Formerly: Harvard Intelligent Probabilistic Systems Group -- Now at Princeton ORG

Efficiently computes derivatives of NumPy code.

Python library that automatically computes exact derivatives of functions written in standard Python and NumPy, handles loops, branches, and recursion without any code rewriting, developed at Harvard.

Spearmint Bayesian optimization codebase

Convolutional nets which can take molecular graphs of arbitrary size as input.

Exploring differentiation with respect to hyperparameters

Kayak is a library for automatic differentiation with applications to deep neural networks.

Implementation in C and Theano of the method Probabilistic Backpropagation for scalable Bayesian inference in deep neural networks.

A project to enable optimization of molecules by transforming them to and from a continuous representation.

Dependent multinomials made easy: stick-breaking with the Pólya-gamma augmentation

LaTeX package for randomizing author order based on a public seed.

Library of common tools for machine learning research.

Fastidious accounting of entropy streams into and out of optimization and sampling algorithms.

Implementation of an algorithm for Markov chain Monte Carlo with data subsampling

MCMC for the Dark Energy Spectroscopic Instrument

Code for performing Bayesian regression with structured sparsity from a Gaussian field.

Gradient-based variational autoencoders to generate class-conditional natural images.

A Numpy wrapper that adds a gpufloat32 dtype to Numpy.

Website for viewing a git repo as a lab notebook. Figures and text files can be included with markdown-like syntax.

A python framework for fitting biophysical models to optically recorded neural signals.

Linefeed-delimited pickle for Unix-style piping of arbitrary Python data

A simple abstraction layer for matrix computations in Python, making it easy to switch between CPU and NVIDIA or Intel coprocessors.

Github page for Harvard Intelligent Probabilistic Systems Group