sgnlp.models.ufd.eval.test_ufd

test_ufd(test_data: List, adaptor_global: sgnlp.models.ufd.modeling.UFDAdaptorGlobalModel, adaptor_domain: sgnlp.models.ufd.modeling.UFDAdaptorDomainModel, maper: sgnlp.models.ufd.modeling.UFDCombineFeaturesMapModel, model: sgnlp.models.ufd.modeling.UFDClassifierModel, criterion: torch.nn.modules.loss.CrossEntropyLoss, cfg: sgnlp.models.ufd.data_class.UFDArguments) → float

Test step method for UFD evaluation

Parameters

• test_data (List) – test dataset embeddings

• adaptor_global (UFDAdaptorGlobalModel) – adaptor global model

• adaptor_domain (UFDAdaptorDomainModel) – adaptor domain model

• maper (UFDCombineFeaturesMapModel) – combine features map model

• model (UFDClassifierModel) – classifier model

• criterion (torch.nn.CrossEntropyLoss) – cross entropy loss criterion

• cfg (UFDArguments) – UFDArguments config load from configuration file

Returns

loss calculated for the test step

Return type

float