metadata
language: en
tags:
- jailbreak-detection
- deberta-v3
- text-classification
model-index:
- name: predict_llama3.2_3b
results:
- task:
type: text-classification
name: Jailbreak Detection
metrics:
- name: F1
type: f1
value: 0.7216
- name: PR-AUC
type: pr_auc
value: 0.7712
- name: ROC-AUC
type: roc_auc
value: 0.9199
- name: Precision
type: precision
value: 0.6306
- name: Recall
type: recall
value: 0.8434
Jailbreak Prediction Model: llama3.2:3b
Fine-tuned DeBERTa-v3-base for detecting unsafe/jailbreak prompts in multi-turn conversations.
Evaluation Results (best fold: 1)
| Metric | Value |
|---|---|
| F1 | 0.7216 |
| PR-AUC | 0.7712 |
| ROC-AUC | 0.9199 |
| Precision | 0.6306 |
| Recall | 0.8434 |
| Best Threshold | 0.20 |
Training Details
- Base model:
microsoft/deberta-v3-base - Target model:
llama3.2:3b - 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): 2096 (unsafe: 401, safe: 1695)