Scitator / animal-olympics-starter-kit

NeurIPS 2019: Animal-AI Olympics
Catalyst starter kit

Environment setup.

(Taken from the official repo).

The Animal-AI package works on Linux, Mac and Windows, as well as most Cloud providers. Note that for submission to the competition we only support linux-based Docker files.

We recommend using a virtual environment specifically for the competition. You will need python3.6 installed (we currently only support python3.6).

The main package is an API for interfacing with the Unity environment. It contains both a gym environment as well as an extension of Unity's ml-agents environments. You can install it via pip: pip install animalai Or you can install it from the source, head to animalai/ folder and run pip install -e .

Additionally download the environment for your system:

OS	Environment link
Linux	download v1.0.0
MacOS	download v1.0.0
Windows	download v1.0.0

You can now unzip the content of the archive to the assets folder and you're ready to go! Make sure the executable AnimalAI.* is in assets/. On linux you may have to make the file executable by running chmod +x assets/AnimalAI.x86_64. Head over to Quick Start Guide for a quick overview of how the environment works.

tl;dr

# system requirements
sudo apt-get install xvfb redis-server

# python requirements
conda create -n animal python=3.6 anaconda
source activate animal
pip install -r ./requirements.txt

# download and unzip env

Catalyst.RL

To train agents on the Animal Olympics environment, we can run Catalyst.RL as usual.

# start db node
redis-server --port 12012

# start trainer node
export GPUS=""  # like GPUS="0" or GPUS="0,1" for multi-gpu training
CUDA_VISIBLE_DEVICES="$GPUS" catalyst-rl run-trainer --config ./configs/_exp_common.yml ./configs/ppo.yml

# start sampler node
CUDA_VISIBLE_DEVICES="" catalyst-rl run-samplers --config ./configs/_exp_common.yml ./configs/ppo.yml --sampler-id=1

# view tensorboard logs
CUDA_VISIBLE_DEVICE="" tensorboard --logdir=./logs

For more information about Catalyst.RL you can check official repo, documentaiton and examples.