The Question Answering module within AdaptNLP

class QACallback[source]

QACallback(xmodel_instances, features) :: Callback

Basic Question Answering Data Callback

Parameters:

  • xmodel_instances : <class 'inspect._empty'>

  • features : <class 'inspect._empty'>

class QAResult[source]

QAResult(examples:List[SquadExample], top_predictions:Union[str, OrderedDict], all_nbest_json:List[OrderedDict], n_best_size:int)

A result class designed for Question Answering models

Parameters:

  • examples : typing.List[transformers.data.processors.squad.SquadExample]

  • top_predictions : typing.Union[str, collections.OrderedDict]

  • all_nbest_json : typing.List[collections.OrderedDict]

  • n_best_size : <class 'int'>

class TransformersQuestionAnswering[source]

TransformersQuestionAnswering(tokenizer:PreTrainedTokenizer, model:PreTrainedModel) :: AdaptiveModel

Adaptive Model for Transformers Question Answering Model

Parameters

  • tokenizer - A tokenizer object from Huggingface's transformers (TODO)and tokenizers *
  • model - A transformer Question Answering model

Parameters:

  • tokenizer : <class 'transformers.tokenization_utils.PreTrainedTokenizer'>

  • model : <class 'transformers.modeling_utils.PreTrainedModel'>

class EasyQuestionAnswering[source]

EasyQuestionAnswering()

Question Answering Module

Usage:

>>> qa = adaptnlp.EasyQuestionAnswering()
>>> qa.predict_qa(query='What is life?', context='Life is NLP.', n_best_size=5, mini_batch_size=1)