def __init__(self,
model_creator,
optimizer_creator,
loss_creator=None,
scheduler_creator=None,
training_operator_cls=None,
config=None,
use_tqdm=False,
scheduler_step_freq=None):
self.model_creator = model_creator
self.optimizer_creator = optimizer_creator
self.loss_creator = loss_creator
self.scheduler_creator = scheduler_creator
self.training_operator_cls = training_operator_cls or TrainingOperator
self.config = {} if config is None else config
self.timers = utils.TimerCollection()
self.epochs = 0
self.models = None
self.optimizers = None
self.criterion = None
self.schedulers = None
self.train_loader = None
self.validation_loader = None
self.training_operator = None
self.use_tqdm = use_tqdm
self.scheduler_step_freq = scheduler_step_freq
self.backend = "torch-local"
self.rank = 0
self.size = 0
def __init__(self,
model_creator,
data_creator,
optimizer_creator,
loss_creator=None,
scheduler_creator=None,
training_operator_cls=None,
config=None,
serialize_data_creation=True,
use_tqdm=False,
scheduler_step_freq=None):
self.model_creator = model_creator
self.optimizer_creator = optimizer_creator
self.loss_creator = loss_creator
self.data_creator = data_creator
self.scheduler_creator = scheduler_creator
self.training_operator_cls = training_operator_cls or TrainingOperator
self.config = {} if config is None else config
self.timers = utils.TimerCollection()
self.epochs = 0
self.models = None
self.optimizers = None
self.criterion = None
self.schedulers = None
self.train_loader = None
self.validation_loader = None
self.training_operator = None
self.serialize_data_creation = serialize_data_creation
self.use_tqdm = use_tqdm
self.scheduler_step_freq = scheduler_step_freq