Experiments for Optimistic Risk-Aware Model-based RL. Bayesian MBRL with various planning strategies.
The dependencies can be installed using conda:
conda env create -f conda_env.yml
conda activate small_mbrl
To train "CVaR-constrained Upper CVaR" on Distributional-Shift environment:
python main2.py train_type=upper-cvar-opt-cvar hydra.run.dir=output/${now:%Y-%m-%d}
Logging is done through Weights and Biases on the public project: 'rabachi/small_mbrl'