LLM Jailbreaking on Mistral CB
72Success Rate First (SRF)Adaptive Probe-based Steering
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Adaptive Probe-based Steering2026.05 | 72 | 81 | 95 | |
| RepE2026.05 | 2 | 2 | 0 | |
| Angular2026.05 | 2 | 2 | 1 | |
| SCAV2026.05 | 1 | 0 | 1 |
| Method | Links | |||
|---|---|---|---|---|
| Adaptive Probe-based Steering2026.05 | 72 | 81 | 95 | |
| RepE2026.05 | 2 | 2 | 0 | |
| Angular2026.05 | 2 | 2 | 1 | |
| SCAV2026.05 | 1 | 0 | 1 |