rixty/KoboldAI-Client
1
1
Fork 0
mirror of https://github.com/KoboldAI/KoboldAI-Client.git synced 2025-06-05 21:59:24 +02:00
Code Issues Projects Releases Wiki Activity

Automatic installation of the quant_cuda module during install_requirements

Browse Source 
Kepler (K40+) and Maxwell support
This commit is contained in:
0cc4m
2023-04-10 22:37:16 +02:00
parent 7efd314428
commit b628aec719
3 changed files with 8 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
install_requirements.bat
View File

@@ -49,6 +49,8 @@ umamba.exe install --no-shortcuts -r B:\python\ -n base -f "%~dp0\environments\h
umamba.exe -r B:\ clean -a -y umamba.exe -r B:\ clean -a -y
rd B:\Python\pkgs /S /Q rd B:\Python\pkgs /S /Q
subst B: /d subst B: /d
call B:\python\condabin\activate
cmd /k "pip install https://github.com/0cc4m/GPTQ-for-LLaMa/releases/download/2023-04-10/quant_cuda-0.0.0-cp38-cp38-win_amd64.whl"
pause pause
exit exit
@@ -60,5 +62,7 @@ umamba.exe create -r miniconda3\ -n base
umamba.exe install --no-shortcuts -r miniconda3 -n base -f environments\huggingface.yml -y --always-copy umamba.exe install --no-shortcuts -r miniconda3 -n base -f environments\huggingface.yml -y --always-copy
umamba.exe clean -a -y umamba.exe clean -a -y
rd miniconda3\Python\pkgs /S /Q rd miniconda3\Python\pkgs /S /Q
call miniconda3\condabin\activate
cmd /k "pip install https://github.com/0cc4m/GPTQ-for-LLaMa/releases/download/2023-04-10/quant_cuda-0.0.0-cp38-cp38-win_amd64.whl"
pause pause
exit exit

3
install_requirements.sh
View File

@@ -5,6 +5,9 @@ wget -qO- https://micromamba.snakepit.net/api/micromamba/linux-64/latest | tar -
bin/micromamba create -f environments/huggingface.yml -r runtime -n koboldai -y bin/micromamba create -f environments/huggingface.yml -r runtime -n koboldai -y
# Weird micromamba bug causes it to fail the first time, running it twice just to be safe, the second time is much faster # Weird micromamba bug causes it to fail the first time, running it twice just to be safe, the second time is much faster
bin/micromamba create -f environments/huggingface.yml -r runtime -n koboldai -y bin/micromamba create -f environments/huggingface.yml -r runtime -n koboldai -y
# Install quant_cuda module for 4-bit
bin/micromamba run -r runtime -n koboldai pip install https://github.com/0cc4m/GPTQ-for-LLaMa/releases/download/2023-04-10/quant_cuda-0.0.0-cp38-cp38-linux_x86_64.whl
exit exit
fi fi
if [[ $1 = "rocm" ]]; then if [[ $1 = "rocm" ]]; then

2
repos/gptq

Submodule repos/gptq updated: 17c46a59ff...50b22e2ba8