def init(self, optimizer=None, model=None, uid=None):
"""
Parameters
----------
optimizer: Dict
Optimizer hyper parameters!s
model: Dict
model hyper parameters
uid: Optional[str]
trial id to use for logging.
When using orion usually it already created a trial for us we just need to append to it
"""
optimizer = select(optimizer, {})
model = select(model, {})
self.model.init(**model)
# list of all parameters this task has
parameters = list()
parameters.append({'params': self.model.parameters()})
# We need to set the device now so optimizer receive cuda tensors
self.set_device(self.device)
self.optimizer.init(self.model.parameters(), **optimizer)
self.hyper_parameters = {'optimizer': optimizer, 'model': model}
# Get all hyper parameters even the one that were set manually
hyperparameters = self.get_current_space()
# Trial Creation and Trial resume
self.metrics.new_trial(hyperparameters, uid)
self.set_device(self.device)
def init(self,
gamma=0.99,
optimizer=None,
lr_schedule=None,
model=None,
uid=None):
"""
Parameters
----------
optimizer: Dict
Optimizer hyper parameters
lr_schedule: Dict
lr schedule hyper parameters
model: Dict
model hyper parameters
gamma: float
reward discount factor
trial: Optional[str]
trial id to use for logging.
When using orion usually it already created a trial for us we just need to append to it
"""
optimizer = select(optimizer, {})
lr_schedule = select(lr_schedule, {})
model = select(model, {})
self.gamma = gamma
self.actor_critic.init(**model)
# We need to set the device now so optimizer receive cuda tensors
self.set_device(self.device)
self.optimizer.init(self.actor_critic.parameters(),
override=True,
**optimizer)
self.lr_scheduler.init(self.optimizer, override=True, **lr_schedule)
self.hyper_parameters = {
'optimizer': optimizer,
'lr_schedule': lr_schedule,
'model': model
}
parameters = {}
parameters.update(optimizer)
parameters.update(lr_schedule)
parameters.update(model)
self.metrics.new_trial(parameters, uid)
self.set_device(self.device)
def __init__(self, classifier, optimizer, lr_scheduler, dataloader, criterion=None, device=None,
storage=None, preprocessor=None, metrics=None):
super(Classification, self).__init__(device=device)
criterion = select(criterion, CrossEntropyLoss())
self._first_epoch = 0
self.current_epoch = 0
self.classifier = classifier
self.optimizer = optimizer
self.lr_scheduler = lr_scheduler
self.dataloader = dataloader
self.criterion = criterion
self.preprocessor = Preprocessor()
# ------------------------------------------------------------------
self.metrics.append(ElapsedRealTime().every(batch=1))
self.metrics.append(SampleCount().every(batch=1, epoch=1))
self.metrics.append(OnlineTrainAccuracy())
self.metrics.append(Speed())
# All metrics must be before ProgressView and CheckPointer
if metrics:
for metric in metrics:
self.metrics.append(metric)
self.metrics.append(ProgressView(self.metrics.get('Speed')))
if storage:
self.metrics.append(CheckPointer(storage=storage))
# ------------------------------------------------------------------
if preprocessor is not None:
self.preprocessor = preprocessor
self.hyper_parameters = {}
def init(self, optimizer=None, lr_schedule=None, model=None, uid=None):
optimizer = select(optimizer, {})
lr_schedule = select(lr_schedule, {})
model = select(model, {})
self.detector.init(**model)
self.set_device(self.device)
self.optimizer.init(self.detector.parameters(),
override=True,
**optimizer)
self.lr_scheduler.init(self.optimizer, override=True, **lr_schedule)
parameters = {}
parameters.update(optimizer)
parameters.update(lr_schedule)
parameters.update(model)
# Trial Creation and Trial resume
self.metrics.new_trial(parameters, uid)
self.set_device(self.device)