Metraipolis › archive › dc_aed2bf54e24b

Discovered, Not Designed: Population Evolution for Collaborative and Compute-Intensive Model Discovery — arXiv:2610.05950

Source: arXiv
Original: https://arxiv.org/abs/2610.05950
License (as archived): CC0-1.0
Archived: 2026-10-06T05:31:22.047Z
SHA-256 of the archived text: e0b7bc8f9308454a663caf500a1c9e4d3d3c63e51c9eba94501c455170fd62bd
Stable URL of this page: https://metraipolis.com/documento/dc_aed2bf54e24b

How to cite

Discovered, Not Designed: Population Evolution for Collaborative and Compute-Intensive Model Discovery — arXiv:2610.05950 — arXiv, https://arxiv.org/abs/2610.05950 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_aed2bf54e24b (Metraipolis archive dc_aed2bf54e24b, sha256 e0b7bc8f9308454a663caf500a1c9e4d3d3c63e51c9eba94501c455170fd62bd)

BibTeX

@misc{dcaed2bf54e24b,
  title = {Discovered, Not Designed: Population Evolution for Collaborative and Compute-Intensive Model Discovery — arXiv:2610.05950},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.05950},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_aed2bf54e24b; sha256 e0b7bc8f9308454a663caf500a1c9e4d3d3c63e51c9eba94501c455170fd62bd},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05950. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM-driven evolution enables iterative model development, but two practical goals remain underexplored: finding model designs that transfer across related tasks and sustaining improvement when training is expensive. We introduce Population Evolution (PE), a collaborative, hierarchical framework that connects ongoing local searches through shared experimental evidence. PE evaluates code changes across related training instances and shares the results to guide subsequent proposals and promotion to larger training scales. For expensive targets, PE searches small training subsets and screens candidates through peer and intermediate evaluations before full-target training. We introduce RMD-Bench to evaluate both settings across ranking, watch-time prediction, RL algorithm discovery, and LLM/VLM pretraining. Compared with standalone evolution at matched source iterations, PE raises mean best local gains from 7.01% to 8.97% in ranking and from 2.84% to 3.85% in watch-time, while improving the best larger-scale outcome in all three joint-discovery families. In watch-time discovery, PE improves best larger-scale gains with four of five harnesses and all four proposers. On new recommendation datasets under shared target-side calibration, every evaluated PE design improves over the reference in mean performance. Under matched total GPU compute, completed LLM discovery runs yield a best relative accuracy gain of 2.48% and 13 successful candidates for PE, versus 0.92% and none for direct evolution. VLM loss reduction reaches 8.78% versus 5.05% under matched total GPU compute.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.05950 — 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 2026-10-06T05:27:05.841Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).