GitHub
plugyawn/Megaprop
Newton-Muon + Preconditioned Optimizers for MoE Training at scale, with out-of-the-box support for MuP and FSDP support for Muon, built on top of Megatron-LM and TransformerEngine.
- Stars
- 35
- Forks
- 2
- Stars gained in 24 hours
- 0
- Forks gained in 24 hours
- 0
Topics: adam, adamw, amd, cuda, gpt, large-language-models, muon, optimization, optimizers, parallelism
Latest data point: Nov 21, 2026 · 35 stars · 2 forks