stanfordmlgroup / ngboost

Natural Gradient Boosting for Probabilistic Prediction
Apache License 2.0
1.64k stars 215 forks source link

Build Fisher reparametrisation documentation for GH Pages #198

Closed mzjp2 closed 3 years ago

mzjp2 commented 3 years ago

Followed @alejandroschuler's steps to build the docs, moved @cs2716's changes from docs/content/5-dev.ipynb to examples/user-guide/content/5-dev.ipynb and then built the docs properly.

I think a much smoother way to do this would be to get the GH Pages GH action to do the building, if possible - so that users can only edit the notebook and leave it be. Something worth exploring.

The diff is large because of a couple of minor changes, perhaps because my jupyter-book version was0.6.5but the original book was built with0.6.5dev0` (which no longer exists)

mzjp2 commented 3 years ago

Docs all look fine locally:

image
alejandroschuler commented 3 years ago

@mzjp2 yes I would love it if GH pages could do the build. iirc, though, there are restrictions on what they support that made local builds the only option. I think this is especially the case now that jupyterbook has moved to a sphinx backend instead of jekyll. If you figure out a way around that I would love to implement it.

mzjp2 commented 3 years ago

@alejandroschuler I think if we move to a model where the docs are served from a gh-pages branch rather than a docs/ directory on master we could use https://jupyterbook.org/publish/gh-pages.html#automatically-host-your-book-with-github-actions