Giters

deepseek-ai / DeepSeek-MoE

DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

关于flash_attn

GXKIM opened this issue · comments

commented

torch 2.1
transformers:4.37.1
显卡:A800
手动安装装进去了，但是检测不到flash_attn
直接用pip 装不进去
去github flash_attn 是支持A800的

image
commented

这个是transformers的检测机制问题，可以在源代码目录试试pip install .