Revision History
Muon (optimizer) · 7 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected Moonlight's architecture (27 layers, hidden size 2,048, 64 routed experts), the description of the update rule and of how Distributed Muon shards its work, a mischaracterization of earlier orthogonalization work by Carlson et al., a misquoted claim that the inexact orthogonalization is beneficial, and the outdated claim that Muon is not in torch.optim (it shipped in PyTorch 2.9 in October 2025) (plus 24 minor fixes)
Version 7 · Nov 5, 2026, 05:32 AM
- Improve internal links (2026-10-13)
Version 6 · Oct 13, 2026, 10:05 AM
- Added 6 contextual internal links
Version 5 · Jul 23, 2026, 03:34 PM
- Render formulas with LaTeX math notation
Version 4 · Jul 11, 2026, 10:31 PM
- Initial AI-authored creation (4194 words).
Version 2 · May 20, 2026, 02:14 PM