vit-facial-expression

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

• Loss: 0.8099
• Accuracy: 0.7378
• Precision: 0.7303
• Recall: 0.7378
• F1: 0.7274

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: 3e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
2.0154	0.4831	100	1.8473	0.3168	0.3005	0.3168	0.1961
1.569	0.9662	200	1.3372	0.5767	0.5110	0.5767	0.4926
1.185	1.4493	300	1.1047	0.6474	0.5805	0.6474	0.5937
1.044	1.9324	400	0.9638	0.6896	0.6787	0.6896	0.6611
0.8813	2.4155	500	0.8928	0.7005	0.6822	0.7005	0.6646
0.7925	2.8986	600	0.8209	0.7301	0.7183	0.7301	0.7186

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.20.3