riva/proto/riva_nlp_pipeline.proto

message ClassScore

string name

Class label

float score

Score

message TextClassificationInput

string sentence

Sentence to classify

string classes(repeated)

Possible class labels to classify the sequence into

string hypothesis_template

The template used to turn each label into an NLI-style hypothesis. Template must include a {}. Default: “This example is {}.”

bool multi_class

Whether or not multiple candidate labels can be true. If False, then the scores are normalized such that the sum of the label likelihoods for each sequence is 1. If True, then the labels are considered independent and probabilities are normalized for each candidate by doing a softmax of the entailment score vs. the contradiction score.

message TextClassificationOutput

string sentence

Sentence which was classified

ClassScore scores(repeated)

Class scores

message TransformTextInput

string sentence

message TransformTextOutput

string sentence