Skip to content

Setup

Installation is documented in the participant primer, not here

One guide, maintained in one place, valid for every edition of the bootcamp:

sarl-plus.github.io/rl-bootcamp-setup

This page is a pointer. If anything here and the primer disagree, the primer is right.

The short version

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.py

The smoke test creates a Gymnasium environment, trains a short Stable-Baselines3 run, renders a frame and writes an .mp4. When the last line reads Everything works. You are ready for the bootcamp. you are ready for all three sessions.

Do this before you arrive. PyTorch and MuJoCo are several GB and the venue Wi-Fi cannot serve a full room at once.

Where to go

Installation Conda, the environment, the smoke test
Linux · macOS · Windows Platform-specific detail
VS Code · PyCharm Interpreter and notebook kernel
Working in the Conda environment Daily commands, Jupyter kernels
Troubleshooting Every failure we have seen, by symptom

Getting the tutorial code

This repository is published at the start of the event — you do not need it to get set up, and that is deliberate: you can be completely ready before the exercises exist.

Once it is available:

git clone https://github.com/SARL-PLUS/RL_Bootcamp_2026_tutorial.git
cd RL_Bootcamp_2026_tutorial
conda activate rlbootcamp

The same rlbootcamp environment runs all three sessions. There is no second environment to create.

The environment.yml in this repository

It is a copy of the primer's, kept identical so that cloning either repository gives you the same environment. Create it from whichever one you have; do not create both.

Where the notebooks are

Session Notebooks
#1 Fundamentals sessions/01-fundamentals/environments/ and .../solutions/
#2 Intermediate sessions/02-intermediate/crippled-ant/notebooks/

Session #1 is delivered entirely as notebooks, so confirm jupyter lab works before you arrive — the notebook check takes two minutes.

Problems specific to this repository

Anything about Conda, Python, MuJoCo, ffmpeg or rendering belongs in the primer's troubleshooting page. Only these are ours:

ModuleNotFoundError: No module named 'envs'

The AirTraffic package is not on the path. Run the scripts from inside the sessions/03-advanced/airtraffic/ directory — scripts/train_4d.py and scripts/score_4d.py already handle it from there — or add it explicitly:

import sys
sys.path.insert(0, "sessions/03-advanced/airtraffic")
from envs import Flight4DEnv
mkdocs serve fails with Unrecognised theme 'material'

The docs toolchain is not part of the participant environment. Only organisers need it:

pip install -r requirements-docs.txt