Search results

[RFC] MuonTR: Muon variant for deterministic Token-Routed architectures

Status: Open.

#65 In KellerJordan/Muon;

· Complexity-ML opened on Apr 2, 2026

ImportError: cannot import name 'MuonWithAuxAdam' from 'muon'

Status: Open.

#61 In KellerJordan/Muon;

· shiwanghua opened on Dec 23, 2025

Supporting QK-Clip (MuonClip) as an additional input

Status: Open.

#59 In KellerJordan/Muon;

· axion66 opened on Nov 27, 2025

Does Muon need clip_grad?

Status: Open.

#56 In KellerJordan/Muon;

· OBJECT907 opened on Nov 4, 2025

optimizer applies weight decay to parameters even when their gradient is None

Status: Open.

#54 In KellerJordan/Muon;

· Muinez opened on Nov 1, 2025

Muon momentum update appears to deviate from the paper; seems to be using a running average

Status: Open.

#52 In KellerJordan/Muon;

· RobotSail opened on Sep 24, 2025

The performernce of Muon optimizer worse than AdamW ?

Status: Open.

#51 In KellerJordan/Muon;

· chenzixiong808-max opened on Sep 19, 2025

Muon for multi-head attention weights

Status: Open.

#50 In KellerJordan/Muon;

· schmidt-ai opened on Sep 16, 2025

Learning rate compared to adamw

Status: Open.

#49 In KellerJordan/Muon;

· davnords opened on Sep 1, 2025

Muon optimizer shows no GPU memory reduction with Accelerator mixed precision (bf16)

Status: Open.

#48 In KellerJordan/Muon;

· gitouni opened on Aug 27, 2025

Does Muon save the state["momentum_buffer"] correctly in a distributed setting?

Status: Open.

#46 In KellerJordan/Muon;

· EIFY opened on Aug 21, 2025

Does Muon support FSDP Zero3

Status: Open.

#45 In KellerJordan/Muon;

· lucaswychan opened on Aug 4, 2025