Uses RUDDER on PPO to solve environments with delayed rewards. Our RUDDER paper is available at https://arxiv.org/abs/1806.07857.

See https://widmi.github.io/ for a quick tutorial on reward redistribution with tensorflow implementation and code-snippets that you can run on your CPU.

Our code is based on the OpenAI Baselines package, in which we included our implementation of RUDDER for PPO for ATARI games with delayed rewards. Additionally, the Tensorflow Layer Library (TeLL) package (v1.0.0) is required for RUDDER.

RUDDER for PPO2 and documentation for the configuration is located in the folder baselines/ppo2_rudder.

Modified/new files:

• baselines/common/atari_wrappers.py
• baselines/common/distributions.py
• baselines/common/vec_env/vec_frame_stack.py
• baselines/ppo2_rudder/
• logger.py
• README.md

Videos are available at: