kvelt00 / IMT-Agent

Integrated Multi-Task Reinforcement Learning Agent Architecture

IMT-Agent

Integrated Multi-Task Reinforcement Learning Agent Architecture

Installation

1. git submodule update --init --recursive

   • pull the go game repository

2. conda env create -f environment.yml

   • installs all the required packages

3. conda activate IMT-Agent

   • activate the conda environment

4. pip install GymGo/.

   • install the gym_go environment

Running

1. python src/go.py
   • will run the imt agent (black) against a standard q-table (white)
   • to change this, modify src/go.py line 91