OctoMed：面向最优多模态医学推理的数据配方
OctoMed: Data Recipes for State-of-the-Art Multimodal Medical Reasoning
对做医学 VLM 的人，这篇给出了可复用的 SFT 数据配方：用长度多样的结构化推理链做监督微调，模型就能在下游任务上自适应地决定推理长度，无需额外监督；可直接迁移到自建医学多模态推理模型的训练数据构造与配比设计上。
原文依据 curating a high-quality, diverse training dataset with varying structured reasoning trace lengths enables the fine-tuned model to self-calibrate its reasoning trajectory lengths
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论高质量且经过精心筛选的数据是训练医学大语言模型的基石，因为它直接影响模型对未见临床任务的泛化能力与鲁棒性。我们研究了训练与数据筛选策略，以在医学领域构建一个鲁棒的多模态推理模型。我们的工作聚焦于监督微调（SFT），并探索了利用结构化推理轨迹的数据配方。使用我们提出的数据配方，我们将实验规模扩展到一个包含超过 800 万个样本、68 亿个回复 token 的数据集，在多样化的分布外医学基准任务上取得了开源模型中的最优性能。结果进一步表明，筛选出一个高质量、多样化且结构化推理轨迹长度各异的训练数据集，能使微调后的模型在无显式监督的情况下，根据下游任务自行校准其推理轨迹长度。我们给出了关键洞见，描述了数据筛选策略，并概述了构建鲁棒医学视觉语言推理系统的后续步骤。
英文原文摘要 1103 字符
High-quality and carefully curated data is a cornerstone of training medical large language models, as it directly impacts both generalization and robustness to unseen clinical tasks. We investigate strategies for training and data curation to develop a robust multimodal reasoning model in the medical domain. Our work focuses on supervised fine-tuning (SFT) and explores data recipes that leverage structured reasoning traces. Using our proposed data recipe, we scale experiments to a dataset of over 8 million examples and 6.8 billion response tokens, achieving state-of-the-art performance among open-source models across diverse out-of-distribution medical benchmark tasks. Our results further indicate that curating a high-quality, diverse training dataset with varying structured reasoning trace lengths enables the fine-tuned model to self-calibrate its reasoning trajectory lengths based on the downstream task, without explicit supervision. We present key insights, describe the data curation strategy, and outline next steps toward developing robust medical vision-language reasoning systems.