models

Models

This package containts two high-level models that can be used to inherit some useful methods and a general standardized structure.

TransfomersModel

TransformersModel only overrides configure_optimizers by returning a better optimizer and the relative scheduler and finally provides a add_argparse_args to automatically add the parameters of the optimizer to the global parser.

Example:

>>> parser = ArgumentParser()
>>> TransformerModel.add_argparse_args(parser)
>>> save_transformers_callback = callbacks.TransformersModelCheckpointCallback(hyperparameters)
>>> hyperparameters = parser.parse_args()