Revision 5e599fb01df65d156a40f7a138ab6627a06a50db authored by Eric Hammy on 07 May 2020, 15:01:50 UTC, committed by GitHub on 07 May 2020, 15:01:50 UTC
* add long_description and project_urls to (#1438)
* add type hints for probability distributions (#1421)
* remove comment (#1441)
* Use a type alias in the fnction signature of leading_transpose (#1442)
* refactor natgrads to be more efficient (#1443)
* Fix dimensions of kernel evaluation of changepoint kernel (#1446)
* Removed unusued imports. (#1450)
* Improve representation of GPflow objects in IPython/Jupyter notebook (#1453)
* includes the repr() string in IPython/Jupyter notebook representation as well (i.e. fully-qualified class name and object hash (memory address), which helps distinguish objects from each other)
* only displays the parameter table when it is not empty
* makes use of default_summary_fmt() for IPython shell
* Convert data structures to tensor in model init method (#1452)
* Use a boolean for full covariance in sample_mvn. (#1448)
* #1452 for GPMC model (#1458)
* release candidate v2.0.2 (#1457)

Co-authored-by: st-- <>
Co-authored-by: joelberkeley-pio <>
Co-authored-by: John Mcleod <>
Co-authored-by: Mark van der Wilk <>
Co-authored-by: Artem Artemev <>
1 parent 2358539
Raw File
# GPflow documentation

## Read documentation online

[![Documentation Status (master)](](
[![Documentation Status (develop)](](

We use readthedocs to build the online documentation, and have separate versions for the `master` and `develop` branches: (for the latest official release e.g. on PyPI) and (for the latest cutting-edge code available from the develop branch on github).

## Compile documentation locally

To compile the GPflow documentation locally:

1. Change to this directory (e.g. `cd doc` if you are in the GPflow git repository's base directory)

2. Install doc dependencies
   pip install sphinx sphinx_rtd_theme numpydoc nbsphinx sphinx_autodoc_typehints ipython jupytext jupyter_client ipywidgets

3. Install pandoc
   pip install pandoc
   If pandoc does not install via pip, or step 5 does not work, go to (the PyPI package depends on the external system-wide installation of pandoc executables)

4. Generate auto-generated files
   * Notebooks (.ipynb): run `make -C source/notebooks -j 4` (here with 4 parallel threads)
   * API documentation (.rst): run `python source/`

5. Compile the documentation
   make html

6. Check documentation locally by opening (in a browser) build/html/index.html

## Setup for automatic documentation generation

Upon each merge to the `develop` branch, this repository's [CircleCI configuration](../.circleci/config.yml) runs the `trigger-docs-generation` step
which triggers a CircleCI build on the [GPflow/docs repository](
This clones the latest GPflow develop branch and compiles all notebooks from jupytext to .ipynb
(setting the `DOCS` environment variable so that notebook optimisations are run to convergence)
and runs the `` script as above to generate the .rst files for API documentation.
(This script is run on CircleCI, instead of ReadTheDocs, as it requires gpflow and hence `tensorflow` and `tensorflow_probability` to be installed, but TensorFlow is too large to be installed inside the ReadTheDocs docker images.)
ReadTheDocs then pulls in these auto-generated files as well as all other files within this doc/ directory to actually build the documentation using Sphinx.
back to top