paper-with-me

Papers

BandPO: Bridging Trust Regions and Ratio Clipping via Probability-Aware Bounds for LLM Reinforcement Learning

2026-03-05 · Yuan Li, Bo Wang, Yufei Gao, Yuqian Yao, Xinyuan Wang, Zhangyue Yin, Xipeng Qiu arxiv

Proximal constraints are fundamental to the stability of the Large Language Model reinforcement learning. While the canonical clipping mechanism in PPO serves as an efficient surrogate for trust regions, we identify a critical bottleneck: fixed bounds strictly constrain the upward update margin of low-probability actions, disproportionately suppressing high-advantage tail strategies and inducing rapid entropy collapse. To address this, we introduce Band-constrained Policy Optimization (BandPO). BandPO replaces canonical clipping with Band, a unified theoretical operator that projects trust regions defined by f-divergences into dynamic, probability-aware clipping intervals. Theoretical analysis confirms that Band effectively resolves this exploration bottleneck. We formulate this mapping as a convex optimization problem, guaranteeing a globally optimal numerical solution while deriving closed-form solutions for specific divergences. Extensive experiments across diverse models and datasets demonstrate that BandPO consistently outperforms canonical clipping and Clip-Higher, while robustly mitigating entropy collapse.

📄 PDF Abstract BibTeX arXiv:2603.04918

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

TROLL: Trust Regions improve Reinforcement Learning for Large Language Models

2025-10-04 · Philipp Becker, Niklas Freymuth, Serge Thilges, Fabian Otto 외 arxiv

Reinforcement Learning (RL) with PPO-like clip objectives has become the standard choice for reward-based fine-tuning of large language models (LLMs). Although recent work has explored improved estimators of advantages a…

Reinforcement LearningMathematical ReasoningCode Generation

MEG abnormalities and mechanisms of surgical failure in neocortical epilepsy

2022-07-13 · Thomas W. Owen, Gabrielle M. Schroeder, Vytene Janiukstyte, Gerard R. Hall 외

Neocortical epilepsy surgery fails to achieve post-operative seizure freedom in 30-40% of cases. It is not fully understood why surgery in some patients is unsuccessful. Comparing interictal MEG bandpower from patients t…

Flow-DPPO: Divergence Proximal Policy Optimization for Flow Matching Models

2026-06-09 · Bowen Ping, Xiangxin Zhou, Penghui Qi, Minnan Luo 외 arxiv

Recent work has demonstrated that online reinforcement learning (RL) can substantially improve the quality and alignment of flow matching models for image and video generation. Methods such as Flow-GRPO and CPS cast the …

Reinforcement LearningVideo Generation

It's Not You, It's Clipping: A Soft Trust-Region via Probability Smoothing for LLM RL

2025-09-25 · Madeleine Dwyer, Adam Sobey, Adriane Chapman arxiv

Training large language models (LLMs) with reinforcement learning (RL) methods such as PPO and GRPO commonly relies on ratio clipping to stabilise updates. While effective at preventing instability, clipping discards inf…

Reinforcement LearningMathematical Reasoning

Trust Region On-Policy Distillation

2026-05-31 · Xingrun Xing, Haoqing Wang, Boyan Gao, Ziheng Li 외 arxiv

On-Policy Distillation (OPD) is a fundamental technique for efficient post-training of large language models (LLMs), with broad applications in agent learning, multi-task enhancement, and model compression. However, OPD …

Mathematical ReasoningModel CompressionCode Generation