Metraipolis › archive › dc_d30a062bd996

Self-Evolving Skills via Surrogate-Guided Solve-and-Reproduce — arXiv:2608.28638

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.28638
License (as archived)
CC0-1.0
Archived
2026-10-17T07:41:45.768Z
SHA-256 of the archived text
144734746b0c6de471ab08b4d4ae54af840f01ddb750362ddc03b9682fdb0c4a
Stable URL of this page
https://metraipolis.com/documento/dc_d30a062bd996

How to cite

Self-Evolving Skills via Surrogate-Guided Solve-and-Reproduce — arXiv:2608.28638 — arXiv (en), https://arxiv.org/abs/2608.28638 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d30a062bd996 (Metraipolis archive dc_d30a062bd996, sha256 144734746b0c6de471ab08b4d4ae54af840f01ddb750362ddc03b9682fdb0c4a)
BibTeX
@misc{dcd30a062bd996,
  title = {Self-Evolving Skills via Surrogate-Guided Solve-and-Reproduce — arXiv:2608.28638},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.28638},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d30a062bd996; sha256 144734746b0c6de471ab08b4d4ae54af840f01ddb750362ddc03b9682fdb0c4a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.28638. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Agent skills are portable packages of instructions and resources an agent consults at deployment. Self-evolving them fails in two ways today. First, skills evolved from scratch underperform human-curated ones and, on a weak model, using no skill at all. Second, an evolution-time pass records one lucky trajectory that a fresh stochastic agent often fails to reproduce at deployment. We present reSolve, a per-task, oracle-in-the-loop framework built on three components. It decouples interactive solving from a self-contained deliverable that is independently re-executed in a fresh container, a protocol we call solve-and-reproduce. It enhances the sparse reward signal with a surrogate verifier that cannot access hidden tests or reference answers. It then runs verifier-guided beam search over a solution-construction graph. Within a fixed harness, a cheap model self-evolves skills that reach $74.9\%$ mean-of-3, $+14.8$ points over the $60.1\%$ human-curated baseline, exceeding the strongest official curated-skill result ($67.3\%$, GPT-5.5/OpenHands). We also report observed failure cases and domain-level results, including performance on the 14 Natural Science tasks, to clarify when the approach does and does not help.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.28638 — 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 2026-10-17T07:28:26.464Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).