Zhejiang/Baidu Study: Decision Models Teach Better Than They Act
A Zhejiang University and Baidu paper finds decision models perform poorly as actors in long-horizon agents, but serving as teachers—even when weak—substantially boosts agent success rates.
2026-11-13 ~ 2026-11-13 · 2 related posts
- Poor-acting decision models can still teach RL agents well on ALFWorld and WebShop — ZhejiangUniversity · 2026-11-13
- Decision models act poorly but teach well: GRPO teacher lifts Qwen 1.5B to 82.9% on ALFWorld — omarsar0 · 2026-11-13