vit_4090_3_7

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

• Loss: 0.0353
• Accuracy: 0.9908
• Precision: 0.9907
• Recall: 0.9908
• F1: 0.9906

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: 0.0001
• train_batch_size: 24
• eval_batch_size: 4
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.3443	1.0	531	0.2528	0.9216	0.9410	0.9216	0.9264
0.1457	2.0	1062	0.1413	0.9590	0.9589	0.9590	0.9584
0.0938	3.0	1593	0.1220	0.9611	0.9599	0.9611	0.9597
0.0737	4.0	2124	0.0718	0.9781	0.9760	0.9781	0.9764
0.0513	5.0	2655	0.1124	0.9731	0.9734	0.9731	0.9721
0.0389	6.0	3186	0.1038	0.9760	0.9766	0.9760	0.9754
0.0314	7.0	3717	0.0736	0.9837	0.9847	0.9837	0.9835
0.0291	8.0	4248	0.0958	0.9802	0.9805	0.9802	0.9800
0.0183	9.0	4779	0.0914	0.9823	0.9828	0.9823	0.9823
0.014	10.0	5310	0.0916	0.9781	0.9783	0.9781	0.9776
0.0171	11.0	5841	0.0554	0.9866	0.9869	0.9866	0.9863
0.009	12.0	6372	0.0778	0.9830	0.9835	0.9830	0.9827
0.0049	13.0	6903	0.0761	0.9873	0.9873	0.9873	0.9871
0.0057	14.0	7434	0.0628	0.9880	0.9879	0.9880	0.9877
0.0032	15.0	7965	0.0563	0.9887	0.9887	0.9887	0.9885
0.0028	16.0	8496	0.0695	0.9852	0.9854	0.9852	0.9850
0.0021	17.0	9027	0.0470	0.9887	0.9887	0.9887	0.9886
0.0006	18.0	9558	0.0403	0.9908	0.9908	0.9908	0.9907
0.0001	19.0	10089	0.0354	0.9922	0.9922	0.9922	0.9920
0.0001	20.0	10620	0.0353	0.9908	0.9907	0.9908	0.9906

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1
• Datasets 3.2.0
• Tokenizers 0.21.1