vit-base-patch16-224-PMI-against-NotPMI

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

• Loss: 0.2463
• Accuracy: 0.9615

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: 0.0005
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.8	3	0.6638	0.5
No log	1.8667	7	0.6295	0.6346
0.7487	2.9333	11	0.6244	0.6346
0.7487	4.0	15	0.5469	0.6731
0.7487	4.8	18	1.5578	0.3654
0.6925	5.8667	22	0.5725	0.6538
0.6925	6.9333	26	1.0066	0.6346
0.6121	8.0	30	0.3489	0.8654
0.6121	8.8	33	0.2112	0.9038
0.6121	9.8667	37	0.2463	0.9615
0.4183	10.9333	41	0.3180	0.8654
0.4183	12.0	45	0.3389	0.8269
0.4183	12.8	48	0.5034	0.7692
0.3724	13.8667	52	0.3657	0.8269
0.3724	14.9333	56	0.2972	0.8846
0.3727	16.0	60	0.3495	0.8462

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1