This repo contains some naive implementations (only RSSM) of a purely model-based reinforcement learning algorithm that solves control tasks from images by efficient planning in a learned latent space.
To install this project, simply run the following command after cloning the repo:
pip install poetry
make installThis project implements Learning Latent Dynamics for Planning from Pixels (RSSM) (Hafner et al., 2019)
To run the training script
cd scripts
python trainer.py --config config/path/hereTo run the evaluation script
cd scripts
python tester.py --config config/path/here