def _train_task(self, train_loader, val_loader):
loops.single_loop(train_loader,
val_loader,
self._multiple_devices,
self._network,
self.unsupervised_training["epochs"],
self._optimizer,
scheduler=self._scheduler,
train_function=self._unsupervised_forward,
eval_function=self._accuracy,
task=self._task,
n_tasks=self._n_tasks,
disable_progressbar=self._disable_progressbar)
if self.new_supervised_training:
logger.info("Finetuning new")
self.finetuning(train_loader, val_loader,
self.new_supervised_training,
[{
"params": self._network.classifier.new_weights
}])
if self.all_supervised_training:
logger.info("Finetuning all")
self.finetuning(
train_loader, val_loader, self.all_supervised_training,
[{
"params": self._network.classifier.parameters()
}])
def finetuning(self, train_loader, val_loader, config, params):
if config["sampling"] == "undersampling" or \
(config["sampling"] == "next_undersampling" and self._task > 0):
self._data_memory, self._targets_memory, _, _ = self.build_examplars(
self.inc_dataset, self._herding_indexes)
loader = self.inc_dataset.get_memory_loader(*self.get_memory())
elif config["sampling"] == "new":
class_ids = list(
range(self._n_classes - self._task_size, self._n_classes))
_, loader = self.inc_dataset.get_custom_loader([class_ids],
mode="train")
else:
loader = train_loader
optimizer = factory.get_optimizer(params, config["optimizer"],
config["lr"], self._weight_decay)
scheduler = factory.get_lr_scheduler(config["scheduling"],
optimizer,
nb_epochs=config["epochs"],
lr_decay=config.get(
"lr_decay", 0.1),
task=self._task)
loops.single_loop(loader,
val_loader,
self._multiple_devices,
self._network,
config["epochs"],
optimizer,
scheduler=scheduler,
train_function=self._supervised_forward,
eval_function=self._accuracy,
task=self._task,
n_tasks=self._n_tasks,
disable_progressbar=self._disable_progressbar)
def _train_task(self, train_loader, val_loader):
loops.single_loop(train_loader,
val_loader,
self._multiple_devices,
self._network,
self._n_epochs,
self._optimizer,
scheduler=self._scheduler,
train_function=self._forward_loss,
eval_function=self._accuracy,
task=self._task,
n_tasks=self._n_tasks)
def _train_task(self, train_loader, val_loader):
# Oracle retrains everytime a model on all data:
_, train_loader = self.inc_dataset.get_custom_loader(list(
range(self._n_classes)),
mode="train")
_, val_loader = self.inc_dataset.get_custom_loader(list(
range(self._n_classes)),
data_source="val")
loops.single_loop(train_loader,
val_loader,
self._multiple_devices,
self._network,
self._n_epochs,
self._optimizer,
scheduler=self._scheduler,
train_function=self._forward_loss,
eval_function=self._accuracy,
task=self._task,
n_tasks=self._n_tasks)