Inference module
- class multimodal_compare.eval.infer.MultimodalVAEInfer(model)
Class that includes methods for direct model testing and evaluation. The users can use this class to test their trained model with their own data (outside Dataloaders), compare multiple models etc.
- eval_statistics()
Runs the official evaluation routine defined in trainer.py. If applicable, calculates statistics
- get_base_path(path)
Finds the base directory of the model
- get_config()
Creates the Config instance based on the provided path
- get_datamodule(load_data=True)
Creates an instance of the DataModule class. Necessary for accessing the specific data processing tools
- get_wrapped_model()
Returns the Trainer class with loaded Datamodule outside Pytorch Lightning
- make_dataloaders()
Loads the train, val (and test, if available) dataloaders within the datamodule class