redline-guard-v3

This model is a fine-tuned version of answerdotai/ModernBERT-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0965
• Auprc: 0.9914
• Auroc: 0.9953
• Tpr@1fpr: 0.9082

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 128
• 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: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Auprc	Auroc	Tpr@1fpr
0.1865	0.1941	2000	0.1469	0.9734	0.9850	0.7717
0.1211	0.3882	4000	0.1419	0.9808	0.9892	0.8232
0.1109	0.5822	6000	0.1030	0.9867	0.9925	0.8727
0.1007	0.7763	8000	0.0974	0.9886	0.9937	0.8874
0.0997	0.9704	10000	0.0895	0.9904	0.9948	0.9050
0.0572	1.1645	12000	0.1256	0.9899	0.9943	0.9058
0.0622	1.3586	14000	0.0962	0.9908	0.9950	0.9098
0.0562	1.5526	16000	0.1030	0.9905	0.9949	0.9018
0.065	1.7467	18000	0.0961	0.9909	0.9951	0.9062
0.0545	1.9408	20000	0.0965	0.9914	0.9953	0.9082

Framework versions

• Transformers 4.57.6
• Pytorch 2.13.0+cu130
• Datasets 5.0.1
• Tokenizers 0.22.2