跳到正文
arXiv cs.CL· Prakhar Mishra, Bhavya Yajush Awasthi, Tanuja, Nishant·· 10 小时前AI 评分24

自奖励强化学习让小语言模型的医学解释结构更清晰

Self-Rewarded Reinforcement Learning for Better-Structured Medical Explanations in Small Language Models

AI 导读

研究团队在指令微调的 Mistral-7B 上,用 LoRA 适配的 PPO 对 2000 条 Meerkat 衍生医学提示词做约 170 步强化学习训练,奖励仅基于自监督信号(偏好更长、词汇多样、标点规范的回复,惩罚过短回复),单张 NVIDIA A100 即可完成。

正文 · 原文

View PDF HTML (experimental)

Abstract:Small language models (SLMs) can run on commodity hardware and support privacy-preserving, on-premises clinical deployment, but their medical explanations are often short and poorly structured compared with large language models. Recent supervised approaches such as Meerkat transfer chain-of-thought reasoning into 7B-8B backbones, yet their training objective does not directly target the style or organization of the generated rationales. Building on the instruction-tuned Mistral-7B backbone, we investigate whether a lightweight reinforcement learning (RL) step can improve explanation structure without human preference labels. We apply Proximal Policy Optimization (PPO) with LoRA adaptation to 2000 Meerkat-derived medical prompts, guided by a simple self-supervised reward that favors longer, lexically diverse, and well-punctuated responses while penalizing very short ones. Training for about 170 PPO steps on a single NVIDIA A100 GPU yields a stable increase in the average reward. Quantitative and qualitative comparisons show that the RL-tuned model produces longer, more organized, multi-point explanations while preserving the underlying medical content. We discuss the promise and the limitations of self-rewarded RL for shaping the style of medical explanations.
Comments: 14 Pages, 10 Figures
Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Cite as: arXiv:2611.24650 [cs.AI]
  (or arXiv:2611.24650v2 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.24650

arXiv-issued DOI via DataCite

Submission history

From: Bhavya Yajush Awasthi [view email]
[v1] Wed, 18 Nov 2026 15:52:24 UTC (48,594 KB)
[v2] Thu, 19 Nov 2026 05:26:05 UTC (48,590 KB)

来源:arXiv cs.CL · arxiv.org