This repo contains the code of a Deep Reinforcement Learning agent that is trained to solve Quadratic Assignment Problems (QAP).
For details and results see the thesis paper. As an introduction to the problem and setup, see the proposal presentation.
Requires python 3.9
Install poetry
pip install poetry
Setup environment
poetry install
poetry shell
From within the environment execute either one of the run_*
scripts or one of the scripts in scripts/*
.
Plots and small experiments.
Within the environment, run
jupyter notebook
then open the browser interface and open a notebook in the evaluation
folder
Shows some information about the trained agent and allows interactive execution of the agents.
bokeh serve evaluation/server
Only cover some critical aspects
pytest