vit-base-patch16-224-in21k_covid_19_ct_scans

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.6385
• Accuracy: 0.9010
• F1: 0.4740
• Auc: 0.4971
• Recall: 0.9943
• Precision: 0.9058

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Auc	Recall	Precision
0.7218	1.0	55	0.3383	0.9062	0.4754	0.5	1.0	0.9062
0.7218	2.0	110	0.3823	0.9062	0.4754	0.5	1.0	0.9062
0.7218	3.0	165	0.3957	0.9062	0.4754	0.5	1.0	0.9062
0.7218	4.0	220	0.4485	0.9062	0.4754	0.5	1.0	0.9062
0.7218	5.0	275	0.4786	0.8958	0.4725	0.4943	0.9885	0.9053
0.7218	6.0	330	0.5316	0.9010	0.4740	0.4971	0.9943	0.9058
0.7218	7.0	385	0.5539	0.9010	0.4740	0.4971	0.9943	0.9058
0.7218	8.0	440	0.5800	0.9010	0.4740	0.4971	0.9943	0.9058
0.7218	9.0	495	0.5977	0.9010	0.4740	0.4971	0.9943	0.9058
0.0987	10.0	550	0.6110	0.9010	0.4740	0.4971	0.9943	0.9058
0.0987	11.0	605	0.6211	0.9010	0.4740	0.4971	0.9943	0.9058
0.0987	12.0	660	0.6288	0.9010	0.4740	0.4971	0.9943	0.9058
0.0987	13.0	715	0.6341	0.9010	0.4740	0.4971	0.9943	0.9058
0.0987	14.0	770	0.6374	0.9010	0.4740	0.4971	0.9943	0.9058
0.0987	15.0	825	0.6385	0.9010	0.4740	0.4971	0.9943	0.9058

Framework versions

• Transformers 4.41.1
• Pytorch 2.0.0+cu117
• Datasets 2.19.1
• Tokenizers 0.19.1