HISR: Hindsight Information Modulated Segmental Process Rewards For Multi-turn Agentic Reinforcement Learning

ACL ARR 2026 January Submission8821 Authors

Keywords: Hindsight Information, Segmental Process Rewards, Agent, Reinforcement Learning

Abstract: While large language models excel in diverse domains, their performance on complex long-horizon agentic decision-making tasks remains limited. Most existing methods concentrate on designing effective reward models (RMs) to advance performance via multi-turn reinforcement learning. However, they suffer from delayed propagation in sparse outcome rewards and unreliable credit assignment with potentially overly fine-grained and unfocused turn-level process rewards. In this paper, we propose ($\textbf{HISR}$) exploiting $\textbf{H}$indsight $\textbf{I}$nformation to modulate $\textbf{S}$egmental process $\textbf{R}$ewards, which closely aligns rewards with sub-goals and underscores significant segments to enhance the reliability of credit assignment. Specifically, a segment-level process RM is presented to assign rewards for each sub-goal in the task, avoiding excessively granular allocation to turns. To emphasize significant segments in the trajectory, a hindsight model is devised to reflect the preference of performing a certain action after knowing the trajectory outcome. With this characteristic, we design the ratios of sequence likelihoods between hindsight and policy model to measure action importance. The ratios are subsequently employed to aggregate segment importance scores, which in turn modulate segmental process rewards, enhancing credit assignment reliability. Extensive experimental results on three publicly benchmarks demonstrate the validity of our method.

Paper Type: Long

Research Area: AI/LLM Agents

Research Area Keywords: LLM agents, reinforcement learning in agents

Contribution Types: NLP engineering experiment

Languages Studied: English

Submission Number: 8821