Metraipolis › archive › dc_15992be8bb28

Initialization Improves LLM-Driven Discovery — arXiv:2610.00707

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.00707
License (as archived)
CC0-1.0
Archived
2026-10-06T12:10:50.010Z
SHA-256 of the archived text
68c49f663c14b254d7742afc61cf50b1703f589ebd6ef31a84715643be493e08
Stable URL of this page
https://metraipolis.com/documento/dc_15992be8bb28

How to cite

Initialization Improves LLM-Driven Discovery — arXiv:2610.00707 — arXiv (en), https://arxiv.org/abs/2610.00707 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_15992be8bb28 (Metraipolis archive dc_15992be8bb28, sha256 68c49f663c14b254d7742afc61cf50b1703f589ebd6ef31a84715643be493e08)
BibTeX
@misc{dc15992be8bb28,
  title = {Initialization Improves LLM-Driven Discovery — arXiv:2610.00707},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.00707},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_15992be8bb28; sha256 68c49f663c14b254d7742afc61cf50b1703f589ebd6ef31a84715643be493e08},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.00707. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large Language Models (LLMs) have been used for novel discovery of algorithms, theorems, drugs, and other tasks through the use of harnesses that prompt an LLM to iteratively optimize an objective. In this work, we study the relationship between the population of previous iterates and eventual discovery success. We generalize past work on harness design to develop a suite of 12 harnesses called 'Modular' and characterize their performance across 5 diverse discovery tasks, finding that discovery success is brittle and sensitive to harness design. We uncover mode collapse, characterized by a dramatic drop in the diversity of iterates, as a common failure mode. We find that popular state-of-the-art harnesses and diversity-inducing harness interventions, which aim to prolong this collapse, yield inconsistent gains. Our results instead uncover that the performance of early discoveries is predictive of eventual success. We therefore propose a universally applicable intervention that performs an initial stage of parallel exploration in order to initialize subsequent iterative optimization. Our method provides consistent gains across many harnesses and target applications, confirming the importance of initialization in LLM-driven discovery.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-02 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.00707 — 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 2026-10-06T10:21:28.103Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).