Muon massively beats Adam on tiny superhuman agents across games, drones, driving
jsuarez · x · 2026-08-29
Researcher jsuarez clarifies their work is not on LLMs: they train tiny superhuman agents across games, drones, driving, and commerce sims. Under a full hyperparameter sweep for both optimizers, Muon is a massive improvement over Adam on these tasks. The thread context is a hypothesis about mode-covering in cross-entropy pretraining vs RL exploration.
Related event: Muon Optimizer Beats Adam on Small Superhuman Agents(2 posts)→
More from Research
- Learn Positional Encodings derivation from first principles — zainhas · 2026-08-30
- COLM Paper Traces Capability Provenance in LLMs via Gradient Attribution — ziv_ravid · 2026-08-30
- Toby Ord paper argues recursive self-improvement has physical limits — Exponential View (Azeem Azhar) · 2026-08-30
- AI Formalization Tools Fable and Sol Spot First Repairable Error in Published Literature — Sauers_ · 2026-08-30
- Mark Schmidt Posts ICML Tutorial Video: Is Numerical Optimization Theory Irrelevant to ML Practice in 2026? — MarkSchmidtUBC · 2026-08-30
- SDF Donut in 46 Lines of Python — voooooogel · 2026-08-30