# SIPO: Selective-Inference Policy Optimization for Tree-Structured Agentic RL

arXiv: https://arxiv.org/abs/2609.34805 · HTML: https://mlsanity.com/paper/2609.34805/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 24 of its 35 references that are in the map.
- First posted on arXiv 28 Sep 2026.

## Placed in

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

## Builds on (most cited first, 24 of 24)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [TreeRL: LLM Reinforcement Learning with On-Policy Tree Search](https://mlsanity.com/paper/2506.11902/) (2025)
- [Hindsight Credit Assignment for Long-Horizon LLM Agents](https://mlsanity.com/paper/2603.08754/) (2026)
- [Agentic Entropy-Balanced Policy Optimization](https://mlsanity.com/paper/2510.14545/) (2025)
- [TIPS: Turn-Level Information-Potential Reward Shaping for Search-Augmented LLMs](https://mlsanity.com/paper/2603.22293/) (2026)
- [Group-Graph Policy Optimization for Long-Horizon Agentic Reinforcement Learning](https://mlsanity.com/paper/2606.22995/) (2026)
- [Branching Policy Optimization: Sandbox-Native Language Agent Reinforcement Learning](https://mlsanity.com/paper/2607.14171/) (2026)
- [TRACE: A Unified Rollout Budget Allocation Framework for Efficient Agentic Reinforcement Learning](https://mlsanity.com/paper/2606.11119/) (2026)
- [Enhancing LLM-based Search Agents via Contribution Weighted Group Relative Policy Optimization](https://mlsanity.com/paper/2604.14267/) (2026)
- [GAGPO: Generalized Advantage Grouped Policy Optimization](https://mlsanity.com/paper/2605.13217/) (2026)
- [AT2PO: Agentic Turn-based Policy Optimization via Tree Search](https://mlsanity.com/paper/2601.04767/) (2026)
- [TSR: Trajectory-Search Rollouts for Multi-Turn RL of LLM Agents](https://mlsanity.com/paper/2602.11767/) (2026)
- [BiPACE: Bisimulation-Guided Policy Optimization with Action Counterfactual Estimation for LLM Agents](https://mlsanity.com/paper/2606.25556/) (2026)
- [Self-Play Meets Skill Evolution: Self-Evolving Search Agents that Pose, Solve, and Remember](https://mlsanity.com/paper/2607.29468/) (2026)
- [TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization](https://mlsanity.com/paper/2601.22776/) (2026)
- [APPO: Agentic Procedural Policy Optimization](https://mlsanity.com/paper/2606.12384/) (2026)
- [Optimizing What Policies Learn From: Recoverability-aware Rollout Intervention Learning](https://mlsanity.com/paper/2608.05080/) (2026)
- [Long-Horizon Plan Execution in Large Tool Spaces through Entropy-Guided Branching](https://mlsanity.com/paper/2604.12126/) (2026)
- [CoEvoKG: Co-Evolving Knowledge Graphs with Self-Evolving Search Agents](https://mlsanity.com/paper/2608.01904/) (2026)
- [Maximizing Rollout Informativeness under a Fixed Budget: A Submodular View of Tree Search for Tool-Use Agentic Reinforcement Learning](https://mlsanity.com/paper/2605.05262/) (2026)
- [Information Gain-based Rollout Policy Optimization: An Adaptive Tree-Structured Rollout Approach for Multi-Turn LLM Agents](https://mlsanity.com/paper/2607.06223/) (2026)
- [Process Reward Informed Tree Rollout for Effective Multi-Turn RL](https://mlsanity.com/paper/2607.15610/) (2026)
- [Early Verdicts, Better Budgets: Sequential Adaptive Rollout Allocation for Compute-Efficient RLVR](https://mlsanity.com/paper/2607.26253/) (2026)

## Abstract (arXiv)

Tree-structured reinforcement learning trains search agents by comparing alternative continuations and propagating terminal rewards to intermediate decisions. Adaptive expansion, however, creates a statistical asymmetry: an incumbent is selected using its own generation statistic, whereas fresh siblings are sampled after selection. When that statistic is associated with return, branch values can reflect selection history as well as continuation quality, even for a shared parent. We propose Selective-Inference Policy Optimization (SIPO), which incorporates this distinction into tree-based credit estimation. Its scale-free branch criterion keeps generation scores and sibling penalties on a consistent relative scale; exchangeable branching supplies multiple fresh continuations from each selected parent; and order-statistic correction adjusts retained incumbent values using selection rank and the estimated score--outcome association. These mechanisms preserve the leaf budget and the host policy optimisation objective. Across seven QA benchmarks using Qwen3-4B, Qwen3-8B, and Qwen2.5-7B, SIPO achieves the highest reported multi-hop and single-hop averages among the compared methods. On Qwen3-8B, it improves these averages over ATtextsuperscript2PO by 1.31 and 1.07 percentage points, respectively, and ranks first on six of seven benchmarks. Component ablations evaluate the individual and combined changes, while early-training paired diagnostics show a selected--fresh value gap alongside a near-zero fresh--fresh reference. Together, these results support accounting for selection history when constructing and evaluating search-agent rollouts. Our code is available at https://github.com/Zenghuang-Fu/SIPO