This repository accompanies a paper published at the Reinforcement Learning Conference 2024:

Towards General Negotiation Strategies with End-to-End Reinforcement Learning
B.M. Renting, T.M. Moerland, H.H. Hoos, C.M. Jonker

Installation

Code was written for Python 3.10 and CUDA 12.2. To install, we recommend the usage of a Python environment:

python3.10 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install --upgrade setuptools wheel
pip install -r requirements.txt

Usage

To train and test GNN based negotiation agents use the ppo.py and evaluate.py.

The default arguments are set to match the settings in the paper. To change/view the command line arguments run python ppo.py --help.

The figures from the paper can be reproduced using the paper_results.py script. A number of tests is defined in the TESTS variable in the script. Each of these tests can be run as follows: python paper_results.py --test_num $TEST_LIST_INDEX. The trained models are included in this repository. Results will be saved in the analysis/data and analysis/figures directories.