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

Loss: 0.5348
Accuracy: 0.8350
Precision: 0.8391
Recall: 0.8350
F1: 0.8350

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: 16
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.99) and epsilon=1e-07
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 1733
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.765	1.0	321	0.9570	0.6917	0.6487	0.6917	0.6531
1.1815	2.0	642	0.6949	0.7559	0.7315	0.7559	0.7314
1.0605	3.0	963	0.6213	0.7715	0.7649	0.7715	0.7530
1.0147	4.0	1284	0.5768	0.7732	0.7807	0.7732	0.7734
0.93	5.0	1605	0.6572	0.7587	0.7940	0.7587	0.7662
0.9793	6.0	1926	0.6165	0.7701	0.7940	0.7701	0.7742
0.8662	7.0	2247	0.6535	0.7240	0.8098	0.7240	0.7456
0.7767	8.0	2568	0.5813	0.7566	0.8124	0.7566	0.7733
0.7572	9.0	2889	0.5385	0.8145	0.8131	0.8145	0.8114
0.7003	10.0	3210	0.5355	0.8027	0.8276	0.8027	0.8093
0.6316	11.0	3531	0.6285	0.7653	0.8322	0.7653	0.7816
0.5723	12.0	3852	0.5775	0.8017	0.8279	0.8017	0.8105
0.4717	13.0	4173	0.5348	0.8350	0.8391	0.8350	0.8350
0.4472	14.0	4494	0.5469	0.8239	0.8442	0.8239	0.8299
0.3764	15.0	4815	0.5738	0.8291	0.8501	0.8291	0.8355
0.3346	16.0	5136	0.5368	0.8436	0.8512	0.8436	0.8461
0.2826	17.0	5457	0.5510	0.8474	0.8489	0.8474	0.8468
0.2659	18.0	5778	0.5467	0.8547	0.8560	0.8547	0.8549
0.2545	19.0	6099	0.6156	0.8433	0.8617	0.8433	0.8487
0.2123	20.0	6420	0.6871	0.8429	0.8499	0.8429	0.8427
0.1655	21.0	6741	0.6139	0.8610	0.8552	0.8610	0.8567
0.1246	22.0	7062	0.6129	0.8675	0.8681	0.8675	0.8677
0.1394	23.0	7383	0.6523	0.8714	0.8675	0.8714	0.8677

Framework versions

Transformers 4.40.0.dev0
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2

sharren/vit-epsilon-1e-7

vit-epsilon-1e-7

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

sharren/
vit-epsilon-1e-7