Metraipolis › archive › dc_fe90fd1d3cfc
OPRD: On-Policy Reminder Distillation for Long-Horizon Coding Agents — arXiv:2610.21496
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.21496
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T10:01:28.199Z
- SHA-256 of the archived text
ab71920a9e0eb354cc0ec4318e1c14ef9fb799303ab2368eba5a0d6dfc2bac8d- Stable URL of this page
https://metraipolis.com/documento/dc_fe90fd1d3cfc
How to cite
OPRD: On-Policy Reminder Distillation for Long-Horizon Coding Agents — arXiv:2610.21496 — arXiv (en), https://arxiv.org/abs/2610.21496 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fe90fd1d3cfc (Metraipolis archive dc_fe90fd1d3cfc, sha256 ab71920a9e0eb354cc0ec4318e1c14ef9fb799303ab2368eba5a0d6dfc2bac8d)
BibTeX
@misc{dcfe90fd1d3cfc,
title = {OPRD: On-Policy Reminder Distillation for Long-Horizon Coding Agents — arXiv:2610.21496},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.21496},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fe90fd1d3cfc; sha256 ab71920a9e0eb354cc0ec4318e1c14ef9fb799303ab2368eba5a0d6dfc2bac8d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21496. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reliable repository-level coding requires agents not only to generate correct code but also to preserve constraints and respond to tool feedback across long interactions. However, our diagnostic analysis shows that agents often fail to act on relevant information already available in context, while final Pass/Fail outcomes provide only sparse process supervision. We introduce On-Policy Reminder Distillation (OPRD), which combines Multi-Scope Reminder supervision with outcome-based routing on reminder-free student rollouts. Successful rollouts provide autoregressive self-imitation targets, whereas failed rollouts receive Reminder-conditioned reverse-KL supervision from a teacher that re-evaluates the same student-generated prefixes. A Global Reminder converts final-test diagnostics into trajectory-level guidance, while Turn-local Reminders apply only to the next decision following the relevant environment observation. The teacher and student share a two-stage initialization from Coding SFT and Instruction-Following (IF) SFT. At inference, the student operates without Reminders. On SWE-bench Verified, SWE-bench Pro, and SWE-bench Multilingual, OPRD improves the benchmark-averaged task resolution rate by 11.3 points over the open-source Qwen3.5-4B base model. Under the same diagnostic protocol, it reduces the aggregate error-event count from 446 to 222, a 50.2% reduction. 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/2610.21496 — 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 2026-10-16T08:44:47.657Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).