🛰️ Daily AI Frontier
‹ back to 2026-07-27

Hyperball May Not Be a Free Lunch

Research Efficiency & Systems

Ranking

Overall 77
Content 80
Popularity 71

Observed public metrics from 1 member.

Merged summary

TL;DR - This paper argues that Hyperball-style optimizers derive much of their behavior from effective step-size evolution rather than inherently better update directions. Their performance still depends critically on learning-rate scheduling.

  • Introduces an angular effective learning rate incorporating parameter-update angle, parameter norm, and update norm.
  • Finds radial update components have limited direct impact on angular displacement under the tested settings.
  • Schedule-matching experiments attribute differences between MuonH and MuonWD mainly to effective step-size dynamics.
  • More aggressive decay improves MuonH early but can hurt later pretraining performance.

Sources (1)

Hyperball May Not Be a Free Lunch

arXiv cs.LG Yihao Xiao, Jialong Sun, Zitian Gao, Zeming Wei, Chutian Wang, Ran Tao, Jiaye Teng, Bryan Dai 2026-07-24 arXiv:2607.22444
Public signals Semantic Scholar citations 1 · Semantic Scholar influential citations 0
Providers: Hugging Face · N/A OpenAlex · N/A Publisher · N/A Semantic Scholar · Citations 1 · Influential citations 0 X · N/A Fetched 2026-08-26 14:45:07.327197 UTC

TL;DR - This paper argues that Hyperball-style optimizers derive much of their behavior from effective step-size evolution rather than inherently better update directions. Their performance still depends critically on learning-rate scheduling.

  • Introduces an angular effective learning rate incorporating parameter-update angle, parameter norm, and update norm.
  • Finds radial update components have limited direct impact on angular displacement under the tested settings.
  • Schedule-matching experiments attribute differences between MuonH and MuonWD mainly to effective step-size dynamics.
  • More aggressive decay improves MuonH early but can hurt later pretraining performance.
item →