modernbert-phishing-classifier

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

• Loss: 0.3066
• Accuracy: 0.9
• Auc: 0.965

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: 0.0002
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Auc
0.3714	1.0	263	0.2936	0.869	0.949
0.2622	2.0	526	0.2681	0.884	0.96
0.2405	3.0	789	0.2642	0.898	0.961
0.2091	4.0	1052	0.2688	0.893	0.963
0.2078	5.0	1315	0.3813	0.882	0.962
0.1887	6.0	1578	0.2667	0.9	0.965
0.1695	7.0	1841	0.2851	0.902	0.964
0.1654	8.0	2104	0.2935	0.902	0.964
0.157	9.0	2367	0.3169	0.904	0.966
0.158	10.0	2630	0.3190	0.896	0.964
0.149	11.0	2893	0.3019	0.893	0.965
0.1437	12.0	3156	0.2995	0.9	0.965
0.1365	13.0	3419	0.3048	0.9	0.965
0.1312	14.0	3682	0.3090	0.898	0.965
0.1304	15.0	3945	0.3066	0.9	0.965

Framework versions

• Transformers 4.48.3
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.0