paper-with-me

Papers

Enhancing Reinforcement Learning Fine-Tuning with an Online Refiner

2026-03-18 · Hao Ma, Zhiqiang Pu, Yang Liu, Xiaolin Ai arxiv

Constraints are essential for stabilizing reinforcement learning fine-tuning (RFT) and preventing degenerate outputs, yet they inherently conflict with the optimization objective because stronger constraints limit the ability of a fine-tuned model to discover better solutions. We propose \textit{dynamic constraints} that resolve this tension by adapting to the evolving capabilities of the fine-tuned model based on the insight that constraints should only intervene when degenerate outputs occur. We implement this by using a reference model as an \textit{online refiner} that takes the response from the fine-tuned model and generates a minimally corrected version which preserves correct content verbatim while fixing errors. A supervised fine-tuning loss then trains the fine-tuned model to produce the refined output. This mechanism yields a constraint that automatically strengthens or relaxes based on output quality. Experiments on dialogue and code generation show that dynamic constraints outperform both KL regularization and unconstrained baselines, achieving substantially higher task rewards while maintaining training stability.

📄 PDF Abstract BibTeX arXiv:2603.18088

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningCode Generation

Similar Papers 제목 키워드 기반

AdaRefiner: Refining Decisions of Language Models with Adaptive Feedback

2023-09-29 · Wanpeng Zhang, Zongqing Lu

Large Language Models (LLMs) have demonstrated significant success across various domains. However, their application in complex decision-making tasks frequently necessitates intricate prompt engineering or fine-tuning, …

Common Sense ReasoningDecision MakingLanguage ModellingPrompt Engineering+2

VG-Refiner: Towards Tool-Refined Referring Grounded Reasoning via Agentic Reinforcement Learning

2025-12-06 · Yuji Wang, Wenlong Liu, Jingxuan Niu, Haoji Zhang 외 arxiv

Tool-integrated visual reasoning (TiVR) has demonstrated great potential in enhancing multimodal problem-solving. However, existing TiVR paradigms mainly focus on integrating various visual tools through reinforcement le…

Reinforcement LearningVisual Reasoning

One Refiner to Unlock Them All: Inference-Time Reasoning Elicitation via Reinforcement Query Refinement

2026-04-28 · Yixiao Zhou, Dongzhou Cheng, zhiliang wu, Yi Yang 외 arxiv

Large Language Models (LLMs) often fail to utilize their latent reasoning capabilities due to a distributional mismatch between ambiguous human inquiries and the structured logic required for machine activation. Existing…

Reinforcement Learning

AutoRefiner: Improving Autoregressive Video Diffusion Models via Reflective Refinement Over the Stochastic Sampling Path

2025-12-12 · Zhengyang Yu, Akio Hayakawa, Masato Ishii, Qingtao Yu 외 arxiv

Autoregressive video diffusion models (AR-VDMs) show strong promise as scalable alternatives to bidirectional VDMs, enabling real-time and interactive applications. Yet there remains room for improvement in their sample …

Reinforcement Learning-Driven Plant-Wide Refinery Planning Using Model Decomposition

2025-04-11 · Zhouchang Li, Runze Lin, Hongye Su, Lei Xie

In the era of smart manufacturing and Industry 4.0, the refining industry is evolving towards large-scale integration and flexible production systems. In response to these new demands, this paper presents a novel optimiz…

Computational EfficiencyDeep Reinforcement Learningreinforcement-learningReinforcement Learning