Metraipolis › archive › dc_4dbe2737fa5b

Can We Trust the Teacher? Decoupled Credit Direction-Magnitude for Self-Distillation — arXiv:2609.34848

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34848
License (as archived)
CC0-1.0
Archived
2026-10-07T20:58:14.622Z
SHA-256 of the archived text
5463a6c7a01392d8b9e01f6e69c16ef8027d688324600056dad04c4444f126a6
Stable URL of this page
https://metraipolis.com/documento/dc_4dbe2737fa5b

How to cite

Can We Trust the Teacher? Decoupled Credit Direction-Magnitude for Self-Distillation — arXiv:2609.34848 — arXiv (en), https://arxiv.org/abs/2609.34848 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4dbe2737fa5b (Metraipolis archive dc_4dbe2737fa5b, sha256 5463a6c7a01392d8b9e01f6e69c16ef8027d688324600056dad04c4444f126a6)
BibTeX
@misc{dc4dbe2737fa5b,
  title = {Can We Trust the Teacher? Decoupled Credit Direction-Magnitude for Self-Distillation — arXiv:2609.34848},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34848},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4dbe2737fa5b; sha256 5463a6c7a01392d8b9e01f6e69c16ef8027d688324600056dad04c4444f126a6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34848. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

RLVR provides reliable trajectory-level credit, while OPSD offers dense supervision for token-level credit. This exposes a fundamental coupling when updating step-level credit direction and magnitude with teacher supervision, preventing steps from receiving reliable credit directions and contribution magnitudes, while making both vulnerable to teacher judgment errors and preference variance, as supported by our theoretical analysis. To separate credit direction from its contribution magnitude, we introduce \textit{Decoupled Credit Self-Distillation (DCSD)}, which theoretically decouples credit direction and magnitude into two reliable signals and uses them to calibrate privileged teacher supervision. Specifically, we design belief-margin probing to determine credit direction and marginal information gain to quantify credit magnitude, enabling step-to-token credit assignment for policy optimization. Across 11 benchmarks, DCSD achieves the best overall scores against GRPO, OPSD, RLSD, and RLCSD. Compared with base models, DCSD improves the overall score by 8.45 points on mathematical reasoning and 7.01 points on multimodal reasoning, while correcting the credit direction for 6\% of tokens and yielding a 1.5$\times$ reduction in token credit magnitude.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.34848 — 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 2026-10-07T20:29:01.686Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).