Version everything

Snapshot your model, your data, your environment, your parameters.

Dotscience versions and saves everything you need to rerun any version of your model.

To reproduce a model, you need to have the exact version of:

  • the model code
  • the datasets the model was trained on
  • the input parameters or hyperparameters supplied to the model

Dotscience saves all of these.

Dotscience also captures the environment used to run your model in, using Docker. This means that dotsicence records all the model’s dependencies, down to the exact version number: everything, in short, to rerun the model.