Top-down program synthesis with a REPL and reinforcement learning

This is the repo for a project I did combining top-down program synthesis with reinforcement learning. The full blog post can be read here.