BabyAI

There are three folders and eight other files

Folders

• levels contains the code for all levels
• rl contains an implementation of the Proximal Policy Optimization (PPO) RL algorithm
• utils contains files for reading and saving logs, demos and models. In this folder, agent.py defines an abstract class for an agent

Files

• arguments.py contains the value of default arguments shared by both imitation and reinforcement learning
• bot.py is a heuristic stack-based bot that can solve all levels
• efficiency.py contains hyperparmeter configurations we use for imitation learning sample efficiency
• evaluate.py contains functions used by IL and RL to evaluate an agent
• imitation.py is our imitation learning implementation
• model.py contains the neural network code
• plotting.py is used in plotting. It also contains Gaussian Process code used in measuring imitation learning sample efficiency