Reinforcement learning with auxiliary inputs.
Code for the corresponding paper published at TMLR.
Simply install everything in requirements.txt.
All experiments are run through the main.py file. Check out the arguments file in unc/args.py for a list of all possible hyperparameter configurations.
Experiments are defined in the hyperparameter files located in scripts/hparams.
Environments are set up such that you just need to specify the environment string
as an argument (see unc.args for more details).
Changes to this base environment are mostly gym.Wrappers around
this environment, in unc.envs.wrappers.