vit_fold_5_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: 1.1346
• Accuracy: 0.9614
• F1 Score: 0.9617
• Recall: 0.9606

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
2.7256	1.0	20	2.7334	0.3280	0.3191	0.3196
2.5857	2.0	40	2.5720	0.4437	0.4325	0.4316
2.3319	3.0	60	2.3313	0.6141	0.6091	0.6021
1.9213	4.0	80	2.0386	0.7042	0.7069	0.7017
1.5856	5.0	100	1.7432	0.7814	0.7834	0.7753
1.3525	6.0	120	1.5248	0.8521	0.8514	0.8461
1.2066	7.0	140	1.3952	0.8907	0.8886	0.8882
1.1562	8.0	160	1.3169	0.9100	0.9097	0.9089
1.1321	9.0	180	1.2701	0.9228	0.9234	0.9251
1.0852	10.0	200	1.2451	0.9260	0.9259	0.9248
1.0566	11.0	220	1.2291	0.9325	0.9330	0.9297
1.0369	12.0	240	1.2170	0.9293	0.9305	0.9284
1.0472	13.0	260	1.1885	0.9389	0.9398	0.9388
0.9939	14.0	280	1.1825	0.9357	0.9366	0.9351
0.9927	15.0	300	1.1735	0.9421	0.9433	0.9428
0.9894	16.0	320	1.1678	0.9486	0.9494	0.9496
1.0003	17.0	340	1.1651	0.9453	0.9464	0.9469
0.9758	18.0	360	1.1610	0.9518	0.9527	0.9540
0.9899	19.0	380	1.1561	0.9486	0.9495	0.9507
0.9770	20.0	400	1.1512	0.9453	0.9461	0.9459
0.9722	21.0	420	1.1555	0.9486	0.9497	0.9515
0.9767	22.0	440	1.1526	0.9453	0.9460	0.9463
0.9686	23.0	460	1.1519	0.9421	0.9432	0.9451
0.9719	24.0	480	1.1459	0.9486	0.9498	0.9507
0.9666	25.0	500	1.1407	0.9550	0.9560	0.9571
0.9709	26.0	520	1.1363	0.9582	0.9591	0.9596
0.9659	27.0	540	1.1423	0.9453	0.9467	0.9482
0.9657	28.0	560	1.1392	0.9518	0.9528	0.9531
0.9702	29.0	580	1.1357	0.9550	0.9556	0.9556
0.9616	30.0	600	1.1353	0.9518	0.9528	0.9531
0.9620	31.0	620	1.1356	0.9550	0.9556	0.9563
0.9716	32.0	640	1.1589	0.9518	0.9518	0.9481
0.9575	33.0	660	1.1339	0.9582	0.9584	0.9587
0.9620	34.0	680	1.1295	0.9550	0.9556	0.9556
0.9616	35.0	700	1.1349	0.9518	0.9524	0.9519
0.9644	36.0	720	1.1389	0.9550	0.9554	0.9544
0.9607	37.0	740	1.1354	0.9582	0.9585	0.9568
0.9618	38.0	760	1.1290	0.9550	0.9556	0.9556
0.9570	39.0	780	1.1356	0.9582	0.9584	0.9587
0.9616	40.0	800	1.1356	0.9582	0.9585	0.9568
0.9633	41.0	820	1.1332	0.9550	0.9556	0.9556
0.9586	42.0	840	1.1289	0.9550	0.9556	0.9556
0.9562	43.0	860	1.1291	0.9550	0.9556	0.9556
0.9560	44.0	880	1.1349	0.9582	0.9584	0.9587
0.9620	45.0	900	1.1353	0.9518	0.9519	0.9506
0.9588	46.0	920	1.1472	0.9550	0.9556	0.9556
0.9620	47.0	940	1.1461	0.9518	0.9526	0.9512
0.9543	48.0	960	1.1380	0.9550	0.9554	0.9544
0.9601	49.0	980	1.1448	0.9486	0.9491	0.9450
0.9565	50.0	1000	1.1297	0.9582	0.9584	0.9587
0.9623	51.0	1020	1.1281	0.9582	0.9586	0.9581
0.9615	52.0	1040	1.1296	0.9582	0.9585	0.9568
0.9650	53.0	1060	1.1335	0.9550	0.9556	0.9556
0.9568	54.0	1080	1.1393	0.9582	0.9585	0.9568
0.9596	55.0	1100	1.1359	0.9582	0.9586	0.9581
0.9552	56.0	1120	1.1346	0.9614	0.9617	0.9606
0.9607	57.0	1140	1.1322	0.9582	0.9585	0.9568
0.9624	58.0	1160	1.1349	0.9582	0.9586	0.9581
0.9583	59.0	1180	1.1375	0.9550	0.9556	0.9556
0.9606	60.0	1200	1.1315	0.9550	0.9556	0.9556
0.9547	61.0	1220	1.1281	0.9582	0.9586	0.9581

Framework versions

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