Sample Efficient Reinforcement Learning Through Human Demonstrations
This folder contains a set of baselines for solving the MineRLObtainDiamond-v0
environment in the NeurIPS 2019 MineRL Competition.
- Main Competition where you can sign-up, read the rules, and check the leaderboard is here.
- Submission Starter Template with documentation and starter code for your first submission to the competition. (The baselines in this repository are based off of this!)
- Documentation for the dataset and the environments is found here!
- Questions about getting started or running these baselines should be directed to the competition forum or discord server
- Technical issues related to the
minerl
python package should be submitted through the MineRL GitHub page!
Prerequisites. To get started with these baselines and incorporate them into your own submissions make sure you have:
- Registered with the competition on the main competition page,
- Cloned your own submission starter template and created a repository at AIcrowd.
Usage. The various baselines associated with this competition are contained within this folder (competition/
) as git
repositories (submodules) which fork the AIcrowd submission template.
To download all of the baseline code run the following:
git clone https://github.com/minerllabs/baselines.git --recurse-submodules
Each competition baseline is structured exactly as a submission to AIcrowd (containing train.py
, test.py
, etc.) and contains a link to the its leaderboard score!
You can additionally base your submission off of one of the existing baselines in this repository by simply forking or cloning one of the baseline repositories in this folder! Good luck :)