/SiMPL

Primary LanguagePython

SiMPL : Skill-based Meta-Policy Learning

[Project Page] [Paper]

Official implementation of the paper "Skill-based Meta-Reinforcement Learning", ICLR 2022.

Requirements

  • Python 3.6+
  • Mujoco
  • D4RL
  • pip packages listed in requirements.txt

Getting Started

  1. Install simpl package from code
cd /project/home
mkdir lib && cd lib
git clone https://github.com/namsan96/SiMPL.git
cd SiMPL
pip install -e .
  1. Copy and paste reproduce scripts to your source directory.
cp -r reproduce /project/home/your/source/directory
  1. Run & modify the scripts
cd /project/home/your/source/directory
python simpl_meta_train.py  --help

Downloading SPiRL Pre-trained Model

Currently, we don't support SPiRL skill extraction in this repository. Here are pre-trained SPiRL for each domain :

Download and feed to SiMPL & SPiRL scripts by --spirl-pretrained-path or -s.