arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.07712cs.LG
收藏

面向优化器的稳健缩放定律

Towards Robust Scaling Laws for Optimizers

Institute of Science and Technology Austria (ISTA)(科学与技术奥地利研究所);Lancaster University(兰卡斯特大学)

Alexandra Volkova, Mher Safaryan, Christoph H. Lampert, Dan Alistarh

AI总结:

本文研究了不同优化器的稳健缩放定律,提出了一种共享幂律指数和优化器特定重缩放因子的更稳健定律,并通过理论分析证明了Chinchilla式缩放定律的自然产生。

摘要:

大型语言模型(LLM)预训练的质量取决于多个因素,包括计算预算和优化算法的选择。经验性缩放定律被广泛用于预测随着模型大小和训练数据增长时的损失,然而,几乎所有现有研究都固定了优化器(通常为AdamW)。同时,新一代优化器(如Muon、Shampoo、SOAP)承诺更快且更稳定的收敛,但它们与模型和数据缩放的关系尚不明确。在本工作中,我们研究了不同优化器的缩放定律。实证上,我们表明1)为每个优化器单独制定Chinchilla式缩放定律是病态的,且具有高度相关的参数。相反,2)我们提出了一种更稳健的定律,具有共享的幂律指数和优化器特定的重缩放因子,这使得不同优化器之间能够直接比较。最后,3)我们为代理任务(凸二次目标)的梯度方法提供了理论分析,证明Chinchilla式缩放定律自然地出现在损失分解为不可约、近似和优化误差的过程中。

英文摘要:

The quality of Large Language Model (LLM) pretraining depends on multiple factors, including the compute budget and the choice of optimization algorithm. Empirical scaling laws are widely used to predict loss as model size and training data grow, however, almost all existing studies fix the optimizer (typically AdamW). At the same time, a new generation of optimizers (e.g., Muon, Shampoo, SOAP) promises faster and more stable convergence, but their relationship with model and data scaling is not yet well understood. In this work, we study scaling laws across different optimizers. Empirically, we show that 1) separate Chinchilla-style scaling laws for each optimizer are ill-conditioned and have highly correlated parameters. Instead, 2) we propose a more robust law with shared power-law exponents and optimizer-specific rescaling factors, which enable direct comparison between optimizers. Finally, 3) we provide a theoretical analysis of gradient-based methods for the proxy task of a convex quadratic objective, demonstrating that Chinchilla-style scaling laws emerge naturally as a result of loss decomposition into irreducible, approximation, and optimization errors.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑