MLPragueImitation

Repository for the 'Accelerating AI Through Human Knowledge: Teaching to Imitate Experts and Win on the Race Track' workshop at ML Prague 2025.

Link to Colab Notebook

The notebook is already available in Google Colab under this link.

Running the Docker Container

The container is available on dockerhub as tkietreiber/mlprague-imitation (Link), you can run it with all necessary arguments via

docker run -itp 8000:8000 -v $(pwd)/recordings:/app/recordings -v $(pwd)/agents:/app/agents tkietreiber/mlprague-imitation

and open frontend.html in a browser. (Just make sure you also created an agents and recordings folder.)

Building the Docker Container

Download the Dockerfile and run

docker build . -t tkietreiber/mlprague-imitation

Create a directories recordings and agents, e.g. with mkdir recordings agents, then you can start the docker container with

docker run -itp 8000:8000 -v $(pwd)/recordings:/app/recordings -v $(pwd)/agents:/app/agents tkietreiber/mlprague-imitation

Finally, open the frontend.html in a browser and play using W,A,S,D.

Running locally

Clone the repository and make sure you have Python 3.12 installed. (We recommend using a conda environment.) Then run

pip install -e .

to install all necessary requirements.