ToSCA: Leveraging Hierarchical Reinforcement Learning on Temporal and Strategic Abstractions of Conversational Agents
A two-level hierarchical reinforcement learning (HRL) framework for conversational agents that bridges the gap between existing token-level and utterance-level RL methods and introduces a dual-granularity reward mechanism that combines the utterance-level satisfaction score with token-level intrinsic self-consistency a...
Xiaoyu Wang, Qing-Qing Gu, Yue Zhao et al.
· 0 citations