![]() Use the python -m and call pip commands to install all mentioned packagesĬreate a requirements.txt file and add all previously mentioned packages to it.Īlso add the following to the requirements.txt file:Īnd add any additional performance enhancing packages for stable diffusion in addition to graphics card performance packages compatible with Windows 11 and stable diffusion.Īdd an install requirements.txt command to the. Install torch for cuda 11.8 using " whatever specific command you want and update all of torch and cuda's associated dependencies inside of a virtual environment named venv. ![]() bat to activate a virtual environment called venv in the same directory. If you're writing a bat to simplify everything make sure you add a pause line after call venv/scripts/activate and do python where to verify python.exe is running in the local directory(aka the venv is active).Īsk chat gpt, your new personal assistant, to do this all for you Always do this in a virtual environment specific to each project so as not to ransack your system ![]() I usually base any software level on my torch version and bring everything else up to that unless the software I'm running requires something else. Upgrade numba and clean-fid that should bring everything up to the same version of torch if I'm not mistaken. Since your version of numpy is higher than, it ultimately means upgrading remg and so on. If torch two is your primary goal you're going to have to bring everything else up to that. From this you're going to have to build a tree priorities.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |