MONAI Bundle Properties

Train properties

	description	required	id
bundle_root	root path of the bundle.	True	bundle_root
device	target device to execute the bundle workflow.	True	device
dataset_dir	directory path of the dataset.	True	dataset_dir
trainer	training workflow engine.	True	train#trainer
network_def	network module for the training.	False	network_def
max_epochs	max number of epochs to execute the training.	True	train#trainer#max_epochs
train_dataset	PyTorch dataset object for the training logic.	True	train#dataset
train_inferer	MONAI Inferer object to execute the model computation in training.	True	train#inferer
train_dataset_data	data source for the training dataset.	False	train#dataset#data
train_handlers	event-handlers for the training logic.	False	train#handlers
train_preprocessing	preprocessing for the training input data.	False	train#preprocessing
train_postprocessing	postprocessing for the training model output data.	False	train#postprocessing
train_key_metric	key metric to compute on the training data.	False	train#key_metric
evaluator	validation workflow engine.	False	validate#evaluator
val_interval	validation interval during the training.	False	val_interval
val_handlers	event-handlers for the validation logic.	False	validate#handlers
val_dataset	PyTorch dataset object for the validation logic.	False	validate#dataset
val_dataset_data	data source for the validation dataset.	False	validate#dataset#data
val_inferer	MONAI Inferer object to execute the model computation in validation.	False	validate#inferer
val_preprocessing	preprocessing for the validation input data.	False	validate#preprocessing
val_postprocessing	postprocessing for the validation model output data.	False	validate#postprocessing
val_key_metric	key metric to compute on the validation data.	False	validate#key_metric

Infer properties

	description	required	id
bundle_root	root path of the bundle.	True	bundle_root
device	target device to execute the bundle workflow.	True	device
dataset_dir	directory path of the dataset.	True	dataset_dir
dataset	PyTorch dataset object for the inference / evaluation logic.	True	dataset
evaluator	inference / evaluation workflow engine.	True	evaluator
network_def	network module for the inference.	True	network_def
inferer	MONAI Inferer object to execute the model computation in inference.	True	inferer
dataset_data	data source for the inference / evaluation dataset.	False	dataset#data
handlers	event-handlers for the inference / evaluation logic.	False	handlers
preprocessing	preprocessing for the input data.	False	preprocessing
postprocessing	postprocessing for the model output data.	False	postprocessing
key_metric	the key metric during evaluation.	False	key_metric

Meta properties

	description	required	id
version	bundle version	True	_meta_#version
monai_version	required monai version used for bundle	True	_meta_#monai_version
pytorch_version	required pytorch version used for bundle	True	_meta_#pytorch_version
numpy_version	required numpy version used for bundle	True	_meta_#numpy_version
description	description for bundle	False	_meta_#description
spatial_shape	spatial shape for the inputs	False	_meta_#network_data_format#inputs#image#spatial_shape
channel_def	channel definition for the prediction	False	_meta_#network_data_format#outputs#pred#channel_def