Question Answering on LogiQA (TSR/ASR Metrics)
74.5Task Success Rate (TSR)STAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| STARAgent Model=Qwen-Plus, Attack Type=Group Collusive Attack2026.05 | 74.5 | 18.25 | |
| G-SafeguardAgent Model=Qwen-Plus, Attack Type=Group Collusive Attack2026.05 | 71.75 | 22 | |
| STARAgent Model=GPT-3.5-Turbo, Attack Type=Group Collusive Attack2026.05 | 52 | 35.5 | |
| STARAgent Model=GPT-4-Turbo, Attack Type=Group Collusive Attack2026.05 | 49.5 | 41.75 | |
| G-SafeguardAgent Model=GPT-4-Turbo, Attack Type=Group Collusive Attack2026.05 | 48.75 | 42.5 | |
| G-SafeguardAgent Model=GPT-3.5-Turbo, Attack Type=Group Collusive Attack2026.05 | 44.5 | 46.25 |