Mirrors
/
EveryDream2trainer
mirror of https://github.com/victorchall/EveryDream2trainer.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

update install doc for local linux

This commit is contained in:
Victor Hall 2023-06-14 15:04:35 -04:00
parent b69793623b
commit 7b6ec89441
1 changed files with 15 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
doc/SETUP.md
View File

@ -63,3 +63,18 @@ Or go to your browser and hit `http://localhost:8888`. The web password is
`test1234` but you can change that in `docker-compose.yml`.
Your current source directory will be moutned to the Jupyter notebook.
## Local Linux install
* Make sure Linux Nvidia driver is up to date and working.
Check that `nvidia-smi` is working and shows your GPU.
How to update your driver may depend on the Linux distribution you use. For Ubuntu, use Gnome and open `Softwrae & Updates`, go to the `additional drivers` tab and select `Using NVIDIA driver metapackage from nvidia-driver-530 (proprietary)`. Currently `530` is the latest version, but you can use latest at your time of install.
*You will need to use the proprietary driver.*
* Install Cuda 11.8. You can use https://developer.nvidia.com/cuda-11-8-0-download-archive?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=22.04&target_type=deb_local for Ubuntu 22.04 for instance. Your install may vary depending on Linux distribution
* `git clone https://github.com/victorchall/EveryDream2trainer`
* `cd EveryDream2trainer`
* `python3 -m venv venv`
* `source venv/bin/activate`
* `pip install -r requirements.txt`
If you are savvy you can configure your own conda environment as well using roughly the above but using a conda env instead of venv.