0.9796	24.0	480	1.1045	0.9583	0.9572	0.9537
0.9692	25.0	500	1.1016	0.9679	0.9671	0.9643
0.9778	26.0	520	1.0916	0.9679	0.9671	0.9643
0.9809	27.0	540	1.0911	0.9647	0.9639	0.9606
0.9709	28.0	560	1.0943	0.9615	0.9608	0.9568
0.9755	29.0	580	1.0872	0.9583	0.9576	0.9531
0.9787	30.0	600	1.0821	0.9615	0.9608	0.9568
0.9836	31.0	620	1.0912	0.9551	0.9540	0.9500
0.9717	32.0	640	1.0837	0.9647	0.9639	0.9606
0.9681	33.0	660	1.0974	0.9583	0.9576	0.9531
0.9797	34.0	680	1.0785	0.9647	0.9639	0.9606
0.9681	35.0	700	1.0790	0.9647	0.9639	0.9606
0.9665	36.0	720	1.0857	0.9647	0.9640	0.9608
0.9720	37.0	740	1.0834	0.9615	0.9604	0.9574
0.9682	38.0	760	1.0980	0.9583	0.9572	0.9537
0.9632	39.0	780	1.0789	0.9679	0.9671	0.9643
0.9639	40.0	800	1.0697	0.9712	0.9704	0.9683
0.9748	41.0	820	1.0892	0.9551	0.9540	0.9500
0.9618	42.0	840	1.0735	0.9647	0.9636	0.9612
0.9627	43.0	860	1.0646	0.9712	0.9700	0.9689
0.9722	44.0	880	1.0720	0.9679	0.9668	0.9651
0.9654	45.0	900	1.0841	0.9583	0.9572	0.9537
0.9714	46.0	920	1.0720	0.9712	0.9704	0.9683
0.9661	47.0	940	1.0861	0.9615	0.9608	0.9568
0.9666	48.0	960	1.0710	0.9712	0.9704	0.9683
0.9630	49.0	980	1.0730	0.9712	0.9704	0.9683
0.9672	50.0	1000	1.0658	0.9679	0.9671	0.9643
0.9726	51.0	1020	1.0684	0.9679	0.9671	0.9643
0.9643	52.0	1040	1.0791	0.9583	0.9576	0.9531
0.9602	53.0	1060	1.0747	0.9712	0.9704	0.9683

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 5.0.0
• Tokenizers 0.22.2