primeqa.qg.models.passage_qg.answer_sampler.AnswerSampler

class primeqa.qg.models.passage_qg.answer_sampler.AnswerSampler

Bases: object

Class for sampling answer tokens from passage, to be used for PassageQG inference.

Methods

create_qg_input	create the input for qg training: If the answers are provided in answers_list, use them.
detect_language	detect the language of an input passage, used later for identifying the right NER model to use
get_named_entities	pick the right NER model to use based on detected language of the passage

create_qg_input(data_list, num_questions_per_instance=5, answers_list=[], id_list=[])

create the input for qg training: If the answers are provided in answers_list, use them. Otherwise sampling named entities as possible answers from text passages.

detect_language(text)

detect the language of an input passage, used later for identifying the right NER model to use

get_named_entities(text)

pick the right NER model to use based on detected language of the passage