This repository contains the code for an imitation learning model and the DAgger algorithm for the CarRacing-v0 Gym Environment. This was part of a homework assignment for the Deep Reinforcement Learning course at NYU Courant taught by Lerrel Pinto.
DAgger helps the imitation learning agent learn correct actions when following sub-optimal trajectories:
Check out this paper to learn more about DAgger.
Clone this repository:
git clone https://github.com/kvgarimella/dagger.git
cd dagger
Install the requirements:
pip install -r requirements.txt
Run DAgger and train your model:
python dagger.py