ez_transfer.postprocessors

Classifcation Postprocessor

class easytransfer.postprocessors.classification_postprocessors.ClassificationPostprocessor(label_enumerate_values, output_schema, thread_num=None, input_queue=None, output_queue=None, prediction_colname='predictions', job_name='CLSpostprocessor')

Postprocessor for text classification, convert label_id to the label_name

process(in_data)

Post-process the model outputs

Parameters:in_data (dict) -- a dict of model outputs Returns:a dict of post-processed model outputs Return type:ret (dict)

Sequence Labeling Postprocessor

class easytransfer.postprocessors.labeling_postprocessors.LabelingPostprocessor(label_enumerate_values, output_schema, thread_num=None, input_queue=None, output_queue=None, prediction_colname='predictions', job_name='LabelingPostprocessor')

Postprocessor for sequence labeling, merge the sub-tokens and output the tag for each word

process(in_data)

Post-process the model outputs

Parameters:in_data (dict) -- a dict of model outputs Returns:a dict of post-processed model outputs Return type:ret (dict)

Text Comprehension Postprocessor

class easytransfer.postprocessors.comprehension_postprocessors.ComprehensionPostprocessor(output_schema, n_best_size=20, max_answer_length=30, thread_num=None, input_queue=None, output_queue=None, job_name='ComprehentionPostprocessor')

Postprocessor for text comprehension, search and improve the answer span

process(in_data)

Post-process the model outputs

Parameters:in_data (dict) -- a dict of model outputs Returns:a dict of post-processed model outputs Return type:ret (dict)