Metraipolis › archive › dc_e478cf5bcdfb

The Evolution Strategies Paradox: Better Models from Worse Perturbations — arXiv:2611.04113

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.04113
License (as archived): CC0-1.0
Archived: 2026-11-04T12:27:41.058Z
SHA-256 of the archived text: 45cf0629e0af98e62596d4b5438af912ae4738b732b641373e69f8abd9dfaa29
Stable URL of this page: https://metraipolis.com/documento/dc_e478cf5bcdfb

How to cite

The Evolution Strategies Paradox: Better Models from Worse Perturbations — arXiv:2611.04113 — arXiv (en), https://arxiv.org/abs/2611.04113 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e478cf5bcdfb (Metraipolis archive dc_e478cf5bcdfb, sha256 45cf0629e0af98e62596d4b5438af912ae4738b732b641373e69f8abd9dfaa29)

BibTeX

@misc{dce478cf5bcdfb,
  title = {The Evolution Strategies Paradox: Better Models from Worse Perturbations — arXiv:2611.04113},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.04113},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e478cf5bcdfb; sha256 45cf0629e0af98e62596d4b5438af912ae4738b732b641373e69f8abd9dfaa29},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04113. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

How is it possible for perturbation-based methods like Evolution Strategies (ES) to optimize billion-parameter large language models (LLMs)? Classical search methods are highly dependent on finding good candidates within generated populations. However, when using ES to fine-tune LLMs, a phenomenon can occur where all offspring models consistently perform worse than the parent model, yet during training, the parent model consistently improves. This paradox occurs across models, tasks, and even modalities. This paper characterizes the paradox, then examines how the local reward landscape and properties of the ES estimator contribute to it: in high-dimensional spaces, exploration leads to worse performance, yet still highlights good directions; aggregating those directions then allows for making local improvements. The result can be seen as a new optimization paradigm where a principled interaction of noise and curvature produces the paradox as a byproduct of effective optimization. These findings provide guidance for taking advantage of perturbation-based methods in LLM fine-tuning and in high-dimensional optimization in general.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.04113 — 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 2026-11-04T12:26:47.127Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).