A Tetris reinforcement learning agent by Nikhil Suri and Soumil Singh. Read about our project purpose, specification, experiments methodology, discussion of results, and usage in our final paper.
This project uses git-lfs. Read about how we managed our large files of learned values after training sessions here.
Cloning this project in its entirety requires git-lfs. To clone a lightweight version which only includes weights learned by our best agent, clone our test branch with git clone --single-branch --branch test <remote-url>. Cloning the test branch does not require git-lfs, though it does require pygame. See our test branch here.
Initial contributions made by Soumil Singh are not linked to his Github account because of configuration issues.