【flash-attention】Building wheel for flash-attn (pyproject.toml) did not run successfully

报错

Building wheel for flash-attn (pyproject.toml) did not run successfully

解决

方法1

git clone [email protected]:Dao-AILab/flash-attention.git
cd /flash-attention
python setup.py install

注意这里会从出现错误提示flash-attention/csrc/cutlass找不到，git下载cutlass失败
所以cd flash-attention/csrc/ 然后 [email protected]:NVIDIA/cutlass.git

重新运行python setup.py install 就可以编译成功了

方法2(推荐)

找到自己对应的配置版本，例如:
cuda：12.2
torch：2.2
python：3.10

pip install https://github.com/Dao-AILab/flash-attention/releases/download/v2.4.2/flash_attn-2.4.2+cu122torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl