Saved in:
Bibliographic Details
Main Authors: Wang, Yutong, Ji, Pengliang, Li, Kaixin, Bi, Baolong, Feng, Tao, Sartoretti, Guillaume
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2508.03018
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916016879042560
author Wang, Yutong
Ji, Pengliang
Li, Kaixin
Bi, Baolong
Feng, Tao
Sartoretti, Guillaume
author_facet Wang, Yutong
Ji, Pengliang
Li, Kaixin
Bi, Baolong
Feng, Tao
Sartoretti, Guillaume
contents Large Language Reasoning Models have demonstrated remarkable success on static tasks, yet their application to multi-round agentic planning in interactive environments faces two fundamental challenges. First, the intractable credit assignment problem renders conventional reinforcement learning ineffective in sparse-reward settings. Second, the computational overhead of verbose, step-by-step reasoning histories is prohibitive. To address these challenges, we propose BPO, a three-stage framework (bootstrapping, extrapolation, and refinement) that establishes a self-improving data flywheel to develop robust reasoning models for long-horizon, sparse-reward environments. Our framework first bootstraps efficient reasoning using the proposed planning quaternions with long-short chain-of-thought fusion. It then extrapolates to out-of-distribution tasks through complexity-stratified curriculum learning. Finally, the model iteratively refines itself by learning exclusively on experiences selected via reward-gated rejection sampling. Experiments on ALFWorld, ScienceWorld, and WebShop demonstrate that our approach achieves state-of-the-art with significant token efficiency, providing a new recipe for reasoning models in agentic planning.
format Preprint
id arxiv_https___arxiv_org_abs_2508_03018
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Policy Optimization: A Data Curation Flywheel for Sparse-Reward Long-Horizon Planning
Wang, Yutong
Ji, Pengliang
Li, Kaixin
Bi, Baolong
Feng, Tao
Sartoretti, Guillaume
Artificial Intelligence
Robotics
Large Language Reasoning Models have demonstrated remarkable success on static tasks, yet their application to multi-round agentic planning in interactive environments faces two fundamental challenges. First, the intractable credit assignment problem renders conventional reinforcement learning ineffective in sparse-reward settings. Second, the computational overhead of verbose, step-by-step reasoning histories is prohibitive. To address these challenges, we propose BPO, a three-stage framework (bootstrapping, extrapolation, and refinement) that establishes a self-improving data flywheel to develop robust reasoning models for long-horizon, sparse-reward environments. Our framework first bootstraps efficient reasoning using the proposed planning quaternions with long-short chain-of-thought fusion. It then extrapolates to out-of-distribution tasks through complexity-stratified curriculum learning. Finally, the model iteratively refines itself by learning exclusively on experiences selected via reward-gated rejection sampling. Experiments on ALFWorld, ScienceWorld, and WebShop demonstrate that our approach achieves state-of-the-art with significant token efficiency, providing a new recipe for reasoning models in agentic planning.
title Beyond Policy Optimization: A Data Curation Flywheel for Sparse-Reward Long-Horizon Planning
topic Artificial Intelligence
Robotics
url https://arxiv.org/abs/2508.03018