Metraipolis › archive › dc_9d36aaa2a66c
Improving Data-Constrained Scaling with Hyperparameter Transfer and Ensembling — arXiv:2610.37842
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.37842
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-01T02:23:16.634Z
- SHA-256 of the archived text
108821890dfe8a27d24613914f29f15daf87b27b7d7fbc1ff931879de4649efa- Stable URL of this page
https://metraipolis.com/documento/dc_9d36aaa2a66c
How to cite
Improving Data-Constrained Scaling with Hyperparameter Transfer and Ensembling — arXiv:2610.37842 — arXiv (en), https://arxiv.org/abs/2610.37842 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9d36aaa2a66c (Metraipolis archive dc_9d36aaa2a66c, sha256 108821890dfe8a27d24613914f29f15daf87b27b7d7fbc1ff931879de4649efa)
BibTeX
@misc{dc9d36aaa2a66c,
title = {Improving Data-Constrained Scaling with Hyperparameter Transfer and Ensembling — arXiv:2610.37842},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.37842},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9d36aaa2a66c; sha256 108821890dfe8a27d24613914f29f15daf87b27b7d7fbc1ff931879de4649efa},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.37842. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
As growing compute budgets outpace the growth of high-quality training data, large language models must be trained with both compute and data constraints in mind. Unfortunately, existing prescriptions for data-constrained scaling suffer from two important limitations. First, past works hold key hyperparameters such as weight decay fixed with scale, despite the fact that optimal hyperparameters vary wildly with model size and number of epochs. Second, past works neglect ensembling, a promising axis for scaling under data constraints. As a surprising consequence, prior prescriptions can lead to loss that increases with compute budget. We address both limitations by fitting power laws that prescribe how to transfer hyperparameters with scale and scaling laws that prescribe how to allocate compute with ensembling. Training on 100M tokens from FineWeb, our recipe reduces validation loss by over 0.15 nats compared to prior recipes, a 14% decrease in perplexity, and achieves monotonic improvements with compute through ensembling. These gains are achieved using prescriptions fit to smaller runs and validated at up to 12$\times$ larger compute budgets. 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/2610.37842 — 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 2026-11-01T02:12:44.084Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).