What can you register ?

To really leverage your Mode Registry, it's really important that you understand what information can be stored in a Model and how they are used in Picsellia.

Here is what a Model looks like when it has all the information filled

All the information that can be stored here are:

  • framework (tensorflow, pytorch, onnx...)
  • files (saved_model, checkpoint, config...)
  • labels (classes from labelmap)
  • base parameters (hyperparameters)
  • source Experiment
  • training data (base Dataset)


Some information can be added when you create or update the model and others (source experiment, training data...) only becomes available when you export an Experiment, we will explain this below

To see how we can get our first model in the Registry, let's head to the next tutorial πŸ‘‡