def __call__(self, engine: Engine, batchdata: Dict[str, torch.Tensor]):
if batchdata is None:
raise ValueError("Must provide batch data for current iteration.")
for j in range(self.max_interactions):
inputs, _ = engine.prepare_batch(batchdata)
inputs = inputs.to(engine.state.device)
engine.network.eval()
with torch.no_grad():
if engine.amp:
with torch.cuda.amp.autocast():
predictions = engine.inferer(inputs, engine.network)
else:
predictions = engine.inferer(inputs, engine.network)
batchdata.update({CommonKeys.PRED: predictions})
batchdata[self.key_probability] = torch.as_tensor(
([1.0 - ((1.0 / self.max_interactions) * j)]
if self.train else [1.0]) * len(inputs))
batchdata = self.transforms(batchdata)
return engine._iteration(engine, batchdata)
def attach(self, engine: Engine) -> None:
engine.add_event_handler(Events.ITERATION_STARTED, self)
def attach(self, engine: Engine) -> None:
if not engine.has_event_handler(self, Events.ITERATION_COMPLETED):
engine.add_event_handler(Events.ITERATION_COMPLETED, self)
def attach(self, engine: Engine) -> None:
engine.add_event_handler(
Events.ITERATION_COMPLETED(every=self.interval), self, "iteration")
engine.add_event_handler(Events.EPOCH_COMPLETED(every=1), self,
"epoch")
def attach(self, engine: Engine) -> None:
if not engine.has_event_handler(self.iteration_completed, Events.EPOCH_COMPLETED):
engine.add_event_handler(Events.EPOCH_COMPLETED, self.iteration_completed)