Installing a trained pipeline from a release in a private repo

[elisno]

I've set up several spaCy projects in private repositories and figured that I could track all of the trained artifacts by trying to set up my own release repository (similar to explosion/spacy-models, but private).

The idea is that I could do:

export OWNER=...
export RELEASE_REPO=...
export MODEL=...

pip install https://github.com/$OWNER/$RELEASE_REPO/releases/download/$MODEL/$MODEL.tar.gz

similarly to what is described in the "Installation via pip" section in the docs

As is, the pip install statement only works if the RELEASE_REPO is public, but I ask if anyone has any experience with doing this for a private repository?

[polm]

I haven't done this, but I assume you could download the .tar.gz with wget by specifying Github credentials somehow? If you do that then you can just pip install the local file.