ktp-spoof-clip

This model is a fine-tuned version of openai/clip-vit-base-patch32 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0740
• Accuracy: 0.9853

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.8889	4	0.5501	0.8088
No log	2.0	9	0.3671	0.8529
0.5611	2.8889	13	0.3852	0.8235
0.5611	4.0	18	0.2422	0.9118
0.4558	4.8889	22	0.3534	0.8824
0.4558	6.0	27	0.1137	0.9412
0.3562	6.8889	31	0.5266	0.7941
0.3562	8.0	36	0.1918	0.9118
0.1201	8.8889	40	0.0301	1.0
0.1201	10.0	45	0.0450	0.9853
0.1201	10.8889	49	0.0327	0.9853
0.0604	12.0	54	0.0898	0.9706
0.0604	12.8889	58	0.0789	0.9853
0.0322	13.3333	60	0.0740	0.9853

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2
• Datasets 2.19.2
• Tokenizers 0.19.1