Skip to content

Repository files navigation

CMSC722_project

The folder eval_plots contains the figures comparing the Goal Skill Network agent against the monolithic agent, as well as the figures for hyperparameter tuning for Tabular SARSA.

The folder eval_results contains tables of evaluation results of each agent.

The folder training_plots contains figures showing the learning curves of the agents during training.

The folder q_values contains the learned action-value functions.

The file box_and_whiskers.py creates the box and whiskers plots.

The file eval_monolithic.py evaluates the monolithic agent.

The file taxi_agent_GSN.py defines the Goal Skill Network agent.

The file taxi_agent_monolithic.py defines the monolithic agent.

The file taxi_env.py modifies the Toy Taxi environment for the GSN agent.

The file taxi_planner_GSN.py defines the GTPyhop planner for the taxi environment, and runs the taxi environment

The file taxi_train_GSN.py trains the Goal Skill Network agent.

The file taxi_train_monolithic.py trains the monolithic taxi agent.

About

"A Hierarchical Goal-Biased Curriculum for Training Reinforcement Learning" (Patra et al., 2022) but for the Taxi Gym Environment

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Used by

Contributors

Languages