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: 0.4778
Accuracy: 0.8311
Precision: 0.8433
Recall: 0.8311
F1: 0.8269

Training procedure

Early stopping is employed with a patience of 10 and validation loss as the stopping criteria.

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
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.6718	1.0	321	0.4934	0.8242	0.8254	0.8242	0.8163
0.4045	2.0	642	0.4778	0.8311	0.8433	0.8311	0.8269
0.2419	3.0	963	0.5133	0.8405	0.8525	0.8405	0.8410
0.1267	4.0	1284	0.6154	0.8495	0.8491	0.8495	0.8448
0.0733	5.0	1605	0.7845	0.8422	0.8421	0.8422	0.8361
0.0446	6.0	1926	0.8767	0.8471	0.8407	0.8471	0.8408
0.0523	7.0	2247	0.8674	0.8450	0.8492	0.8450	0.8407
0.0388	8.0	2568	0.9754	0.8398	0.8566	0.8398	0.8387
0.0402	9.0	2889	0.9370	0.8492	0.8547	0.8492	0.8461
0.0283	10.0	3210	0.9218	0.8509	0.8496	0.8509	0.8483
0.0451	11.0	3531	0.9872	0.8474	0.8400	0.8474	0.8401
0.0549	12.0	3852	1.0203	0.8457	0.8488	0.8457	0.8416

Framework versions

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

sharren/vit-lr-0.0001

vit-lr-0.0001

Training procedure

Training hyperparameters

Training results

Framework versions

sharren/
vit-lr-0.0001