The ModelDraft is the programmatic interface equivalent to the model creation page in the Predibase UI. Using the ModelDraft object, you are able to edit the config and dataset before starting a training job by accessing the objects' properties. ModelDraft is a precursor to a model and is a key component enabling an iterative approach to model development. Checkout
ModelDraft.train for more details on how this object is used.
repo The repo object that the model which created this draft belongs to
config The current config set for this model draft. If
.train()is called, this is the config that will be used for training.
dataset The current dataset object set for this model draft. If
.train()is called, this is the dataset that will be used for training.