@author: Vincent Marois & Tomasz Kornuta

To use Mi-Prometheus, we need to first install PyTorch. Refer to the official installation guide of PyTorch for its installation. It is easily installable via conda, or you can compile it from source to optimize it for your machine.

Mi-Prometheus is not (yet) available as a pip package, or on conda. However, we provide the script and recommend to use it for the installation of Mi-Prometheus. First please clone the MI-Prometheus repository:

git clone
cd mi-prometheus/

Then, install the dependencies by running:

python install

This command will install all dependencies of Mi-Prometheus via pip. If you plan to develop and introduce changes, please call the following command instead:

python develop

This will enable you to change the code of the existing problems/models/workers and still be able to run them by calling the associated mip-* commands. More in that subject can be found in the following blog post on dev_mode.

Please note that it will not install PyTorch, as we have observed inconsistent and erratic errors when we were installing it from pip. Use conda, or compile it from source instead. The indicated way by the PyTorch team is conda.

The will register Mi-Prometheus as a package in your Python environment so that you will be able to import it:

>>> import miprometheus as mip

And then you will be able to access the API, for instance:

>>> datadict = mip.utils.DataDict()

Additionally, also creates aliases for the workers, so that you can use them as regular commands:

mip-offline-trainer --c path/to/your/config/file

The currently available commands are:

  • mip-offline-trainer
  • mip-online-trainer
  • mip-tester
  • mip-grid-trainer-cpu
  • mip-grid-trainer-gpu
  • mip-grid-tester-cpu
  • mip-grid-tester-gpu
  • mip-grid-analyzer
  • mip-index-splitter

Use –h to see the available flags for each command.

You can then delete the cloned repository and use these commands to run a particular worker with your configurations files.

Please note that we provide multiple configuration files in configs/ (which we use daily for our research & development). Feel free to copy this folder somewhere and use these files as you would like. We are investigating on how we could make these files usable in an easier way.