Question Answering on TriviaQA (TQA) (test) (Robust Accuracy)
75.4Robust AccuracySampling + MIS
Evaluation Results
| Method | Links | |
|---|---|---|
| Sampling + MISModel=GPT-4o-mini, Poison Position=25, k=502025.09 | 75.4 | |
| Sampling + MISModel=GPT-4o-mini, Poison Position=50, k=502025.09 | 74.6 | |
| Sampling + MISModel=Mistral-7B, Poison Position=50, k=502025.09 | 67.6 | |
| AstuteRAGModel=GPT-4o-mini, Poison Position=50, k=502025.09 | 66.4 | |
| Sampling + MISModel=Mistral-7B, Poison Position=25, k=502025.09 | 65.8 | |
| AstuteRAGModel=GPT-4o-mini, Poison Position=25, k=502025.09 | 65.8 | |
| Sampling + MISModel=Llama3.2-3B, Poison Position=25, k=502025.09 | 64 | |
| RobustRAGModel=GPT-4o-mini, Poison Position=25, k=502025.09 | 63.8 | |
| RobustRAGModel=GPT-4o-mini, Poison Position=50, k=502025.09 | 63 | |
| RobustRAGModel=GPT-4o-mini, Poison Position=1, k=502025.09 | 62.6 | |
| Sampling + MISModel=Llama3.2-3B, Poison Position=50, k=502025.09 | 62 | |
| Vanilla RAGModel=GPT-4o-mini, Poison Position=25, k=502025.09 | 61.8 | |
| AstuteRAGModel=GPT-4o-mini, Poison Position=1, k=502025.09 | 61 | |
| Vanilla RAGModel=GPT-4o-mini, Poison Position=50, k=502025.09 | 60.8 | |
| AstuteRAGModel=Llama3.2-3B, Poison Position=1, k=502025.09 | 60.6 | |
| Sampling + MISModel=GPT-4o-mini, Poison Position=1, k=502025.09 | 59.6 | |
| AstuteRAGModel=Mistral-7B, Poison Position=1, k=502025.09 | 57.8 | |
| RobustRAGModel=Llama3.2-3B, Poison Position=25, k=502025.09 | 57.8 | |
| RobustRAGModel=Llama3.2-3B, Poison Position=1, k=502025.09 | 57.6 | |
| RobustRAGModel=Llama3.2-3B, Poison Position=50, k=502025.09 | 57.6 | |
| Sampling + MISModel=Mistral-7B, Poison Position=1, k=502025.09 | 56.2 | |
| AstuteRAGModel=Mistral-7B, Poison Position=25, k=502025.09 | 55.4 | |
| InstructRAGModel=GPT-4o-mini, Poison Position=50, k=502025.09 | 55.2 | |
| AstuteRAGModel=Mistral-7B, Poison Position=50, k=502025.09 | 55 | |
| InstructRAGModel=GPT-4o-mini, Poison Position=25, k=502025.09 | 54.6 | |
| RobustRAGModel=Mistral-7B, Poison Position=50, k=502025.09 | 54.4 | |
| RobustRAGModel=Mistral-7B, Poison Position=25, k=502025.09 | 54.2 | |
| RobustRAGModel=Mistral-7B, Poison Position=1, k=502025.09 | 53.4 | |
| Sampling + MISModel=Llama3.2-3B, Poison Position=1, k=502025.09 | 53.2 | |
| AstuteRAGModel=Llama3.2-3B, Poison Position=25, k=502025.09 | 53 | |
| InstructRAGModel=Mistral-7B, Poison Position=1, k=502025.09 | 51.2 | |
| Vanilla RAGModel=Mistral-7B, Poison Position=1, k=502025.09 | 49.6 | |
| Vanilla RAGModel=Llama3.2-3B, Poison Position=1, k=502025.09 | 49.2 | |
| AstuteRAGModel=Llama3.2-3B, Poison Position=50, k=502025.09 | 46.6 | |
| InstructRAGModel=Llama3.2-3B, Poison Position=1, k=502025.09 | 46.4 | |
| InstructRAGModel=GPT-4o-mini, Poison Position=1, k=502025.09 | 44.2 | |
| Vanilla RAGModel=GPT-4o-mini, Poison Position=1, k=502025.09 | 42.4 | |
| InstructRAGModel=Mistral-7B, Poison Position=25, k=502025.09 | 41.8 | |
| InstructRAGModel=Llama3.2-3B, Poison Position=25, k=502025.09 | 38.6 | |
| Vanilla RAGModel=Llama3.2-3B, Poison Position=25, k=502025.09 | 37.6 | |
| InstructRAGModel=Mistral-7B, Poison Position=50, k=502025.09 | 35.4 | |
| InstructRAGModel=Llama3.2-3B, Poison Position=50, k=502025.09 | 33.8 | |
| Vanilla RAGModel=Llama3.2-3B, Poison Position=50, k=502025.09 | 25.6 | |
| Vanilla RAGModel=Mistral-7B, Poison Position=25, k=502025.09 | 22.6 | |
| Vanilla RAGModel=Mistral-7B, Poison Position=50, k=502025.09 | 11.6 |