swordphish-url-detection

This model is a fine-tuned version of distilbert/distilbert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0618
• Accuracy: 0.9782
• F1: 0.9717
• Precision: 0.9715
• Recall: 0.9719

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: 2e-05
• train_batch_size: 2048
• eval_batch_size: 2048
• 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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.1675	0.4452	500	0.1054	0.9601	0.9485	0.9423	0.9548
0.1016	0.8905	1000	0.0829	0.9694	0.9603	0.9580	0.9627
0.083	1.3357	1500	0.0747	0.9710	0.9627	0.9528	0.9729
0.0753	1.7809	2000	0.0697	0.9734	0.9658	0.9580	0.9736
0.0682	2.2262	2500	0.0660	0.9757	0.9685	0.9676	0.9695
0.063	2.6714	3000	0.0647	0.9759	0.9689	0.9631	0.9748
0.0596	3.1167	3500	0.0631	0.9771	0.9704	0.9681	0.9727
0.0553	3.5619	4000	0.0620	0.9776	0.9709	0.9699	0.9720
0.0554	4.0071	4500	0.0619	0.9784	0.9719	0.9737	0.9702
0.0524	4.4524	5000	0.0617	0.9781	0.9716	0.9710	0.9722
0.0527	4.8976	5500	0.0618	0.9782	0.9717	0.9715	0.9719

Framework versions

• Transformers 4.56.0
• Pytorch 2.8.0+cu129
• Datasets 4.4.1
• Tokenizers 0.22.0