@article{howtoscalemixtureofexpertsfrommuptothema,
  title = {How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization},
  author = {Leena Chennuru Vankadara and Moritz Haas and Luke Hayward and Sebastian Bordt and Alessandro Breccia},
  year = {2026},
  eprint = {2605.14200},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.14200},
}