Python language support

rx provides built-in support for Python projects that use pip and/or conda for dependency management. If you have a requirements.txt or environment.yml file, rx will automatically set up an environment on your worker based on that file.

Pip

To automatically keep a pip environment up-to-date, the Docker image you chose must have Python interpreter installed (version 3.7+). Consider using one of the Python images. By deafult, the virtual environment is in /root/venv on your worker and is activated prior to rx running any commands.

Whenever your requirements.txt file changes, rx will automatically download new packages into your virtual environment. These packages will be built for the worker’s architecture (e.g., Cuda) and can often be installed pre-built for popular packages/architectures, speeding up setup times.

See the getting started example for more info on requirements.txt.

Pip environments have the environment variable VIRTUAL_ENV set and prefix the PATH with /root/venv/bin.

Conda

To automatically keep a Conda environment up-to-date, the Docker image you choose must have Conda installed. Consider using the rx’s own Conda-optimized container by specifying in your config:

image:
  repository: 'runrx/conda'
remote:
  hardware:
    processor: 'gpu'