jalr/tabbyAPI-ollama
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Requirements: Update Flash Attention 2

Browse source 
Use 2.3.4 from tgw. However, keep the 2.3.3 wheels in requirements
if the newer wheels don't work for now.

Signed-off-by: kingbri <bdashore3@proton.me>
This commit is contained in:
kingbri 2023-11-21 22:12:55 -05:00
parent 71b9a53336
commit d25310e55d
1 changed files with 18 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

22
requirements.txt
View file

@ -6,7 +6,21 @@ sse_starlette
uvicorn
# Wheels
https://github.com/bdashore3/flash-attention/releases/download/2.3.3/flash_attn-2.3.3+cu122-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
https://github.com/bdashore3/flash-attention/releases/download/2.3.3/flash_attn-2.3.3+cu122-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.3/flash_attn-2.3.3+cu122torch2.1cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.3/flash_attn-2.3.3+cu122torch2.1cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
# Flash Attention 2. If the wheels don't work, comment these out, uncomment the old wheels, and run `pip install -r requirements.txt`
# Windows FA2 from https://github.com/jllllll/flash-attention/releases
https://github.com/jllllll/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu121torch2.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
https://github.com/jllllll/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu121torch2.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
# Linux FA2 from https://github.com/Dao-AILab/flash-attention/releases
https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu122torch2.1cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.4/flash_attn-2.3.4+cu122torch2.1cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
# Old flash attention 2 wheels. Uncomment these and if the new ones don't work!
# IMPORTANT: Make sure to re-comment the new wheels before re-running `pip install -r requirements.txt`
# https://github.com/bdashore3/flash-attention/releases/download/2.3.3/flash_attn-2.3.3+cu122-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
# https://github.com/bdashore3/flash-attention/releases/download/2.3.3/flash_attn-2.3.3+cu122-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
# https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.3/flash_attn-2.3.3+cu122torch2.1cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
# https://github.com/Dao-AILab/flash-attention/releases/download/v2.3.3/flash_attn-2.3.3+cu122torch2.1cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"