pytorch_pfn_extras.engine.Evaluator#

class pytorch_pfn_extras.engine.Evaluator(handler, models, *, progress_bar=False, metrics=None, profile=None)#

Bases: object

Methods

__init__(handler, models, *[, progress_bar, ...])

run(loader, *[, eval_len])

Executes the evaluation loop.
Parameters:

  • handler (BaseHandler) –

  • models (Union[Module, Mapping[str, Module]]) –

  • progress_bar (bool) –

  • metrics (Optional[Sequence[MetricType]]) –

  • profile (Optional[profile]) –

__init__(handler, models, *, progress_bar=False, metrics=None, profile=None)#
Parameters:

  • handler (BaseHandler) –

  • models (Union[Module, Mapping[str, Module]]) –

  • progress_bar (bool) –

  • metrics (Optional[Sequence[MetricType]]) –

  • profile (Optional[profile]) –

run(loader, *, eval_len=None)#

Executes the evaluation loop.

Parameters:

  • loader (torch.utils.data.DataLoader) – A data loader for evaluation.

  • eval_len (int, optional) – The number of iterations per one evaluation epoch.

Return type:

None