vit-beta2-0.9995

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: 0.5000
• Accuracy: 0.8519
• Precision: 0.8575
• Recall: 0.8519
• F1: 0.8527

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=(0.9,0.9995) and epsilon=1e-08
• 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
1.7395	1.0	321	0.9629	0.6845	0.6040	0.6845	0.5948
1.1477	2.0	642	0.7630	0.7174	0.7438	0.7174	0.7138
1.0466	3.0	963	0.6211	0.7660	0.7543	0.7660	0.7373
0.9366	4.0	1284	0.6310	0.7354	0.7934	0.7354	0.7518
0.9629	5.0	1605	0.5584	0.7982	0.7934	0.7982	0.7865
0.9753	6.0	1926	0.5606	0.7920	0.7922	0.7920	0.7916
0.8629	7.0	2247	0.7154	0.6969	0.8040	0.6969	0.7238
0.7817	8.0	2568	0.6276	0.7597	0.8105	0.7597	0.7743
0.7355	9.0	2889	0.5259	0.8110	0.8215	0.8110	0.8133
0.6878	10.0	3210	0.5347	0.7885	0.8388	0.7885	0.8022
0.6206	11.0	3531	0.5543	0.7958	0.8375	0.7958	0.8079
0.5628	12.0	3852	0.6465	0.7819	0.8285	0.7819	0.7960
0.4607	13.0	4173	0.6070	0.8190	0.8411	0.8190	0.8195
0.4558	14.0	4494	0.6088	0.8138	0.8424	0.8138	0.8215
0.3984	15.0	4815	0.5507	0.8308	0.8486	0.8308	0.8343
0.3541	16.0	5136	0.5112	0.8412	0.8528	0.8412	0.8448
0.3203	17.0	5457	0.5000	0.8519	0.8575	0.8519	0.8527
0.2685	18.0	5778	0.5630	0.8429	0.8574	0.8429	0.8472
0.214	19.0	6099	0.5670	0.8405	0.8607	0.8405	0.8464
0.2195	20.0	6420	0.6209	0.8478	0.8557	0.8478	0.8500
0.1598	21.0	6741	0.5597	0.8630	0.8585	0.8630	0.8590
0.1174	22.0	7062	0.5682	0.8637	0.8617	0.8637	0.8624
0.1341	23.0	7383	0.6186	0.8662	0.8638	0.8662	0.8622
0.1307	24.0	7704	0.5961	0.8727	0.8710	0.8727	0.8701
0.1075	25.0	8025	0.5912	0.8738	0.8715	0.8738	0.8717
0.0741	26.0	8346	0.5991	0.8755	0.8711	0.8755	0.8713
0.1476	27.0	8667	0.5900	0.8745	0.8705	0.8745	0.8714

Framework versions

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