/diverse-ensemble-citylearn

The codebase replicating the results of the "Ensembling Diverse Policies Improves Generalizability of Reinforcement Learning Algorithms in Continuous Control Tasks" paper and similarly titled thesis work.

Primary LanguagePythonMIT LicenseMIT

diverse-ensemble-citylearn

The codebase replicating the results of the "Ensembling Diverse Policies Improves Generalizability of Reinforcement Learning Algorithms in Continuous Control Tasks" paper and similarly titled thesis work.

Installation

Python 3.9 is recommended

pip install -r requirements.txt
pip install citylearn==1.3.6 --no-deps

Running an experiment

python train.py --wandb-entity <wandb team/user name>