Presentation Information

[B-16-07]A Study on Latency-Aware Expert Selection for MoE Inference

〇Hiroto Iwai1, Keito Yokoyama1, Kazuma Aoyama1, Shota Inoue1, Hiroyuki Ohsaki1 (1. Kwansei Gakuin Univ.)

Keywords:

LLM,MoE,expert,routing

Password required to view