Metraipolis › archive › dc_cbd4fb474e40
Beyond Policy Discrepancy: Capability-Aware Multi-Teacher On-Policy Distillation — arXiv:2610.26950
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.26950
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-21T07:07:20.737Z
- SHA-256 of the archived text
82ab71f6853d97e8647d0e7d8c27159df55589218afb17d42451c8750a364277- Stable URL of this page
https://metraipolis.com/documento/dc_cbd4fb474e40
How to cite
Beyond Policy Discrepancy: Capability-Aware Multi-Teacher On-Policy Distillation — arXiv:2610.26950 — arXiv (en), https://arxiv.org/abs/2610.26950 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cbd4fb474e40 (Metraipolis archive dc_cbd4fb474e40, sha256 82ab71f6853d97e8647d0e7d8c27159df55589218afb17d42451c8750a364277)
BibTeX
@misc{dccbd4fb474e40,
title = {Beyond Policy Discrepancy: Capability-Aware Multi-Teacher On-Policy Distillation — arXiv:2610.26950},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.26950},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cbd4fb474e40; sha256 82ab71f6853d97e8647d0e7d8c27159df55589218afb17d42451c8750a364277},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.26950. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multi-teacher on-policy distillation (MOPD) integrates domain specialists into a shared student, but determining how strongly each teacher should supervise remains challenging. Existing discrepancy-based allocation can assign strong supervision even when the student already matches or exceeds the teacher's task performance, since behavioral mismatch does not necessarily imply a remaining capability deficit. We propose Capability-Aware Multi-Teacher On-Policy Distillation (CA-MOPD), which combines specialist merging with capability-guided allocation to account for what the student has already learned. CA-MOPD first merges reinforcement-learning specialists into a strong student initialization, then uses verifier-estimated teacher--student capability gaps to adjust domain-level distillation strength. Supervision decreases as capability gaps close and increases when estimated student capability regresses, while dense token-level guidance remains provided by the teachers. Across six benchmarks spanning mathematics, code generation, and instruction following, CA-MOPD achieves an average score of 32.22, outperforming naive MOPD, Open-MOPD, and the merged initialization by 0.86, 0.83, and 0.74 points, respectively. Controlled comparisons further support the benefits of merged initialization and capability-gap guidance under matched training budgets. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.26950 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-21T07:05:43.511Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).