Metraipolis › archive › dc_a294c783a280

Accelerating LLM Pretraining with Zeroth-Order Curvature on Idle Serving Compute — arXiv:2611.02688

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02688
License (as archived)
CC0-1.0
Archived
2026-11-03T11:08:20.267Z
SHA-256 of the archived text
735c51667cc732624d179a245100f9cf06003535d2875bb5960da4b00824c8bf
Stable URL of this page
https://metraipolis.com/documento/dc_a294c783a280

How to cite

Accelerating LLM Pretraining with Zeroth-Order Curvature on Idle Serving Compute — arXiv:2611.02688 — arXiv (en), https://arxiv.org/abs/2611.02688 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a294c783a280 (Metraipolis archive dc_a294c783a280, sha256 735c51667cc732624d179a245100f9cf06003535d2875bb5960da4b00824c8bf)
BibTeX
@misc{dca294c783a280,
  title = {Accelerating LLM Pretraining with Zeroth-Order Curvature on Idle Serving Compute — arXiv:2611.02688},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02688},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a294c783a280; sha256 735c51667cc732624d179a245100f9cf06003535d2875bb5960da4b00824c8bf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02688. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM pretraining is increasingly compute-intensive, while substantial inference serving compute remains idle during off-peak periods. However, such serving hardware is difficult to use for conventional first-order training due to its memory, networking and other system requirements. Zeroth-order (ZO) optimization offers a natural interface to this forward-only compute resource, but its gradient estimates break down in pretraining. Unlike fine-tuning, which can confine updates to a small parameter subspace, pretraining must update every parameter, and the alignment of a ZO gradient estimate with the true gradient shrinks as the number of updated parameters grows. We make a different observation: noisy ZO estimates can still provide useful curvature information, even when they are too inaccurate to serve as gradients. Based on this insight, we propose ZO-Curv, which leverages idle forward-only devices to estimate curvature and uses it to adapt and refine the update used during first-order training, thereby accelerating the process. We provide convergence guarantees that characterize the effect of finite-query curvature estimation. Experiments on LLMs from 125M to 1.5B parameters show that ZO-Curv consistently accelerates pretraining. At 770M parameters, ZO-Curv reaches AdamW's 20K-step validation loss in 3.9 hours instead of 7.5 hours, while achieving a lower final loss of 2.70 versus 2.84 under the same step budget. These results show that idle model serving compute can accelerate LLM pretraining when zeroth-order queries are used to estimate curvature rather than gradients directly.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.02688 — 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 2026-11-03T10:59:46.119Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).