Using natural language and program abstractions to instill human inductive biases in machines

Open in new window