Install the toolkit, learn the mechanics, arrive ready.
The permanent setup-and-basics companion to the hands-on tutorial track of the Reinforcement Learning Bootcamp. This is the single source of truth for installation — every edition's tutorial repository links here rather than repeating it.
git clone https://github.com/SARL-PLUS/rl-bootcamp-setup.git
cd rl-bootcamp-setup
conda env create -f environment.yml
conda activate rlbootcamp
python scripts/smoke_test.pyIf the last command prints Everything works. You are ready for the bootcamp.
you are set. If not, see Troubleshooting.
Do this before the event. PyTorch and MuJoCo are large downloads, and venue Wi-Fi cannot serve a full room at once.
environment.yml |
The rlbootcamp Conda environment — the whole toolkit in one file. |
scripts/smoke_test.py |
One command that proves your machine is ready. |
scripts/plot_training.py |
Turns a run's TensorBoard logs into the figures used in the docs. |
examples/ |
A complete worked example: baseline → train → evaluate → video. |
docs/ |
The handbook source (MkDocs Material). |
python examples/01_random_agent.py # measure the baseline first
python examples/02_train.py # train SAC on Pendulum-v1 (~5-10 min)
python examples/03_evaluate.py runs/<run>/model.zip --videoThe tutorial exercises, environments and solutions. Those are distributed to participants at each event, in their own repository — deliberately, since the exercises are worth more if you have not read the answers first.
This repository is intentionally year-agnostic: dates, venue and schedule live on the event website for the edition you are attending.
pip install -r requirements-docs.txt
mkdocs serve # http://127.0.0.1:8000Corrections and improvements are very welcome — especially troubleshooting entries for setups we have not seen. Please keep contributions generic: no edition-specific content, and nothing that reveals tutorial exercises.
MIT. If you use this to run your own workshop, attribution is appreciated.