训练领域专家模型的无推理轨迹蒸馏方法

中文摘要 AI 辅助

专家蒸馏通过教师生成的推理轨迹，有效地将领域专业知识迁移给学生模型。然而，当这些专家模型仅基于问答对进行训练，而没有显式的推理监督时，是什么决定了它们生成的轨迹？在本工作中，我们表明专家优化隐式地从这一潜在轨迹空间中进行选择。为了隔离并观察这一潜在分布，我们将学生蒸馏并非作为下游目标，而是作为一种不可知探针——因为学生模型从专家那里不继承任何参数化或优化约束，仅继承采样得到的轨迹本身。通过这一探针，我们的实证分析揭示了一种紧密的支配关系：在27组专家-学生配对中，它们的专业化-泛化特征曲线具有极强的相关性。至关重要的是，显式控制专家的分布漂移，会系统地推动教师及其蒸馏学生沿着领域精度与通用能力保留之间的可控权衡移动。在化学、物理和多语言设置中，蒸馏学生系统地反映了这些专家诱导的特征曲线，即使跨越不同的模型家族也是如此。我们的发现为专家训练建立了新的视角：当缺乏黄金推理时，调优选择直接控制传递给下游模型的潜在监督。

英文摘要

Specialist distillation effectively transfers domain expertise to student models via teacher-generated reasoning trajectories. However, when these specialists are trained solely on question--answer pairs without explicit reasoning supervision, what governs the trajectories they generate? In this work, we show that specialist optimization implicitly selects from this latent trajectory space. To isolate and observe this latent distribution, we leverage student distillation not as a downstream goal, but as an agnostic probe---since students inherit no parameterization or optimization constraints from the specialist, inheriting only the sampled trajectories themselves. Through this probe, our empirical analysis unveils a tight governing relationship: across 27 specialist--student pairings, their specialization--generalization profiles correlate exceptionally strongly. Crucially, explicitly controlling the specialist's distributional drift systematically shifts both the teacher and its distilled student along a controllable trade-off between domain precision and general-capability retention. Across chemistry, physics, and multilingual settings, distilled students systematically reflect these specialist-induced profiles, even across divergent model families. Our findings establish a new view of specialist training: when gold reasoning is absent, tuning choices directly control the latent supervision passed to downstream models.

发表机构

• University of British Columbia（不列颠哥伦比亚大学）
• University of Helsinki（赫尔辛基大学）
• ELLIS Institute Finland（芬兰ELLIS研究所）
• University of Turku（图尔库大学）
• Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）

机构由 AI 辅助整理，请以论文原文为准。