williamyuanv0 / holdem-agent

In this project, we implement a PPO-based RL agent for no-limit texas hold'em. This is joint work with Rahul Bhatia, Steven Friedman, and Nicole Giannopoulou as part of the Advanced Machine Learning class at Columbia University in Fall 2020.

holdem-agent

This project contains the final project work for Mark Aksen, Nicole Giannopoulou, Rahul Bhatia, and Steven Friedman in the Fall 2020 session of Advanced Machine Learning at Columbia University.

Overview of Directories

• experiments
  • Contains the output of the training experiments for each model type
• models
  • Contains trained model results for each type of model for comparison scripts
• notebooks
  • Contains Jupyter Notebooks that summarize the behavior of different agents in various scenarios
• src
  • Contains Python Source Code for training the new PPO Model, as well as a python environment defined by a Pipfile for running this training and evaluation