Indirect Prompt Injection Defense on BIPIA English
41.1ASRQwen3.5-9B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3.5-9BDefense=No Defense (ND)2026.05 | 41.1 | — | |
| BanglaLLaMA-3-8BDefense=No Defense (ND)2026.05 | 41.1 | — | |
| Gemma-3-12B-ITDefense=No Defense (ND)2026.05 | 31.7 | — | |
| Macro avg.Defense=No Defense (ND)2026.05 | 31.1 | — | |
| TigerLLM-9B-ITDefense=No Defense (ND)2026.05 | 30.1 | — | |
| Qwen2.5-7B-InstructDefense=No Defense (ND)2026.05 | 28.7 | — | |
| Gemma-3-12B-ITDefense=MIPIAD (D)2026.05 | 28 | 3.7 | |
| TigerLLM-9B-ITDefense=MIPIAD (D)2026.05 | 26.7 | 3.4 | |
| Qwen3-8BDefense=No Defense (ND)2026.05 | 26.1 | — | |
| BanglaLLaMA-3-8BDefense=MIPIAD (D)2026.05 | 24.9 | 16.2 | |
| Qwen2.5-7B-InstructDefense=MIPIAD (D)2026.05 | 23.7 | 5 | |
| Macro avg.Defense=MIPIAD (D)2026.05 | 21.2 | 9.9 | |
| BLOOMZ-7B1-MTDefense=No Defense (ND)2026.05 | 18.7 | — | |
| Qwen3-8BDefense=MIPIAD (D)2026.05 | 17.3 | 8.8 | |
| BLOOMZ-7B1-MTDefense=MIPIAD (D)2026.05 | 17.2 | 1.5 | |
| Qwen3.5-9BDefense=MIPIAD (D)2026.05 | 10.7 | 30.4 |