seizure_vit_jlb_231126_ff_raw_combo_multichannel

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the JLB-JLB/seizure_detection_224x224_raw_freq_combo_multichannel dataset. It achieves the following results on the evaluation set:

• Loss: 0.6323
• Roc Auc: 0.7311

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: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Roc Auc
0.3219	0.17	1000	0.6337	0.7311
0.3625	0.34	2000	0.6381	0.7038
0.3435	0.51	3000	0.6890	0.7010
0.2538	0.68	4000	0.7449	0.6927
0.2545	0.85	5000	0.7448	0.7049
0.1387	1.02	6000	1.0363	0.7127
0.1765	1.19	7000	0.8541	0.7223
0.1385	1.36	8000	1.0059	0.7228
0.1602	1.53	9000	0.8951	0.7070
0.1158	1.7	10000	1.0356	0.7220
0.0973	1.87	11000	1.0616	0.7132

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.1
• Datasets 2.15.0
• Tokenizers 0.15.0