Question Answering on TQA poison @ Position 10, k=10 (test)
71Robustness AccuracyMIS
Evaluation Results
| Method | Links | |
|---|---|---|
| MISModel=GPT-4o-mini, k=102025.09 | 71 | |
| AstuteRAGModel=GPT-4o-mini, k=102025.09 | 67.2 | |
| MISModel=Mistral-7B, k=102025.09 | 64.4 | |
| MISModel=Llama3.2-3B, k=102025.09 | 64.2 | |
| RobustRAGModel=GPT-4o-mini, k=102025.09 | 62.4 | |
| RobustRAGModel=Llama3.2-3B, k=102025.09 | 59 | |
| RobustRAGModel=Mistral-7B, k=102025.09 | 58.4 | |
| Vanilla RAGModel=GPT-4o-mini, k=102025.09 | 57.6 | |
| AstuteRAGModel=Mistral-7B, k=102025.09 | 56.6 | |
| InstructRAGModel=GPT-4o-mini, k=102025.09 | 52 | |
| AstuteRAGModel=Llama3.2-3B, k=102025.09 | 50.8 | |
| InstructRAGModel=Llama3.2-3B, k=102025.09 | 37.4 | |
| InstructRAGModel=Mistral-7B, k=102025.09 | 36.8 | |
| Vanilla RAGModel=Llama3.2-3B, k=102025.09 | 35 | |
| Vanilla RAGModel=Mistral-7B, k=102025.09 | 17.6 |