推理效率对比问题

PKU-YuanGroup / MoE-LLaVA

Mixture-of-Experts for Large Vision-Language Models

https://arxiv.org/abs/2401.15947

Apache License 2.0

1.9k stars 121 forks source link

推理效率对比问题 #47

Open aprilehannibal opened 6 months ago

aprilehannibal commented 6 months ago

Describe the issue

MoE的一个优势是推理效率，但是在paper里没看到和同size的VLM做对比，请问具体情况如何？

LinB203 commented 6 months ago

We don't have that intention while the MoE implementation needs to be driven under Deepspeed engine, which could lead to unfair comparisons. We look forward to having more community institutes working on this.