Exfiltration on ACIARENA Science
40.1UASelf Consistency
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Self ConsistencyBase LLM=GPT-4o-mini2026.04 | 40.1 | 41.09 | |
| LLM DebateBase LLM=GPT-4o-mini2026.04 | 37.5 | 11.67 | |
| AgentVerseBase LLM=GPT-4o-mini2026.04 | 35.53 | 11.36 | |
| AutoGenBase LLM=GPT-4o-mini2026.04 | 25.95 | 23.24 | |
| CAMELBase LLM=GPT-4o-mini2026.04 | 25.64 | 0.57 |