def __init__(self, stage_name: str, is_train: bool,
data_producer: DataProducer):
super().__init__(name=stage_name)
self.data_loader = None
self.data_producer = data_producer
self._losses = None
self._is_train = is_train
self._last_result = None
self._epoch_end_event = events_container.add_event(
'EPOCH_END', Event(self))
self._epoch_start_event = events_container.add_event(
'EPOCH_START', Event(self))
self._batch_processed = events_container.add_event(
'BATCH_PROCESSED', Event(self))
def __init__(self, trainer: Trainer):
self._rules, self._prev_states = [], None
self._best_state_achieved_event = events_container.add_event(
"BEST_STATE_ACHIEVED", Event(self))
events_container.event(
trainer, 'TRAIN_DONE').add_callback(lambda t: self.reset())
def __init__(self, train_config: BaseTrainConfig, fsm: FileStructManager, device: torch.device = None):
MessageReceiver.__init__(self)
self._fsm = fsm
self.__epoch_num, self._cur_epoch_id = 100, 0
self._train_config = train_config
self._data_processor = TrainDataProcessor(self._train_config, device)
self._lr = LearningRate(self._data_processor.get_lr())
self._epoch_end_event = events_container.add_event('EPOCH_END', Event(self))
self._epoch_start_event = events_container.add_event('EPOCH_START', Event(self))
self._train_done_event = events_container.add_event('TRAIN_DONE', Event(self))
self._add_message('NEED_STOP')
def __init__(self, name: str):
self._name = name
self._stage_end_event = events_container.add_event(
'STAGE_END', Event(self))
def __init__(self):
self._metrics = []
self._metrics_groups = []
self._reset_metrics_event = events_container.add_event('BEFORE_METRICS_RESET', Event(self))