| Code as Policies is a representation of a robot-centric language model-generated program executing on a physical system. CaP extends PaLM-SayCan to enable language models to accomplish more complex robotic tasks with a complete representation of generic Python code. With CaP, Google recommends using language models to directly code bots with a few hints. Experiments demonstrate that CaP output code performs better than directly learning robotic tasks and outputting natural language actions. CaP allows a single system to perform a variety of complex and diverse robotic tasks without task-specific training. for… |
#Natural #language #code #generation #system #Code #Policies