Sam

Code for our paper "Visualizing and Understanding Atari Agents" (https://goo.gl/AMAoSc)

Visualizing how deep networks make decisions

A 1D analogue of the MNIST dataset for measuring spatial biases and answering Science of Deep Learning questions.

Code for our paper "Hamiltonian Neural Networks"

My academic blog

Training a transformer to generate cursive handwriting

Realistic Handwriting with Tensorflow

Nature's Cost Function (NCF). Finding paths of least action with gradient descent.

Temporal abstraction for autoregressive sampling

Optimizing neural networks in subspaces

Code for playing with random dot stereograms.

A high-performance Atari A3C agent in 180 lines of PyTorch

Studying Cell Growth with Neural Cellular Automata

Generative Adversarial Networks for the MNIST dataset

Learning the Enigma with Recurrent Neural Networks

A Python script that converts boolean expressions to flowcharts or directed acyclic graphs.

We simulate a wind tunnel, place a rectangular occlusion in it, and then use gradient descent to turn the occlusion into a wing.

A convolutional neural network implemented in pure numpy.

Coding structural optimization, from scratch, in 200 lines of Python

Differentiable Neural Computer in TensorFlow

Fork of codebase from Miles Cranmer's GitHub

Exploring the quantum nature of light with compton scattering

Simple MNIST baselines for 1) numpy backprop 2) dense nns 3) cnns 3) seq2seq

A neural network quantum ground state solver

The idea here was to teach an RNN to draw, pixel by pixel, over a template image using DDPG

A central location for my reinforcement learning experiments

Neural network experiments written purely in numpy

I use a one-layer neural network trained on the MNIST dataset to give an intuition for how common regularization techniques affect learning.

Forays into the world of deep learning using TensorFlow

A numerical model of fractal dynamics