1
Solving Moe Load Imbalance in LLM Training via Optimal Transport
用最优传输理论动态放置专家副本,从拓扑感知角度根治MoE训练负载失衡,值得算法工程师细读。
Article URL: https://arxiv.org/abs/2608.03676 Comments URL: https://news.ycombinator.com/item?id=49330030 Points: 1 # Comments: 0