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: 0.5117
Accuracy: 0.8221
Precision: 0.8347
Recall: 0.8221
F1: 0.8100

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=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine_with_restarts
lr_scheduler_warmup_steps: 770
num_epochs: 12
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.5999	1.0	321	0.5250	0.8141	0.8100	0.8141	0.8011
0.4483	2.0	642	0.5117	0.8221	0.8347	0.8221	0.8100
0.3425	3.0	963	0.5709	0.8270	0.8208	0.8270	0.8107
0.176	4.0	1284	0.5579	0.8575	0.8549	0.8575	0.8544
0.0183	5.0	1605	0.5555	0.8773	0.8762	0.8773	0.8759
0.0273	6.0	1926	0.8166	0.8415	0.8369	0.8415	0.8328
0.1186	7.0	2247	0.6380	0.8617	0.8598	0.8617	0.8591
0.0166	8.0	2568	0.6608	0.8731	0.8716	0.8731	0.8709
0.0003	9.0	2889	1.0045	0.8460	0.8438	0.8460	0.8306
0.1258	10.0	3210	0.6712	0.8499	0.8527	0.8499	0.8490
0.0137	11.0	3531	0.7952	0.8738	0.8765	0.8738	0.8740
0.0004	12.0	3852	0.7956	0.8731	0.8708	0.8731	0.8699

Framework versions

Transformers 4.39.0.dev0
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2

sharren/vit-lr-cosine-restarts

vit-lr-cosine-restarts

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

sharren/
vit-lr-cosine-restarts