I have a function which is meant to be extendable, and would like to ensure that all input parameters keep being logged without the need to manually check/add them. Is there a way to do this automatically?

def __init__(self, model_name: str, num_classes: int, device: str = 'cuda:0', learning_rate: float = 5e-5,
                 do_layer_freeze: bool = True, extra_class_layers: Optional[Union[int, list]] = None,
                 fine_tune_dropout_rate: float = 0):

For scikit learn models this happens naturally using the autologger but doesn’t seem to with pytroch (Lightning). Currently, I am having to do it the manual way:

mlflow.log_params({'model_name': model_name,
                           'num_classes': num_classes,
                           'learning_rate': learning_rate,
                           'do_layer_freeze': do_layer_freeze,
                           'extra_class_layers': extra_class_layers,
                           'fine_tune_dropout_rate': fine_tune_dropout_rate})