Experiments

Experiments are the outermost class to control your training, it wraps your NetworkTrainer and provides utilities for cross-validation.

AbstractExperiment

class AbstractExperiment(n_epochs, *args, **kwargs)[source]

Bases: sphinx.ext.autodoc.importer._MockObject

Abstract Class Representing a single Experiment (must be subclassed for each Backend)

kfold(num_epochs: int, data: List[delira.data_loading.data_manager.BaseDataManager], num_splits=None, shuffle=False, random_seed=None, **kwargs)[source]

Runs K-Fold Crossvalidation

Parameters:
  • num_epochs (int) – number of epochs to train the model
  • data (list of BaseDataManager) – list of datamanagers (will be split for crossvalidation)
  • num_splits (None or int) – number of splits for kfold if None: len(data) splits will be validated
  • shuffle (bool) – whether or not to shuffle indices for kfold
  • random_seed (None or int) – random seed used to seed the kfold (if shuffle is true), pytorch and numpy
  • **kwargs – additional keyword arguments (completely passed to self.run())
static load(file_name)[source]

Loads whole experiment

Parameters:file_name (str) – file_name to load the experiment from
Raises:NotImplementedError – if not overwritten in subclass
run(train_data: Union[delira.data_loading.data_manager.BaseDataManager, delira.data_loading.data_manager.ConcatDataManager], val_data: Union[delira.data_loading.data_manager.BaseDataManager, delira.data_loading.data_manager.ConcatDataManager, None] = None, params: Optional[delira.training.parameters.Parameters] = None, **kwargs)[source]

trains single model

Parameters:
  • train_data (BaseDataManager or ConcatDataManager) – data manager containing the training data
  • val_data (BaseDataManager or ConcatDataManager) – data manager containing the validation data
  • parameters (Parameters, optional) – Class containing all parameters (defaults to None). If not specified, the parameters fall back to the ones given during class initialization
Raises:

NotImplementedError – If not overwritten in subclass

save()[source]

Saves the Whole experiments

Raises:NotImplementedError – If not overwritten in subclass
setup(*args, **kwargs)[source]

Abstract Method to setup a AbstractNetworkTrainer

Raises:NotImplementedError – if not overwritten by subclass

PyTorchExperiment

class PyTorchExperiment(params: delira.training.parameters.Parameters, model_cls: delira.models.abstract_network.AbstractPyTorchNetwork, name=None, save_path=None, val_score_key=None, optim_builder=<function create_optims_default_pytorch>, checkpoint_freq=1, trainer_cls=<class 'delira.training.pytorch_trainer.PyTorchNetworkTrainer'>, **kwargs)[source]

Bases: delira.training.experiment.AbstractExperiment

Single Experiment for PyTorch Backend

kfold(num_epochs: int, data: List[delira.data_loading.data_manager.BaseDataManager], num_splits=None, shuffle=False, random_seed=None, **kwargs)

Runs K-Fold Crossvalidation

Parameters:
  • num_epochs (int) – number of epochs to train the model
  • data (list of BaseDataManager) – list of datamanagers (will be split for crossvalidation)
  • num_splits (None or int) – number of splits for kfold if None: len(data) splits will be validated
  • shuffle (bool) – whether or not to shuffle indices for kfold
  • random_seed (None or int) – random seed used to seed the kfold (if shuffle is true), pytorch and numpy
  • **kwargs – additional keyword arguments (completely passed to self.run())
static load(file_name)[source]

Loads whole experiment

Parameters:file_name (str) – file_name to load the experiment from
run(train_data: Union[delira.data_loading.data_manager.BaseDataManager, delira.data_loading.data_manager.ConcatDataManager], val_data: Union[delira.data_loading.data_manager.BaseDataManager, delira.data_loading.data_manager.ConcatDataManager, None], params: Optional[delira.training.parameters.Parameters] = None, **kwargs)[source]

trains single model

Parameters:
  • train_data (BaseDataManager or ConcatDataManager) – holds the trainset
  • val_data (BaseDataManager or ConcatDataManager or None) – holds the validation set (if None: Model will not be validated)
  • params (Parameters) – the parameters to construct a model and network trainer
  • **kwargs – holds additional keyword arguments (which are completly passed to the trainers init)
Returns:

trainer of trained network

Return type:

AbstractNetworkTrainer

Raises:

ValueError – Class has no Attribute params and no parameters were given as function argument

save()[source]

Saves the Whole experiments

setup(params: delira.training.parameters.Parameters, **kwargs)[source]

Perform setup of Network Trainer

Parameters:
  • params (Parameters) – the parameters to construct a model and network trainer
  • **kwargs – keyword arguments