taichikuri/gail_ppo

Generative Adversarial Imitation Learning

GAIL_PPO

Generative Adversarial Imitation Learning implementation with Tensorflow2
This implementation only supports CartPole environment(OpenAI gym).

このリポジトリは逆強化学習型模倣学習アルゴリズムGenerative Adversarial Imitation LearningをTensorflow2で実装したものです。
学習環境についてはCartPole-v0でのみ検証済みです。
GAILについて解説したブログはこちらになります。

Relevant Papers

• Generative Adversarial Imitation Learning, Jonathan Ho, Stefano Ermon, 2016
  https://arxiv.org/abs/1606.03476
• Proximal Policy Optimization Algorithms, Schulman et al. 2017
  https://arxiv.org/abs/1707.06347

Requirements

• Python3
• tensorflow2
• gym
• tqdm

Usage

• clone this repo

$ git clone https://github.com/morikatron/GAIL_PPO.git

• change directory and run(using human demonstrations)

$ cd GAIL_PPO
$ python algo/run.py

• run(using ppo demonstrations)

$ python algo/run.py ppo

• make demo

$ python algo/make_demo.py  # human demonstrations
$ python algo/generate_demo.py  # ppo demonstrations

Performance Example