A tool to help train RL agents in novel environments.
Polycraft AI Lab (PAL)
A tool to help train reinforcement learning models to handle novel environments.
Polycraft AI Lab consists of a wrapper for Polycraft World game environments. These environments can used to train RL models that respond to novel tasks and scenarios.
1a. Install via package
First, download Polycraft AI Lab using pip:
pip install polycraft-lab
This downloads the
polycraft-lab package from pip, which contains tools to
easily set up and manage the Polycraft game client.
To install the most recent changes (experimental version):
pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://test.pypi.org/simple/ polycraft-lab
1b. Local Installation
Alternatively, to install the latest version from source:
git clone https://github.com/PolycraftWorld/polycraft-ai-lab pip install ./polycraft-ai-lab
If you installed PAL using method (1a), this isn't required.
2. Import and Use
Now train your agent like you would do with any other gym-style environment:
from polycraft_lab.envs.helpers import setup_env env = setup_env('pogo_stick') observation = env.reset() for _ in range(1000): env.render() action = env.action_space.sample() # your agent here (this takes random actions) observation, reward, done, info = env.step(action) if done: observation = env.reset() env.close()
Polycraft AI Lab also contains a wrapper [WIP] to start experiment creation from the command line. The following begins the experiment creation process by launching Minecraft:
python -m polycraft_lab.ect --create EXPERIMENT_NAME --launch
python -m polycraft_lab.ect --create EXPERIMENT_NAME will allow you to
create simpler experiments, allowing configuration of more high-level domain
attributes, like action space and a preconfigured goal, such as finding
diamonds as quickly as possible.
Clone out the repository:
git clone https://github.com/PolycraftWorld/polycraft-ai-lab.git
Optionally, you can create a virtual environment to store dependencies.
In any case, install the dependencies:
pip install -r requirements.txt
Alternatively, a virtual environment can be created with the necessary dependencies by running:
cd polycraft-ai-lab pipenv install
Polycraft AI Lab will be distributed using pip.
The easy way to upload to upload to the test PyPI index:
To release to the live PyPI index:
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Hashes for polycraft_lab-0.1.0a0-py3-none-any.whl