Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Ma, Shichao, Ma, Zhiyuan, Yang, Ming, Li, Xiaofan, Wu, Xing, Du, Jintao, Cheng, Yu, Wang, Weiqiang, Liu, Qiliang, Zhou, Zhengyang, Wang, Yang
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2601.22776
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918429386080256
author Ma, Shichao
Ma, Zhiyuan
Yang, Ming
Li, Xiaofan
Wu, Xing
Du, Jintao
Cheng, Yu
Wang, Weiqiang
Liu, Qiliang
Zhou, Zhengyang
Wang, Yang
author_facet Ma, Shichao
Ma, Zhiyuan
Yang, Ming
Li, Xiaofan
Wu, Xing
Du, Jintao
Cheng, Yu
Wang, Weiqiang
Liu, Qiliang
Zhou, Zhengyang
Wang, Yang
contents Multi-turn tool-integrated reasoning enables Large Language Models (LLMs) to solve complex tasks through iterative information retrieval. However, current reinforcement learning (RL) frameworks for search-augmented reasoning predominantly rely on sparse outcome-level rewards, leading to a "Double Homogenization Dilemma." This manifests as (1) Process homogenization, where the thinking, reasoning, and tooling involved in generation are ignored. (2) Intra-group homogenization, coarse-grained outcome rewards often lead to inefficiencies in intra-group advantage estimation with methods like Group Relative Policy Optimization (GRPO) during sampling. To address this, we propose Turn-level Stage-aware Policy Optimization (TSPO). TSPO introduces the First-Occurrence Latent Reward (FOLR) mechanism, allocating partial rewards to the step where the ground-truth answer first appears, thereby preserving process-level signals and increasing reward variance within groups without requiring external reward models or any annotations. Extensive experiments demonstrate that TSPO significantly outperforms state-of-the-art baselines, achieving average performance gains of 24% and 13.6% on Qwen2.5-3B and 7B models, respectively. Code is available at https://github.com/Flipped-May/TSPO.
format Preprint
id arxiv_https___arxiv_org_abs_2601_22776
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization
Ma, Shichao
Ma, Zhiyuan
Yang, Ming
Li, Xiaofan
Wu, Xing
Du, Jintao
Cheng, Yu
Wang, Weiqiang
Liu, Qiliang
Zhou, Zhengyang
Wang, Yang
Artificial Intelligence
Multi-turn tool-integrated reasoning enables Large Language Models (LLMs) to solve complex tasks through iterative information retrieval. However, current reinforcement learning (RL) frameworks for search-augmented reasoning predominantly rely on sparse outcome-level rewards, leading to a "Double Homogenization Dilemma." This manifests as (1) Process homogenization, where the thinking, reasoning, and tooling involved in generation are ignored. (2) Intra-group homogenization, coarse-grained outcome rewards often lead to inefficiencies in intra-group advantage estimation with methods like Group Relative Policy Optimization (GRPO) during sampling. To address this, we propose Turn-level Stage-aware Policy Optimization (TSPO). TSPO introduces the First-Occurrence Latent Reward (FOLR) mechanism, allocating partial rewards to the step where the ground-truth answer first appears, thereby preserving process-level signals and increasing reward variance within groups without requiring external reward models or any annotations. Extensive experiments demonstrate that TSPO significantly outperforms state-of-the-art baselines, achieving average performance gains of 24% and 13.6% on Qwen2.5-3B and 7B models, respectively. Code is available at https://github.com/Flipped-May/TSPO.
title TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization
topic Artificial Intelligence
url https://arxiv.org/abs/2601.22776