vit-bmr-tuned

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: 0.3154
• Accuracy: 0.8519
• Precision: 0.8610
• Recall: 0.8519
• F1: 0.8517

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
No log	1.0	160	0.3154	0.8519	0.8610	0.8519	0.8517
No log	2.0	320	0.3871	0.9012	0.9019	0.9012	0.9010
No log	3.0	480	0.5655	0.8765	0.8795	0.8765	0.8759
0.1901	4.0	640	0.5268	0.8765	0.8765	0.8765	0.8765
0.1901	5.0	800	0.5546	0.8765	0.8765	0.8765	0.8765
0.1901	6.0	960	0.5745	0.8765	0.8765	0.8765	0.8765
0.0038	7.0	1120	0.5918	0.8765	0.8765	0.8765	0.8765
0.0038	8.0	1280	0.6056	0.8889	0.8890	0.8889	0.8888
0.0038	9.0	1440	0.6164	0.8765	0.8771	0.8765	0.8763
0.0019	10.0	1600	0.6274	0.8765	0.8771	0.8765	0.8763
0.0019	11.0	1760	0.6356	0.8765	0.8771	0.8765	0.8763
0.0019	12.0	1920	0.6426	0.8765	0.8771	0.8765	0.8763
0.0013	13.0	2080	0.6485	0.8765	0.8771	0.8765	0.8763
0.0013	14.0	2240	0.6529	0.8765	0.8771	0.8765	0.8763
0.0013	15.0	2400	0.6566	0.8765	0.8771	0.8765	0.8763
0.0011	16.0	2560	0.6592	0.8765	0.8771	0.8765	0.8763
0.0011	17.0	2720	0.6610	0.8765	0.8771	0.8765	0.8763
0.0011	18.0	2880	0.6618	0.8765	0.8771	0.8765	0.8763
0.0010	19.0	3040	0.6621	0.8765	0.8771	0.8765	0.8763
0.0010	20.0	3200	0.6622	0.8765	0.8771	0.8765	0.8763

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2