Metraipolis › archive › dc_fde1259402c8
EvoGuard: Mitigating Degradation in Online Skill Evolution — arXiv:2611.06444
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.06444
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-05T18:18:48.328Z
- SHA-256 of the archived text
a6c30eb47202f71eb2467c3b886b028483ca8201914b9495bea52d8f20418624- Stable URL of this page
https://metraipolis.com/documento/dc_fde1259402c8
How to cite
EvoGuard: Mitigating Degradation in Online Skill Evolution — arXiv:2611.06444 — arXiv (en), https://arxiv.org/abs/2611.06444 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fde1259402c8 (Metraipolis archive dc_fde1259402c8, sha256 a6c30eb47202f71eb2467c3b886b028483ca8201914b9495bea52d8f20418624)
BibTeX
@misc{dcfde1259402c8,
title = {EvoGuard: Mitigating Degradation in Online Skill Evolution — arXiv:2611.06444},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.06444},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fde1259402c8; sha256 a6c30eb47202f71eb2467c3b886b028483ca8201914b9495bea52d8f20418624},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.06444. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Skills let language agents reuse experience across tasks as natural-language documents. Recent work evolves skills from interaction, but existing methods have limitations when skills evolve online. Prior methods update a skill offline from a batch of trajectories collected over many tasks. Some then adopt the new skill without any check, which can degrade the skill over time. Others keep it only if it improves performance on a held-out validation set, which is too slow and costly to run after every update. We introduce EvoGuard, a pairwise skill selector that mitigates skill degradation at low latency and cost. We first run the agent iteratively to build pools of skills and score each skill on a private test set to obtain utility labels. We pair the skills and use rejection sampling to keep only pairs with a clear utility gap. The selector is trained on the logits of two label tokens to pick the better skill. At deployment it decides in a single forward pass, which is significantly faster than evaluation on the private test set. Compared with adopting every new skill, EvoGuard improves success by 11.0% on ALFWorld, 5.9% on ScienceWorld, and 2.7% on LiveCodeBench. 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/2611.06444 — 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 2026-11-05T18:13:03.946Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).