Metraipolis › archive › dc_765cc2174cde

Pretraining Data Scale Reverses the Muon-AdamW Fine-Tuning Gap — arXiv:2611.03633

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.03633
License (as archived): CC0-1.0
Archived: 2026-11-03T11:23:26.108Z
SHA-256 of the archived text: db1b6eb6508977c08784240211ca88328a1d25e5ec69fda571709c9256fc4204
Stable URL of this page: https://metraipolis.com/documento/dc_765cc2174cde

How to cite

Pretraining Data Scale Reverses the Muon-AdamW Fine-Tuning Gap — arXiv:2611.03633 — arXiv (en), https://arxiv.org/abs/2611.03633 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_765cc2174cde (Metraipolis archive dc_765cc2174cde, sha256 db1b6eb6508977c08784240211ca88328a1d25e5ec69fda571709c9256fc4204)

BibTeX

@misc{dc765cc2174cde,
  title = {Pretraining Data Scale Reverses the Muon-AdamW Fine-Tuning Gap — arXiv:2611.03633},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.03633},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_765cc2174cde; sha256 db1b6eb6508977c08784240211ca88328a1d25e5ec69fda571709c9256fc4204},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.03633. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Muon has recently emerged as a promising optimizer for pretraining large language models, offering a speedup over AdamW in minimizing pretraining loss. However, it remains unclear whether Muon's advantage persists after fine-tuning. In this paper, we study the downstream behavior of Muon and AdamW and uncover a surprising trend: beyond a sufficiently large pretraining data budget, AdamW outperforms Muon after fine-tuning. Moreover, as model size increases, AdamW overtakes Muon at progressively smaller pretraining token-parameter ratios. To explain this reversal, we view fine-tuning as a perturbation of the outputs and introduce the margin-to-spread ratio, which measures how far the correct logit lies above the wrong ones relative to how widely the wrong logits are spread. In language models, Muon has the larger margin-to-spread ratio at small data budgets and AdamW at large ones, mirroring the fine-tuning trend. In a linear associative memory, we prove that after one update the margin-to-spread ratio of gradient descent gains on that of Muon as the data budget grows, and simulations show that this trend persists after extended training.

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/2611.03633 — 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 2026-11-03T11:02:37.734Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).