Abstention on PopQA
81.6Abstain AccuracyR2C
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| R2CRAG Framework=ReSearch, Confidence Threshold=0.92025.10 | 81.6 | 84.4 | |
| R2CRAG Framework=Search-o1, Confidence Threshold=0.92025.10 | 80.4 | 86.42 | |
| P(true)RAG Framework=ReSearch, Confidence Threshold=0.92025.10 | 78.8 | 81.2 | |
| R2CRAG Framework=SelfAsk, Confidence Threshold=0.92025.10 | 77.2 | 82.35 | |
| ReaCRAG Framework=ReSearch, Confidence Threshold=0.92025.10 | 77.2 | 81.25 | |
| R2CRAG Framework=ReAct, Confidence Threshold=0.92025.10 | 77 | 83.69 | |
| R2CRAG Framework=Search-R1, Confidence Threshold=0.92025.10 | 77 | 80.8 | |
| ReaCRAG Framework=Search-o1, Confidence Threshold=0.92025.10 | 75.6 | 81.24 | |
| SelfCRAG Framework=ReSearch, Confidence Threshold=0.92025.10 | 75.2 | 81.71 | |
| SelfCRAG Framework=Search-o1, Confidence Threshold=0.92025.10 | 75 | 80.5 | |
| ReaCRAG Framework=ReAct, Confidence Threshold=0.92025.10 | 74.4 | 79.01 | |
| SelfCRAG Framework=ReAct, Confidence Threshold=0.92025.10 | 74 | 79.1 | |
| P(true)RAG Framework=Search-o1, Confidence Threshold=0.92025.10 | 72.6 | 76.5 | |
| P(true)RAG Framework=Search-R1, Confidence Threshold=0.92025.10 | 70.8 | 71.03 | |
| P(true)RAG Framework=SelfAsk, Confidence Threshold=0.92025.10 | 70.6 | 73.6 | |
| P(true)RAG Framework=ReAct, Confidence Threshold=0.92025.10 | 70.4 | 72.07 | |
| ReaCRAG Framework=SelfAsk, Confidence Threshold=0.92025.10 | 69.2 | 73.54 | |
| SelfCRAG Framework=SelfAsk, Confidence Threshold=0.92025.10 | 68.6 | 72.6 | |
| ReaCRAG Framework=Search-R1, Confidence Threshold=0.92025.10 | 67.8 | 67.6 | |
| SelfCRAG Framework=Search-R1, Confidence Threshold=0.92025.10 | 65.4 | 62.31 | |
| RrrCRAG Framework=SelfAsk, Confidence Threshold=0.92025.10 | 61.4 | 62.52 | |
| RrrCRAG Framework=ReAct, Confidence Threshold=0.92025.10 | 60.8 | 71.59 | |
| RrrCRAG Framework=Search-R1, Confidence Threshold=0.92025.10 | 60.6 | 55.12 | |
| RrrCRAG Framework=ReSearch, Confidence Threshold=0.92025.10 | 59.2 | 54.86 | |
| RrrCRAG Framework=Search-o1, Confidence Threshold=0.92025.10 | 56.6 | 56.33 |