Metraipolis › archive › dc_5293cf86509d
Speculative Self-Distillation enables Efficient Knowledge Internalization — arXiv:2611.12016
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.12016
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-10T18:43:51.197Z
- SHA-256 of the archived text
f6e14079e3d162392a43ec10362ecc0df45a411849c5626ef2248ede2c8871ef- Stable URL of this page
https://metraipolis.com/documento/dc_5293cf86509d
How to cite
Speculative Self-Distillation enables Efficient Knowledge Internalization — arXiv:2611.12016 — arXiv (en), https://arxiv.org/abs/2611.12016 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5293cf86509d (Metraipolis archive dc_5293cf86509d, sha256 f6e14079e3d162392a43ec10362ecc0df45a411849c5626ef2248ede2c8871ef)
BibTeX
@misc{dc5293cf86509d,
title = {Speculative Self-Distillation enables Efficient Knowledge Internalization — arXiv:2611.12016},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.12016},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5293cf86509d; sha256 f6e14079e3d162392a43ec10362ecc0df45a411849c5626ef2248ede2c8871ef},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.12016. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Many real-world applications require language models to internalize knowledge that was absent from pretraining, to incorporate continually updated facts, or to learn new domain skills. Self-distillation has emerged as a natural approach for this task. A model conditioned on the source material acts as the teacher, and an unconditioned student is trained to reproduce the teacher's behavior closed-book. We propose Speculative Self-Distillation (SSD), an efficient per-token mixed-policy method for self-distillation. The student generates each rollout by default, while the teacher intervenes only at positions where its next-token distribution substantially diverges from the student's. Unlike off-policy distillation with teacher-generated prefixes, this design keeps training rollouts close to the student's test-time behavior, while avoiding the low-signal states that fully on-policy rollouts drift into. In effect, SSD lets the teacher steer the rollout exactly where the privileged context changes the prediction. On three closed-book QA benchmarks measuring fact recall, knowledge updates, and skill acquisition, SSD nearly matches on-policy distillation while using 38-69% fewer supervised tokens and consistently outperforms off-policy distillation on the held-out test set. On a forgetting stress test, SSD retains the most general capability while also achieving the best in-domain accuracy and using less supervision than the on-policy baseline. Overall, SSD establishes a new accuracy-efficiency-forgetting Pareto frontier for self-distillation. 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); Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2611.12016 — 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 2026-11-10T18:37:37.672Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).