Question Answering on TriviaQA In-Distribution
96.09Exact Match (EM)SkillFlow
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SkillFlowMethod Category=Ours2026.05 | 96.09 | 96.94 | |
| Qwen3.5-AFlowMethod Category=AFlow, Backbone=Qwen3.52026.05 | 92.97 | 93.25 | |
| FlowSteerMethod Category=Agent+RL2026.05 | 91.41 | 92.06 | |
| AgentFlowMethod Category=Agent+RL2026.05 | 89.84 | 90.47 | |
| SkillRLMethod Category=Agent+RL2026.05 | 89.06 | 90.63 | |
| v4-flashMethod Category=Baseline2026.05 | 74.02 | 81.99 | |
| Qwen3.5-SFTMethod Category=SFT, Backbone=Qwen3.52026.05 | 47.24 | 55.73 | |
| Qwen3.5-GRPOMethod Category=GRPO, Backbone=Qwen3.52026.05 | 47.24 | 56.93 | |
| Qwen3.5-9BMethod Category=Baseline2026.05 | 44.88 | 54.3 |