Question Answering on HotpotQA (ACC_clean, ACC, ASR)
57Accuracy (Clean)RobustRAG
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RobustRAGBase Model=Mistral-7B, Variant=Keyword2026.05 | 57 | 56 | 24 | |
| Vanilla RAGBase Model=Mistral-7B2026.05 | 55 | 17 | 48 | |
| Vanilla RAGBase Model=Vicuna-7B2026.05 | 55 | 23 | 43 | |
| RAGForensicsBase Model=Vicuna-7B2026.05 | 51 | 51 | 6 | |
| TrustRAGBase Model=Mistral-7B2026.05 | 50 | 45 | 10 | |
| PRA-RAGBase Model=Mistral-7B, Variant=cat2026.05 | 50 | 41 | 11 | |
| PRA-RAGBase Model=Vicuna-7B2026.05 | 50 | 49 | 0 | |
| PRA-RAGBase Model=Mistral-7B2026.05 | 49 | 37 | 3 | |
| PRA-RAGBase Model=Vicuna-7B, Variant=cat2026.05 | 49 | 37 | 13 | |
| RAGForensicsBase Model=Mistral-7B2026.05 | 45 | 46 | 10 | |
| RobustRAGBase Model=Vicuna-7B, Variant=Keyword2026.05 | 44 | 31 | 10 | |
| TrustRAGBase Model=Vicuna-7B2026.05 | 36 | 30 | 7 | |
| ASTUTE RAGBase Model=Mistral-7B2026.05 | 35 | 35 | 9 | |
| ASTUTE RAGBase Model=Vicuna-7B2026.05 | 34 | 26 | 30 | |
| InstructRAGBase Model=Vicuna-7B, Variant=ICL2026.05 | 31 | 9 | 11 | |
| InstructRAGBase Model=Mistral-7B, Variant=ICL2026.05 | 30 | 22 | 26 |