General Question Answering on Natural Questions (NQ) v1.0 (test)
40.4Exact MatchReCal
Evaluation Results
| Method | Links | |
|---|---|---|
| ReCalRouting Strategy=RL-based Routing, Task Decomposition=false2026.06 | 40.4 | |
| ReCal+Routing Strategy=RL-based Routing, Task Decomposition=false2026.06 | 39 | |
| Router-R1Routing Strategy=RL-based Routing, Task Decomposition=false2026.06 | 33.4 | |
| Search-R1Routing Strategy=No Routing, Task Decomposition=false2026.06 | 32.8 | |
| Prompt LLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 30 | |
| RAGRouting Strategy=No Routing, Task Decomposition=false2026.06 | 29.8 | |
| Largest LLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 29.6 | |
| RouterDCRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 27.8 | |
| GraphRouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 27.6 | |
| KNN RouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 26.2 | |
| Prompt LLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=true2026.06 | 25.8 | |
| MLP RouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 25.2 | |
| KNN RouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=true2026.06 | 23.6 | |
| RouteLLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 23 | |
| SFTRouting Strategy=No Routing, Task Decomposition=false2026.06 | 21.2 | |
| CoTRouting Strategy=No Routing, Task Decomposition=false2026.06 | 12.6 | |
| VanillaRouting Strategy=No Routing, Task Decomposition=false2026.06 | 9.2 |