This repository implements the demonstration extrapolation method from Imitation Learning to Outperform Demonstrators by Directly Extrapolating Demonstrations. The implementation of SDN are based on their official codebase, the implementations of PPO and GAIL are based on
pytorch-a2c-ppo-acktr-gail by ikostrikov, and the implementation of GAIFO is based on
opolo-code by illidanlab. Please refer to run.sh
for running the demonstration extrapolation method.
- python: 3.7.5
- Torch: 1.3.1
- gym: 0.15.4