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: 0.4961
Accuracy: 0.8252
Precision: 0.8358
Recall: 0.8252
F1: 0.8286

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=5e-09
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
1.7673	1.0	321	0.9546	0.6890	0.6435	0.6890	0.6302
1.1951	2.0	642	0.7244	0.7438	0.7325	0.7438	0.7199
1.0711	3.0	963	0.6499	0.7552	0.7394	0.7552	0.7224
0.9944	4.0	1284	0.5907	0.7590	0.7962	0.7590	0.7683
0.9231	5.0	1605	0.6988	0.7084	0.8054	0.7084	0.7306
0.9889	6.0	1926	0.5912	0.7746	0.7919	0.7746	0.7808
0.8818	7.0	2247	0.6374	0.7569	0.8001	0.7569	0.7697
0.7973	8.0	2568	0.6658	0.7580	0.7925	0.7580	0.7684
0.7525	9.0	2889	0.5220	0.8044	0.8124	0.8044	0.8068
0.6938	10.0	3210	0.5634	0.7899	0.8335	0.7899	0.7980
0.6354	11.0	3531	0.4961	0.8252	0.8358	0.8252	0.8286
0.5602	12.0	3852	0.5486	0.8141	0.8276	0.8141	0.8185
0.44	13.0	4173	0.6554	0.8141	0.8442	0.8141	0.8155
0.4704	14.0	4494	0.5704	0.8235	0.8431	0.8235	0.8287
0.4275	15.0	4815	0.5563	0.8141	0.8459	0.8141	0.8230
0.3511	16.0	5136	0.5933	0.8072	0.8402	0.8072	0.8166
0.2853	17.0	5457	0.5246	0.8436	0.8542	0.8436	0.8470
0.2691	18.0	5778	0.5257	0.8509	0.8551	0.8509	0.8519
0.2134	19.0	6099	0.6391	0.8332	0.8553	0.8332	0.8404
0.224	20.0	6420	0.6297	0.8488	0.8537	0.8488	0.8497
0.1843	21.0	6741	0.6199	0.8582	0.8561	0.8582	0.8541

Framework versions

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

sharren/vit-epsilon-5e-9

vit-epsilon-5e-9

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

sharren/
vit-epsilon-5e-9