torchpack.train package

Submodules

torchpack.train.exception module

exception torchpack.train.exception.StopTraining

Bases: Exception

An exception thrown to stop training.

torchpack.train.summary module

class torchpack.train.summary.Summary

Bases: object

add_image(name: str, tensor: Union[numpy.ndarray, torch.Tensor], *, max_to_keep: Optional[int] = None) → None

add_scalar(name: str, scalar: Union[int, float, numpy.integer, numpy.floating], *, max_to_keep: Optional[int] = None) → None

items() → Iterable[Tuple[str, Deque[Tuple[int, Any]]]]

keys() → Iterable[str]

set_trainer(trainer: None) → None

values() → Iterable[Deque[Tuple[int, Any]]]

torchpack.train.trainer module

class torchpack.train.trainer.Trainer

Bases: object

Base class for a trainer.

after_epoch() → None

after_step(output_dict: Dict[str, Any]) → None

after_train() → None

before_epoch() → None

before_step(feed_dict: Dict[str, Any]) → None

before_train() → None

load_state_dict(state_dict: Dict[str, Any]) → None

run_step(feed_dict: Dict[str, Any]) → Dict[str, Any]

state_dict() → Dict[str, Any]

train(dataflow: torch.utils.data.dataloader.DataLoader, *, num_epochs: int = 9999999, callbacks: Optional[List[torchpack.callbacks.callback.Callback]] = None) → None

train_with_defaults(dataflow: torch.utils.data.dataloader.DataLoader, *, num_epochs: int = 9999999, callbacks: Optional[List[torchpack.callbacks.callback.Callback]] = None) → None

trigger_epoch() → None

trigger_step() → None

Module contents