Name: Patrick Kidger
Type: User
Company: Cradle.bio
Bio: Building a JAX ecosystem: Equinox, Diffrax, Lineax, etc. ---- SciML, numerics, neural ODEs ---- Solo traveller, martial artist, scuba diver!
Twitter: PatrickKidger
Location: Zürich
Blog: https://kidger.site
Patrick Kidger's Projects
Github Action to: Check version / Test / git tag / GitHub Release / Deploy to PyPI
Keras implementation of AdaBound
The very start of a game I was toying with before I got distracted by the PhD...
Unbearably fast near-real-time runtime type-checking in pure Python.
Simple PyTorch helpers. (I think we've probably all written one of these for ourselves!)
The Python programming language
Code for "Deep Signature Transforms" (NeurIPS 2019)
Numerical differential equation solvers in JAX. Autodifferentiable and GPU-capable. https://docs.kidger.site/diffrax/
Master's Dissertation: Polynomial Approximation of Holomorphic Functions
Elegant easy-to-use neural networks + scientific computing in JAX. https://docs.kidger.site/equinox/
A Python package for computer vision models in the Equinox ecosystem.
A CLI tool. Extract regexes from stdout (e.g. URLs) and invoke commands on them (e.g. open the webpage).
Code for "'Hey, that's not an ODE:' Faster ODE Adjoints via Seminorms" (ICML 2021)
Code for "Generalised Interpretable Shapelets for Irregular Time Series"
Named Tensors for Legible Deep Learning in JAX
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Differentiable Fluid Dynamics Package
Type annotations and runtime checking for shape and dtype of JAX/NumPy/PyTorch/etc. arrays. https://docs.kidger.site/jaxtyping/
Tools for working with Keras.
Applying machine learning to help numerically solve the Camassa-Holm equation.
Legibile, Scalable, Reproducible Foundation Models with Named Tensors and Jax
Linear solvers in JAX and Equinox. https://docs.kidger.site/lineax
Counts lines of Python code.
Round robin matching algorithm.
Modify which files MkDocs includes or excludes.
Make posters from Markdown files.
A simple machine learning project for weather observations.
Code for "Neural Controlled Differential Equations for Irregular Time Series" (Neurips 2020 Spotlight)
Nonlinear optimisation (root-finding, least squares, ...) in JAX+Equinox. https://docs.kidger.site/optimistix/