Metraipolis › archive › dc_799cc8543fd8
Adaptive-Policy Distillation: Knowledge Injection under Information Asymmetry — arXiv:2611.00852
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00852
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:09:48.855Z
- SHA-256 of the archived text
fd3226dbf3d9c7e6f8aa0f74ed3b005699153502a0cc2871c36ebee3aa725128- Stable URL of this page
https://metraipolis.com/documento/dc_799cc8543fd8
How to cite
Adaptive-Policy Distillation: Knowledge Injection under Information Asymmetry — arXiv:2611.00852 — arXiv (en), https://arxiv.org/abs/2611.00852 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_799cc8543fd8 (Metraipolis archive dc_799cc8543fd8, sha256 fd3226dbf3d9c7e6f8aa0f74ed3b005699153502a0cc2871c36ebee3aa725128)
BibTeX
@misc{dc799cc8543fd8,
title = {Adaptive-Policy Distillation: Knowledge Injection under Information Asymmetry — arXiv:2611.00852},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00852},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_799cc8543fd8; sha256 fd3226dbf3d9c7e6f8aa0f74ed3b005699153502a0cc2871c36ebee3aa725128},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00852. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
On-policy distillation (OPD), which uses reverse-KL supervision on student rollouts to transfer a context-conditioned model's behavior into student model parameters for closed-book inference, has shown strong results. However, we find that this success does not translate to knowledge injection: across two document-QA benchmarks, OPD injects document knowledge far less efficiently than simple off-policy distillation. We trace this reversal to two failures that are structural, yet sparse across token positions. Reverse KL weights its gradient by the student distribution and therefore barely lifts document-specific tokens the student considers unlikely; after forward KL restores this signal, student rollouts expose later positions on the target trajectory only after generating the preceding tokens correctly, so coverage deteriorates with target length. Motivated by the concentration of teacher-student divergence at these hard positions, we propose Adaptive-Policy Distillation (APD), whose per-token gate jointly shifts the rollout toward the teacher and the objective toward forward KL where needed, withdrawing both interventions as the student learns. Without labels or a verifier, APD improves over OPD by up to 54.6%, achieving the best scores of 0.558 F1 on QASPER and 0.486 MRR on BioASQ. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.00852 — 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 2026-11-03T10:59:52.276Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).