Mixture-of-experts routing

ReLibra

ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning

On the frontierfirst seen May 9, 2026

Current frontier — recent, not yet superseded in the knowledge base

0 papers critique it · 0 beat it on benchmarks

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.