Multi-hop Question Answering on HotpotQA (Tokens and Time)
17,910Output TokensRAG-Critic
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RAG-CriticBaseline=Search-o12026.04 | 17,910 | 4,735 | |
| RAG-CriticBaseline=Search-R12026.04 | 15,203 | 5,511 | |
| RAG-CriticBaseline=ReAct2026.04 | 14,196 | 5,036 | |
| Step-wiseBaseline=ReAct2026.04 | 11,524 | 1,031 | |
| RerunBaseline=ReAct2026.04 | 8,714 | 713 | |
| RerunBaseline=Search-o12026.04 | 7,844 | 633 | |
| Step-wiseBaseline=Search-o12026.04 | 6,861 | 611 | |
| RerunBaseline=Search-R12026.04 | 6,457 | 652 | |
| Step-wiseBaseline=Search-R12026.04 | 5,848 | 530 | |
| DR-RAGBaseline=Search-R12026.04 | 4,594 | 270 | |
| DR-RAGBaseline=ReAct2026.04 | 4,514 | 279 | |
| DR-RAGBaseline=Search-o12026.04 | 4,340 | 274 |