Prebuild flash attention wheels for linux!

[kunibald413]

thank you lord!

show the guy some love:
https://github.com/mjun0812/flash-attention-prebuild-wheels

[eric-haibin-lin]

Thanks! But i got the following error when using prebuilt ones:

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
Cell In[39], line 1
----> 1 import flash_attn

File [/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/flash_attn/__init__.py:3](https://jupyterlab-01je1y37fd9z67ns6q1yqncqk0.studio.lightning.ai/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/flash_attn/__init__.py#line=2)
      1 __version__ = "2.6.3"
----> 3 from flash_attn.flash_attn_interface import (
      4     flash_attn_func,
      5     flash_attn_kvpacked_func,
      6     flash_attn_qkvpacked_func,
      7     flash_attn_varlen_func,
      8     flash_attn_varlen_kvpacked_func,
      9     flash_attn_varlen_qkvpacked_func,
     10     flash_attn_with_kvcache,
     11 )

File [/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py:10](https://jupyterlab-01je1y37fd9z67ns6q1yqncqk0.studio.lightning.ai/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py#line=9)
      6 import torch.nn as nn
      8 # isort: off
      9 # We need to import the CUDA kernels after importing torch
---> 10 import flash_attn_2_cuda as flash_attn_cuda
     12 # isort: on
     14 def maybe_contiguous(x):

ImportError: [/usr/lib/x86_64-linux-gnu/libc.so.6](https://jupyterlab-01je1y37fd9z67ns6q1yqncqk0.studio.lightning.ai/usr/lib/x86_64-linux-gnu/libc.so.6): version `GLIBC_2.32' not found (required by [/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/flash_attn_2_cuda.cpython-310-x86_64-linux-gnu.so](https://jupyterlab-01je1y37fd9z67ns6q1yqncqk0.studio.lightning.ai/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/flash_attn_2_cuda.cpython-310-x86_64-linux-gnu.so))