CodeLLaMA Installation - Windows & WSL2

gistfile1.txt

#
# I have successfully run this with Anaconda on Windows directly & on WSL2/Ubuntu 22.04 with MiniConda
#

# Make sure you have Anaconda installed
# This tutorial assumes you have an Nvidia GPU, but you can find the non-GPU version on the Textgen WebUI github
# More information found here: https://github.com/oobabooga/text-generation-webui

# Make sure you have the CUDA 11.8 & cuDNN 11.x installed
# CUDA Toolkit 11.8: https://developer.nvidia.com/cuda-18-7-0-download-archive
# cuDNN 11.x: https://developer.nvidia.com/rdp/cudnn-download

# Create a conda environment
conda create -n tg python=3.10.9
conda activate tg

# Install pytorch with GPU support
conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

# Install text-generation-webui
git clone https://github.com/oobabooga/text-generation-webui
cd text-generation-webui
python -m pip install -r requirements.txt

# check to make sure CUDA is enabled
python -m torch.utils.collect_env

# To test in the interpreter
$ python
>>> import torch
>>> torch.cuda.is_available()
>>> torch.zeros(1).cuda()

# Start a local server, then click on the 'Running on local URL: http://127.0.0.1:xxxx'
python server.py

# After the page comes up, click on Model in the top nav
# Go get this Model: https://huggingface.co/TheBloke/WizardCoder-Python-7B-V1.0-GPTQ