vit-weight-decay-1e-5

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: 0.4632
• Accuracy: 0.8460
• Precision: 0.8510
• Recall: 0.8460
• F1: 0.8480

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
• 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
1.7624	1.0	321	0.9686	0.7077	0.6754	0.7077	0.6680
1.1455	2.0	642	0.7167	0.7340	0.7348	0.7340	0.7184
1.0313	3.0	963	0.6458	0.7583	0.7586	0.7583	0.7305
0.9864	4.0	1284	0.5631	0.7774	0.7907	0.7774	0.7821
0.931	5.0	1605	0.5847	0.7850	0.7882	0.7850	0.7784
0.9641	6.0	1926	0.5276	0.7899	0.7935	0.7899	0.7906
0.8935	7.0	2247	0.7242	0.7226	0.7970	0.7226	0.7430
0.7589	8.0	2568	0.6404	0.7445	0.7985	0.7445	0.7604
0.7225	9.0	2889	0.5415	0.7975	0.8100	0.7975	0.7986
0.6964	10.0	3210	0.5357	0.7871	0.8323	0.7871	0.8009
0.6232	11.0	3531	0.5579	0.8003	0.8272	0.8003	0.8084
0.5781	12.0	3852	0.6126	0.7847	0.8315	0.7847	0.7978
0.4713	13.0	4173	0.6180	0.8259	0.8343	0.8259	0.8161
0.4834	14.0	4494	0.5668	0.8096	0.8426	0.8096	0.8181
0.3886	15.0	4815	0.4632	0.8460	0.8510	0.8460	0.8480
0.3654	16.0	5136	0.6023	0.8065	0.8375	0.8065	0.8168
0.2904	17.0	5457	0.5002	0.8537	0.8626	0.8537	0.8558
0.2865	18.0	5778	0.5731	0.8332	0.8583	0.8332	0.8408
0.2122	19.0	6099	0.6130	0.8325	0.8606	0.8325	0.8411
0.2227	20.0	6420	0.6097	0.8485	0.8531	0.8485	0.8494
0.179	21.0	6741	0.5650	0.8693	0.8633	0.8693	0.8639
0.1257	22.0	7062	0.5759	0.8714	0.8712	0.8714	0.8707
0.1265	23.0	7383	0.6089	0.8710	0.8684	0.8710	0.8688
0.1146	24.0	7704	0.6169	0.8769	0.8737	0.8769	0.8744
0.1368	25.0	8025	0.5994	0.8745	0.8743	0.8745	0.8739

Framework versions

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