← 研踪 PaperTrailarXiv:2611.23673 · v1
Communicative On-Policy Distillation: Distilling Black-Box Teacher Capabilities through Communication for Long-Horizon Tasks
首次提交 2026-11-18 08:56
版本更新 2026-11-18 08:56
雷达观测 2026-11-19 19:08
摘要
On-policy distillation (OPD) is a powerful paradigm for improving language-model agents on long-horizon tasks, yet its conventional formulation relies on teacher logits that text-only black- box models do not expose. To address this limitation, we introduce Communicative On-Policy Distillation (COPD), which uses natural-language communication as a medium for distilling a stronger model's capabilities into a smaller student. During task execution, the student au- tonomously decides when to consult a black-box teacher and formulates open-ended natural- language requests without predefined question templates or intervention schedules. Successful student-teacher interactions are retained as privileged references to supervise fresh adviser-free student rollouts through on-policy self-distillation, jointly with task-reward reinforcement learning. On CAMG-II, COPD achieves 2.46 times the gain of the strongest black-box OPD baseline. The capabilities it learns through communication generalize to Terminal-Bench 2.1 and SWE-Bench Pro V2, improving task success without adviser access.
作者备注:13 pages, 8 figures, 2 tables. Project page: https://liruiluo.github.io/communicative-opsd/
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 08:56 · 最新观测版本
Communicative On-Policy Distillation: Distilling Black-Box Teacher Capabilities through Communication for Long-Horizon Tasks
On-policy distillation (OPD) is a powerful paradigm for improving language-model agents on long-horizon tasks, yet its conventional formulation relies on teacher logits that text-only black- box models do not expose. To address this limitation, we introduce Communicative On-Policy Distillation (COPD), which uses natural-language communication as a medium for distilling a stronger model's capabilities into a smaller student. During task execution, the student au- tonomously decides when to consult a black-box teacher and formulates open-ended natural- language requests without predefined question templates or intervention schedules. Successful student-teacher interactions are retained as privileged references to supervise fresh adviser-free student rollouts through on-policy self-distillation, jointly with task-reward reinforcement learning. On CAMG-II, COPD achieves 2.46 times the gain of the strongest black-box OPD baseline. The capabilities it learns through communication generalize to Terminal-Bench 2.1 and SWE-Bench Pro V2, improving task success without adviser access.
查看此版本解读 →查看此版本原文 ↗