arXiv cs.AI· Yuanda Xu, Hejian Sang, Zhengze Zhou, Shayan Mohajer Hamidi, Xiaomin Li, Rohit Jain, Alborz Geramifard·· 12 小时前AI 评分29
Δ-MOPD：多教师在线策略蒸馏中传递教师减基线的 logit 偏移
Composing What Each Teacher Learned: Delta Distillation for Multi-Teacher On-Policy Learning
AI 导读
论文提出 Δ-MOPD，将每位教师的 teacher-minus-base logit 偏移重新锚定在学生冻结初始化上，替代多教师在线策略蒸馏（MOPD）中常用的端点策略监督。
正文 · 原文
Abstract:Multi-teacher on-policy distillation (MOPD) is used in two settings. In common-domain composition, several teachers score each student rollout from one prompt domain and their signals form a single target; in routed-domain distillation, prompts from different domains are assigned to the corresponding specialist. Both settings usually transfer each teacher's endpoint policy, which mixes what post-training changed with preferences inherited from the teacher's base. We introduce $\Delta$-MOPD, which transfers each teacher's teacher-minus-base logit shift re-anchored at the student's frozen initialization, and compare it with endpoint supervision in both settings while holding teacher selection fixed. We first expose the mechanism that impedes endpoint transfer: inherited base pull can exceed the post-training shift. Removing it reduces the teacher-term norm ratio and target--student KL.
Across our experiments, the results suggest that shift targets are particularly useful when teacher signals are combined at a state. With three composed teachers, $\Delta$-MOPD exceeds endpoint composition by $4.11$ Math and $1.95$ five-benchmark points; with two, it matches endpoint accuracy. Under phased routing, it achieves higher mean performance in both phase orders and reduces the observed order gap from $10.50$ to $6.42$ points. Under interleaved routing, where each update involves one teacher, the two targets perform comparably. The phased results provide supporting evidence that the benefit may extend to signals accumulated across training phases. Target construction is thus an independent design axis in MOPD, complementary to teacher selection. The experiment repo could be find at this https URL
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.10460 [cs.LG] |
| (or arXiv:2610.10460v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.10460 arXiv-issued DOI via DataCite |
Submission history
From: Yuanda Xu [view email]
[v1]
Wed, 7 Oct 2026 17:27:46 UTC (83 KB)
[v2]
Tue, 20 Oct 2026 16:51:01 UTC (84 KB)
来源：arXiv cs.AI · arxiv.org