extorch.utils.stats

class extorch.utils.stats.AverageMeter

Bases: object

is_empty() → bool

reset() → None

update(val: Union[float, int], n: int = 1) → None

class extorch.utils.stats.OrderedStats

Bases: object

avgs() → Optional[collections.OrderedDict]

items()

update(stats, n: int = 1) → None

extorch.utils.stats.accuracy(outputs: torch.Tensor, targets: torch.Tensor, topk: Tuple[int] = (1,)) → List[torch.Tensor]

extorch.utils.stats.get_params(model: torch.nn.modules.module.Module, only_trainable: bool = False) → int

Get the parameter number of the model.

Parameters

only_trainable (bool) – If only_trainable is true, only trainable parameters will be counted.