paper-with-me

Papers

Data-Efficient RLVR via Off-Policy Influence Guidance

2025-10-30 · Erle Zhu, Dazhi Jiang, Yuan Wang, Xujun Li, Jiale Cheng, Yuxian Gu, Yilin Niu, Aohan Zeng, Jie Tang, Minlie Huang, Hongning Wang arxiv

Data selection is a critical aspect of Reinforcement Learning with Verifiable Rewards (RLVR) for enhancing the reasoning capabilities of large language models (LLMs). Current data selection methods are largely heuristic-based, lacking theoretical guarantees and generalizability. This work proposes a theoretically-grounded approach using influence functions to estimate the contribution of each data point to the learning objective. To overcome the prohibitive computational cost of policy rollouts required for online influence estimation, we introduce an off-policy influence estimation method that efficiently approximates data influence using pre-collected offline trajectories. Furthermore, to manage the high-dimensional gradients of LLMs, we employ sparse random projection to reduce dimensionality and improve storage and computation efficiency. Leveraging these techniques, we develop \textbf{C}urriculum \textbf{R}L with \textbf{O}ff-\textbf{P}olicy \text{I}nfluence guidance (\textbf{CROPI}), a multi-stage RL framework that iteratively selects the most influential data for the current policy. Experiments on models up to 7B parameters demonstrate that CROPI significantly accelerates training. On a 1.5B model, it achieves a 2.66x step-level acceleration while using only 10\% of the data per stage compared to full-dataset training. Our results highlight the substantial potential of influence-based data selection for efficient RLVR.

📄 PDF Abstract BibTeX arXiv:2510.26491

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Learning to Reason under Off-Policy Guidance

2025-04-21 · Jianhao Yan, Yafu Li, Zican Hu, Zhi Wang 외

Recent advances in large reasoning models (LRMs) demonstrate that sophisticated behaviors such as multi-step reasoning and self-reflection can emerge via reinforcement learning with verifiable rewards~(\textit{RLVR}). Ho…

MathReinforcement Learning (RL)

Agent-RLVR: Training Software Engineering Agents via Guidance and Environment Rewards

2025-06-13 · Jeff Da, Clinton Wang, Xiang Deng, Yuntao Ma 외

Reinforcement Learning from Verifiable Rewards (RLVR) has been widely adopted as the de facto method for enhancing the reasoning capabilities of large language models and has demonstrated notable success in verifiable do…

MathNavigate

On-policy Distillation with Verifiable Reward

2026-08-25 · Wenze Lin, Jiale Zhao, Xitai Jiang, Songde Rao 외 hf

Reinforcement Learning with Verifiable Rewards (RLVR) and on-policy distillation (OPD) have become two widely adopted paradigms for post-training large language models. However, RLVR suffers from sparse task-level feedba…

Reinforcement Learning

Think Outside the Policy: In-Context Steered Policy Optimization

2025-10-30 · Hsiu-Yuan Huang, Chenming Tang, Weijie Liu, Clive Bai 외 arxiv

Existing Reinforcement Learning from Verifiable Rewards (RLVR) methods, such as Group Relative Policy Optimization (GRPO), have achieved remarkable progress in improving the reasoning capabilities of Large Reasoning Mode…

Reinforcement LearningMathematical Reasoning

Teaching the Way, Not the Answer: Privileged Tutoring Distillation for Multimodal Policy Optimization

2026-06-05 · Shizhe Xiang, Ke An, Wenlong Yu, Yue Liu 외 arxiv

Recent post-training methods, particularly Reinforcement Learning with Verifiable Rewards (RLVR), have significantly enhanced the reasoning ability of Large Vision-Language Models (LVLMs). However, the sparse nature of v…

Reinforcement LearningMultimodal Reasoning