Pipette speeds up your experiments by figuring out which parts of your workflow have to be re-run, and which parts can be read from earlier results. It also organizes the results of the tasks in the workflow, so that you can run different steps on different machines, locally, or in the cloud. You no longer need to track files in spreadsheets, and invent clever naming schemes to keep everything straight.

You can get started by following the example at tests/premierleague.py.

Detailed documentation is available at https://allenai.github.io/pipette/.

Add pipette to your own project with pip install https://github.com/allenai/pipette/releases/download/v1.4.1/pipette-1.4.1-py3-none-any.whl.