[D] What JAX NN library to use?

• equinox

  31 1,809 9.2 Python

  Elegant easy-to-use neural networks + scientific computing in JAX. https://docs.kidger.site/equinox/

See CONTRIBUTING.md for details on getting started, and how to build the documentation locally.

• diffrax

  21 1,237 8.2 Python

  Numerical differential equation solvers in JAX. Autodifferentiable and GPU-capable. https://docs.kidger.site/diffrax/

Okay, that's easy-to-use. What about generality? The evidence here is Diffrax, which is a library of differential equation solvers... built around Equinox. How so? It's because Equinox is something more general than a "neural network library". It's a parameterised function library. So whilst neural networks are functions parameterised by their weights/biases/etc., Equinox is general enough to handle far more complicated use-cases; in this case all of the Diffrax solvers are Equinox modules. So are all of the adaptive step size controllers. And the adjoint methods. And so on.

• InfluxDB

  www.influxdata.com featured

  Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.

  

• furo

  7 2,466 8.6 Sass

  A clean customizable documentation theme for Sphinx

On another note, what did you dislike in Sphinx ? I found it pretty easy to work with until now and there are quite nice themes, like Furo (https://github.com/pradyunsg/furo), which is actually pretty similar to your current docs theme. I used it recently for one of my projects (see https://francois-rozet.github.io/piqa/piqa.psnr.html).

• mkdocs_include_exclude_files

  1 1 0.0 Python

  Modify which files MkDocs includes or excludes.

(I already have a couple of these. See e.g. [pytkdocs_tweaks](github.com/patrick-kidger/pytkdocs_tweaks) and include_exclude_files.)

Suggest a related project