IMT-Agent
Integrated Multi-Task Reinforcement Learning Agent Architecture
Installation
-
git submodule update --init --recursive
- pull the go game repository
-
conda env create -f environment.yml
- installs all the required packages
-
conda activate IMT-Agent
- activate the conda environment
-
pip install GymGo/.
- install the gym_go environment
Running
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