Metraipolis › archive › dc_6a04bb8fc802

Knowing When Not to Reuse: Conditional Experience Transfer in Autonomous LLM Post-Training — arXiv:2608.26730

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.26730
License (as archived)
CC0-1.0
Archived
2026-10-17T14:22:53.983Z
SHA-256 of the archived text
78aaeadf174595ede21359d6b337f0ac8ba3c104da41f3bfd545ffef01420a37
Stable URL of this page
https://metraipolis.com/documento/dc_6a04bb8fc802

How to cite

Knowing When Not to Reuse: Conditional Experience Transfer in Autonomous LLM Post-Training — arXiv:2608.26730 — arXiv (en), https://arxiv.org/abs/2608.26730 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6a04bb8fc802 (Metraipolis archive dc_6a04bb8fc802, sha256 78aaeadf174595ede21359d6b337f0ac8ba3c104da41f3bfd545ffef01420a37)
BibTeX
@misc{dc6a04bb8fc802,
  title = {Knowing When Not to Reuse: Conditional Experience Transfer in Autonomous LLM Post-Training — arXiv:2608.26730},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.26730},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6a04bb8fc802; sha256 78aaeadf174595ede21359d6b337f0ac8ba3c104da41f3bfd545ffef01420a37},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.26730. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models offer broad capabilities, but adapting them to evolving domains, tools, and requirements often entails repeated post-training. Autonomous systems automate parts of this process by proposing updates, training candidates, and using evaluation feedback to select subsequent proposals. As evidence accumulates, a central problem emerges: which past update evidence remains actionable after subsequent training has changed the parent model? An update's effect depends on its parent, data, and training stage. Treating past success as context-free permission can waste compute. If the resulting child is promoted, it can also degrade the subsequent training trajectory. We formulate this problem as conditional experience transfer and introduce Boundary-Calibrated Intervention Transfer (BCIT), a method that authorizes experience reuse before weight-changing training. BCIT binds an observed effect to its source context, checks applicability conditions, vetoes candidates with named hard conflicts, and obtains current-state evidence through a bounded training trial when needed. Fully trained candidates still face a shared adoption rule, and only observed events extend memory. On one 4B model adapted across finance reasoning, text-to-SQL, and function calling, candidate updates exhibit heterogeneous target and retention effects across the evaluated contexts. Under matched candidates, evidence, and compute, BCIT authorizes fewer harmful updates and attains higher equal-budget final-model quality than the evaluated alternatives. These results support treating experience authorization as a distinct problem in autonomous post-training.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.26730 — 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 2026-10-17T13:16:53.512Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).