yonad2008/predict_llama2_7b

Jailbreak Prediction Model: llama2:7b

Fine-tuned DeBERTa-v3-base for detecting unsafe/jailbreak prompts in multi-turn conversations.

Evaluation Results (best fold: 4)

Metric	Value
F1	0.9014
PR-AUC	0.9574
ROC-AUC	0.9897
Precision	0.8533
Recall	0.9552
Best Threshold	0.35

Training Details

• Base model: microsoft/deberta-v3-base
• Target model: llama2:7b
• Datasets: HarmBench
• K-Folds: 5
• Epochs: 5
• Learning Rate: 2e-05
• Max Length: 512
• Input format: turns only

Dataset Size (before turn expansion)

Original rows (after cleaning and balancing): 1630 (unsafe: 340, safe: 1290)