Metraipolis › archive › dc_25097b29a910

Learning from Repaired Reasoning: Root-Cause-Guided On-Policy Distillation — arXiv:2610.03515

Source
arXiv
Original
https://arxiv.org/abs/2610.03515
License (as archived)
CC0-1.0
Archived
2026-10-05T22:51:48.502Z
SHA-256 of the archived text
bef3fbbcd40444d8cc6c6a1c3db6ee44b98c0a7eba052aa8e00e17dfe8789597
Stable URL of this page
https://metraipolis.com/documento/dc_25097b29a910

How to cite

Learning from Repaired Reasoning: Root-Cause-Guided On-Policy Distillation — arXiv:2610.03515 — arXiv, https://arxiv.org/abs/2610.03515 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_25097b29a910 (Metraipolis archive dc_25097b29a910, sha256 bef3fbbcd40444d8cc6c6a1c3db6ee44b98c0a7eba052aa8e00e17dfe8789597)
BibTeX
@misc{dc25097b29a910,
  title = {Learning from Repaired Reasoning: Root-Cause-Guided On-Policy Distillation — arXiv:2610.03515},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.03515},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_25097b29a910; sha256 bef3fbbcd40444d8cc6c6a1c3db6ee44b98c0a7eba052aa8e00e17dfe8789597},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.03515. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

On-policy self-distillation (OPSD) uses reference solutions as privileged hindsight to supervise student-generated reasoning trajectories. However, reference-based guidance may explain a correct solution without addressing why the student's own reasoning fails. This reasoning mismatch between the guidance provided and the correction needed can encourage the student to borrow correct conclusions while leaving its reasoning errors unresolved. Moreover, applying the same hindsight throughout the trajectory risks a distillation trap, where unnecessary constraints on valid reasoning compete with correction of substantive errors. To address these issues, we propose Root-Cause-Guided On-Policy Distillation (RC-OPD), which uses repairs of the student's own reasoning to provide guidance that addresses its specific errors while building on valid progress. For each failed attempt, RC-OPD locates the earliest substantive error, develops a local correction, and uses the corrected intermediate result as an anchor for the valid prefix. An iterative diagnosis--repair--continuation process tests the repairs through student continuation, identifying further errors within a fixed repair budget. For repair chains that reach a correct answer, root--cause--guided distillation uses failure diagnoses and corrective goals to supervise the erroneous segments, while anchor-guided distillation supports the corresponding valid prefixes with reasoning chains leading to the repaired intermediate results. We evaluate RC-OPD across multiple datasets and model scales. Extensive experiments and analyses show that it mitigates reasoning mismatch and the distillation trap, yielding substantial performance gains.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.03515 — 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 2026-10-05T22:46:10.588Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).