Code for the paper "Learning Powerful Policies by Using Consistent Dynamics Model" presented at Workshop on "Structure & Priors in Reinforcement Learning" at ICLR 2019.
Arxiv link coming soon!
- Copy
config/sample.config.yaml
toconfig/config.yaml
. For testing, no changes are needed in the newly createdconfig.yaml
. - Install requirements using
pip3 install -r requirements.txt
. - Following packages needs to be installed from source:
- From the root dir, run
PYTHONPATH=$PWD python3 codes/app/main.py
- All the logic has a single entry point
codes/app/main.py
. What task is to be performed is controlled via the config files.
- If you get an error related to
patchelf
, tryconda install -c anaconda patchelf