Question Answering on NGQA
69.29F1Oracle
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Oracle2026.04 | 69.29 | 27 | |
| EvolveRouterPrompt Optimization=Enabled, Adaptive K=Enabled2026.04 | 54.04 | 9.67 | |
| EvolveRouter (w/o Adaptive K)Prompt Optimization=Enabled, Adaptive K=Disabled2026.04 | 53.28 | 9.67 | |
| EvolveRouter (w/o PO)Prompt Optimization=Disabled, Adaptive K=Enabled2026.04 | 50.3 | 9 | |
| AgentRouter2026.04 | 49.95 | 8.67 | |
| MLP Router2026.04 | 48.8 | 4.33 | |
| Majority Vote2026.04 | 48.42 | 4 | |
| Best Agent2026.04 | 48.41 | 4 | |
| HybridLLM2026.04 | 48.05 | 2.33 | |
| Best LLM2026.04 | 48.03 | 3.17 | |
| Average2026.04 | 47.5 | 4.04 | |
| Random Router2026.04 | 47.41 | 4.03 | |
| RouteLLM2026.04 | 47.2 | 4 | |
| CascadeRouting2026.04 | 47.06 | 3.33 | |
| GraphRouter2026.04 | 46.88 | 4.17 | |
| KNN Router2026.04 | 46.11 | 4 | |
| LLM-Blender2026.04 | 42.15 | 1.67 |