# TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization

arXiv: https://arxiv.org/abs/2601.22776 · HTML: https://mlsanity.com/paper/2601.22776/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning.
- Builds on 32 of its 57 references that are in the map.
- First posted on arXiv 30 Jan 2026.

## Placed in

- LLM reasoning / [Reinforcement learning for reasoning](https://mlsanity.com/lr/reinforcement-learning-for-reasoning/)
- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 30 of 32)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Toolformer: Language Models Can Teach Themselves to Use Tools](https://mlsanity.com/paper/2302.04761/) (2023)
- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [HybridFlow: A Flexible and Efficient RLHF Framework](https://mlsanity.com/paper/2409.19256/) (2024)
- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps](https://mlsanity.com/paper/2011.01060/) (2020)
- [Text Embeddings by Weakly-Supervised Contrastive Pre-training](https://mlsanity.com/paper/2212.03533/) (2022)
- [Measuring and Narrowing the Compositionality Gap in Language Models](https://mlsanity.com/paper/2210.03350/) (2022)
- [Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions](https://mlsanity.com/paper/2212.10509/) (2022)
- [Group-in-Group Policy Optimization for LLM Agent Training](https://mlsanity.com/paper/2505.10978/) (2025)
- [The Lessons of Developing Process Reward Models in Mathematical Reasoning](https://mlsanity.com/paper/2501.07301/) (2025)
- [ToolRL: Reward is All Tool Learning Needs](https://mlsanity.com/paper/2504.13958/) (2025)
- [ZeroSearch: Incentivize the Search Capability of LLMs without Searching](https://mlsanity.com/paper/2505.04588/) (2025)
- [Reinforcing Multi-Turn Reasoning in LLM Agents via Fine-Grained Reward Structure and Credit Assignment](https://mlsanity.com/paper/2505.11821/) (2025)
- [SPA-RL: Reinforcing LLM Agents via Stepwise Progress Attribution](https://mlsanity.com/paper/2505.20732/) (2025)
- [Process vs. Outcome Reward: Which is Better for Agentic RAG Reinforcement Learning](https://mlsanity.com/paper/2505.14069/) (2025)
- [Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective](https://mlsanity.com/paper/2510.10150/) (2025)
- [CriticSearch: Fine-Grained Credit Assignment for Search Agents via a Retrospective Critic](https://mlsanity.com/paper/2511.12159/) (2025)
- [ReCreate: Reasoning and Creating Domain Agents Driven by Experience](https://mlsanity.com/paper/2601.11100/) (2026)
- [ExpSeek: Self-Triggered Experience Seeking for Web Agents](https://mlsanity.com/paper/2601.08605/) (2026)
- [Kardia-R1: Unleashing LLMs to Reason toward Understanding and Empathy for Emotional Support via Rubric-as-Judge Reinforcement Learning](https://mlsanity.com/paper/2512.01282/) (2025)
- [Lying with Truths: Open-Channel Multi-Agent Collusion for Belief Manipulation via Generative Montage](https://mlsanity.com/paper/2601.01685/) (2026)
- [Talk2Image: A Multi-Agent System for Multi-Turn Image Generation and Editing](https://mlsanity.com/paper/2508.06916/) (2025)
- [POLO: Preference-Guided Multi-Turn Reinforcement Learning for Lead Optimization](https://mlsanity.com/paper/2509.21737/) (2025)
- [Tapas Are Free! Training-Free Adaptation of Programmatic Agents via LLM-Guided Program Synthesis in Dynamic Environments](https://mlsanity.com/paper/2508.11425/) (2025)

## Built on by (most cited first, 5 of 5)

- [Progress-conditioned Group Policy Optimization for Long-Horizon Agentic Tasks](https://mlsanity.com/paper/2607.22724/) (2026)
- [MolMem: Memory-Augmented Agentic Reinforcement Learning for Sample-Efficient Molecular Optimization](https://mlsanity.com/paper/2604.12237/) (2026)
- [Dissecting Failure Dynamics in Large Language Model Reasoning](https://mlsanity.com/paper/2604.14528/) (2026)
- [HindSearch: Trajectory-Level Hindsight Critique for Search-Augmented Reinforcement Learning](https://mlsanity.com/paper/2608.01597/) (2026)
- [SIPO: Selective-Inference Policy Optimization for Tree-Structured Agentic RL](https://mlsanity.com/paper/2609.34805/) (2026)

## Abstract (arXiv)

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.