This repository creates the s9pk
package that is installed to run GPT4All
on embassyOS.
The process of building the package is the same as for pretty much any other start9 wrapper repository. If you have the dependencies installed, just run make
.
The process for version-bumping oobabooga or a dependency is a little complicated. My advice is to:
- launch the
start9/ai-base
docker container with /data mounted to some persistent volume - in the container's
/data
directory, clone the desired revision of text-generation-webui - checkout the same revision in the text-generation-webui submodule
- in the container, run
pip install -r requirements.txt
- run the server and check that everything works
- in the container, run
pip freeze
- copy the results into
requirements-versions.txt
and delete unneeded dependencies such astorch
andnvidia-*