General Question Answering on PopQA (test)
49.3EMWorkflow-R1-Search
Evaluation Results
| Method | Links | |
|---|---|---|
| Workflow-R1-SearchMulti-turn=true, Backbone=Qwen2.5-32B-Instruct2026.02 | 49.3 | |
| ReCalRouting Strategy=RL-based Routing, Task Decomposition=false2026.06 | 46.6 | |
| ReCal+Routing Strategy=RL-based Routing, Task Decomposition=false2026.06 | 46.6 | |
| Search-R1 (GRPO)Multi-turn=true, Backbone=Qwen2.5-32B-Instruct2026.02 | 42.7 | |
| Router-R1Routing Strategy=RL-based Routing, Task Decomposition=false2026.06 | 40 | |
| Search-R1 (PPO)Multi-turn=true, Backbone=Qwen2.5-32B-Instruct2026.02 | 39.7 | |
| RAGRouting Strategy=No Routing, Task Decomposition=false2026.06 | 36.6 | |
| Largest LLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 35.4 | |
| Prompt LLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 34 | |
| Search-R1Routing Strategy=No Routing, Task Decomposition=false2026.06 | 32.4 | |
| RouterDCRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 28.2 | |
| GraphRouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 28 | |
| Prompt LLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=true2026.06 | 25.6 | |
| Workflow-R1Multi-turn=true, Backbone=Qwen2.5-32B-Instruct2026.02 | 24.6 | |
| AFlowMulti-turn=false, Backbone=Qwen2.5-32B-Instruct2026.02 | 23.2 | |
| KNN RouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=true2026.06 | 23.2 | |
| KNN RouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 22.2 | |
| MLP RouterRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 22.2 | |
| SC (CoT×5)Multi-turn=false, Backbone=Qwen2.5-32B-Instruct2026.02 | 21.4 | |
| MedPromptMulti-turn=false, Backbone=Qwen2.5-32B-Instruct2026.02 | 21.1 | |
| CoTMulti-turn=false, Backbone=Qwen2.5-32B-Instruct2026.02 | 20.5 | |
| MaASMulti-turn=false, Backbone=Qwen2.5-32B-Instruct2026.02 | 19.3 | |
| RouteLLMRouting Strategy=Heuristic & Discriminative Routing, Task Decomposition=false2026.06 | 19.2 | |
| CoTRouting Strategy=No Routing, Task Decomposition=false2026.06 | 16 | |
| SFTRouting Strategy=No Routing, Task Decomposition=false2026.06 | 16 | |
| DirectMulti-turn=false, Backbone=Qwen2.5-32B-Instruct2026.02 | 13.3 | |
| VanillaRouting Strategy=No Routing, Task Decomposition=false2026.06 | 12.2 |