Question Answering on Bio poison @ Position 10, k=10 (test)
79.9Robustness Score (LLM-J)MIS
Evaluation Results
| Method | Links | |
|---|---|---|
| MISModel=GPT-4o-mini, k=102025.09 | 79.9 | |
| InstructRAGModel=GPT-4o-mini, k=102025.09 | 75.3 | |
| Vanilla RAGModel=GPT-4o-mini, k=102025.09 | 72.9 | |
| MISModel=Llama3.2-3B, k=102025.09 | 71.3 | |
| AstuteRAGModel=GPT-4o-mini, k=102025.09 | 67.9 | |
| MISModel=Mistral-7B, k=102025.09 | 65.7 | |
| RobustRAGModel=GPT-4o-mini, k=102025.09 | 64.8 | |
| RobustRAGModel=Mistral-7B, k=102025.09 | 57.3 | |
| InstructRAGModel=Mistral-7B, k=102025.09 | 57.1 | |
| RobustRAGModel=Llama3.2-3B, k=102025.09 | 51.1 | |
| Vanilla RAGModel=Mistral-7B, k=102025.09 | 46.1 | |
| AstuteRAGModel=Mistral-7B, k=102025.09 | 42.9 | |
| AstuteRAGModel=Llama3.2-3B, k=102025.09 | 40.3 | |
| Vanilla RAGModel=Llama3.2-3B, k=102025.09 | 36.5 | |
| InstructRAGModel=Llama3.2-3B, k=102025.09 | 33.1 |