CASRL

Context-aware safe reinforcement learning in non-stationary environments

Python3 requirements

scipy
matplotlib
pandas
gym==0.17.2
torch==1.5.1

To install the dependencies:

pip install -r requirements.txt

Running experiments on Cartpole-swingup

python casrl_cartpole.py

Running experiments on Healthcare

First install my fork of assistive_gym environment. Then run

python casrl_healthcare.py

Parameters are stored in in ./config