Skip to content

Reasoning agent project: Policy Networks for Non-Markovian Reinforcement Learning Rewards

Notifications You must be signed in to change notification settings

PietroManganelliConforti/ReasoningAgent_project

Repository files navigation

Reasoning agent project: Policy Networks for Non-Markovian Reinforcement Learning Rewards

Setup (tested on python 3.8.10 and 3.8.12)

  • Install environment:
git clone https://github.com/ireneb97/RA_project.git
cd RA_project
docker pull whitemech/lydia:latest
echo '#!/usr/bin/env sh' > lydia
echo 'docker run -v$(pwd):/home/default whitemech/lydia lydia "$@"' >> lydia
sudo chmod u+x lydia
sudo mv lydia /usr/local/bin/
  • Create and initialize environment:
python3 -m venv ./venv
source venv/bin/activate
  • Install dependencies:
pip install -r requirements.txt

Train an agent

In folder config are stored some configurations we have used. We suggest to not to change those files as they already store the best values for each map and algorithm pair. You can run one of them (e.g. config1.cfg) by running the command:

python3 main.py --config_file config1.cfg

Test an agent

In folder model we saved our trained agents, you can run one of them by using this command:

python3 main.py --trained_model_path model/ppo

Miscellaneous

  • You can read in details about this project here, inside our report.
  • You can see our video presentation here.
  • You can see the slides we used in our presentation here.

Authors

About

Reasoning agent project: Policy Networks for Non-Markovian Reinforcement Learning Rewards

Topics

Resources

Stars

Watchers

Forks

Languages