Paper Proposes Compute-Efficient Hyperparameter Transfer for MoE
kakaocorp · hf · 2026-08-24
This paper introduces a two-step hyperparameter transfer framework designed to predict optimal learning rates for large-scale Mixture-of-Experts models. By scaling across model widths and token budgets, it enables efficient pretraining without the need for costly sweeps.
More from Infra
- Copper to optical: optical backplanes seen as necessary as racks move to optics — BenBajarin · 2026-10-16
- Polymarket prices 28% odds a US state enacts data center moratorium by year-end — Polymarket · 2026-10-16
- Pennsylvania residents reject $10,000-per-household offer for massive Amazon data center — Polymarket · 2026-10-16
- PyTorch now runs natively on AWS Trainium with zero code changes — PyTorch · 2026-10-16
- Chinese labs run on ~1/10 US compute; ByteDance pays 2-4x to hoover up SEA supply — robleclerc · 2026-10-16
- Zymtrace raises $12.2M and hires ML engineers to build agents that fix GPU bottlenecks — cfregly · 2026-10-16