Preprint
Sep 2026
World Model-Guided Reinforcement Learning via Counterfactual User Engagement Simulation
Experiments show that UEWM provides reliable and transferable reward signals across domains, and that WMG-RL enables a compact 1.7B student policy to match or surpass much larger LLMs on downstream recommendation tasks.
Ang Li, Xin Xu, Bin Liang et al.
· 0 citations