hf-scripts / exl2-windows-local /instructions.txt
Anthonyg5005's picture
publish finished auto exl2 script
9bcb78e
raw
history blame
1.09 kB
install the CUDA toolkit
Nvidia Maxwell or higher
https://developer.nvidia.com/cuda-downloads?target_os=Windows&target_arch=x86_64
Nvidia Kepler or higher
https://developer.nvidia.com/cuda-11-8-0-download-archive?target_os=Windows&target_arch=x86_64
Haven't done much testing but Visual Studio with desktop development for C++ might be required. I've gotten cl.exe errors on a previous install
make sure you setup the environment by using windows-setup.bat
after everything is done just download a model using download-model.bat
to quant, use convert-model-auto.bat. Enter the model's folder name, then the BPW for the model
You can always pause the quantization process by pressing Ctrl + C and typing exit. All progress will be stored in the WD (working directory) folder. You can resume where you left off by running the convert-model-auto.bat script with the same arguments you used before.
Credit to turboderp for creating exllamav2 and the exl2 quantization method.
https://github.com/turboderp
Credit to oobabooga the original download script.
https://github.com/oobabooga