Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machines

Open in new window