Privilege Escalation on Tool selection benchmark
82Certified Upper BoundAnomaly mon.
Evaluation Results
| Method | Links | |
|---|---|---|
| Anomaly mon.Mitigation=Anomaly mon., Selector model=Llama-3.1, Generator model=Gemma-32025.10 | 82 | |
| SecAlignMitigation=SecAlign, Selector model=Llama-3.1, Generator model=Gemma-32025.10 | 77 | |
| StruQMitigation=StruQ, Selector model=Llama-3.1, Generator model=Gemma-32025.10 | 71 | |
| PerplexityMitigation=Perplexity, Selector model=Llama-3.1, Generator model=Gemma-32025.10 | 68 | |
| Defended retr.Mitigation=Defended retr., Selector model=Llama-3.1, Generator model=Gemma-32025.10 | 67 | |
| NoneMitigation=None, Selector model=Llama-3.1, Generator model=Gemma-32025.10 | 65 |