Question Answering on NQ poison @ Position 1, k=10 (test)
64.2Robustness AccuracyMIS
Evaluation Results
| Method | Links | |
|---|---|---|
| MISModel=GPT-4o-mini, k=102025.09 | 64.2 | |
| Vanilla RAGModel=GPT-4o-mini, k=102025.09 | 58.6 | |
| RobustRAGModel=GPT-4o-mini, k=102025.09 | 58 | |
| InstructRAGModel=Mistral-7B, k=102025.09 | 57.6 | |
| AstuteRAGModel=GPT-4o-mini, k=102025.09 | 57 | |
| MISModel=Mistral-7B, k=102025.09 | 56.8 | |
| MISModel=Llama3.2-3B, k=102025.09 | 56.8 | |
| Vanilla RAGModel=Mistral-7B, k=102025.09 | 56.4 | |
| AstuteRAGModel=Mistral-7B, k=102025.09 | 52.6 | |
| RobustRAGModel=Llama3.2-3B, k=102025.09 | 50.6 | |
| InstructRAGModel=GPT-4o-mini, k=102025.09 | 49.2 | |
| AstuteRAGModel=Llama3.2-3B, k=102025.09 | 48.6 | |
| Vanilla RAGModel=Llama3.2-3B, k=102025.09 | 47.8 | |
| InstructRAGModel=Llama3.2-3B, k=102025.09 | 47.4 | |
| RobustRAGModel=Mistral-7B, k=102025.09 | 44.4 |