/episodic-policy-minings

Primary LanguagePython

Status: Archive (code is provided as-is, no updates expected)

Exploration by Random Network Distillation

Yuri Burda*, Harri Edwards*, Amos Storkey, Oleg Klimov
*equal contribution

OpenAI
University of Edinburgh

Installation and Usage

The following command should train an RND agent on Montezuma's Revenge

python run_atari.py --gamma_ext 0.999

To use more than one gpu/machine, use MPI (e.g. mpiexec -n 8 python run_atari.py --num_env 128 --gamma_ext 0.999 should use 1024 parallel environments to collect experience on an 8 gpu machine).

Blog post and videos