Improve documentation

[Kucharssim]

A lot of the user-facing code has still some documentation missing. Based on a discussion with @paul-buerkner, I am listing here what I see missing just so that we have an overview of it and can tackle it as we go.

Missing docs important for users

• DeepSet.__init__
• DeepSet.__call__
• CouplingFlow.__init__
• FlowMatching.__init__
• ContinuousApproximator.__init__
• ContinuousApproximator.sample
• ContinuousApproximator.log_prob
• ModelComparisonApproximator.__init__
• ModelComparisonApproximator.predict
• Approximator.fit
• Adapter - everything?
• BasicWorkflow
  - .fit_offline, .fit_disk, .fit_online
  - .simulate, .simulate_adapted
  - .sample
  - .log_prob
  - .plot_diagnostics
  - .compute_diagnostics

[paul-buerkner]

Great, thank you! Everyone, whenever you have a few minutes, consider documenting something from the above list :-)

[vpratz]

When adding docs, please adhere to the numpydoc styleguide (Example). This ensures that the docs will be rendered correctly in the end.
You can also check this yourself by locally building the docs. Starting in the root directory of the repository, run (inside your virtual environment):

pip install -e .[docs] # installs necessary dependencies
cd docsrc
make local  # build docs from your local state and copies them to ../docs, currently takes about 45 seconds)
make view-docs  # starts a Python webserver to view the website (browsing via file:// protocol can cause problems)

If you want faster build times, temporarily comment out some modules in docsrc/source/api/bayesflow.rst. This will reduce the number of pages that is built. You can always rebuild using make local, if something goes wrong run make clean or make clean-all to go back to a clean state.

[vpratz]

I think we got most of it covered (please proof-read and improve if you find the time), thanks @stefanradev93!
The ModelComparisonApproximator is still missing, but I'm not sufficiently familiar with the code to write good docstrings. @Kucharssim or @LarsKue, could you provide the docstrings for that class?

[Kucharssim]

Awesome!

Yes, I will do model comparison

[vpratz]

Should we close this issue, or keep it open for other doc improvement related requests?

[paul-buerkner]

Let's close it. If new things come up, we can open an issue for that or just fix it.