metainfo.py

# Copyright (c) Alibaba, Inc. and its affiliates.


def get_member_set(_class):
    """Get member names set."""
    return set(
        getattr(_class, _a)
        for _a in dir(_class)
        if not _a.startswith('_') and isinstance(getattr(_class, _a), str)
    )


class Tasks:
    """Names for different tasks"""

    word_segmentation = 'word-segmentation'
    part_of_speech = 'part-of-speech'
    named_entity_recognition = 'named-entity-recognition'
    relation_extraction = 'relation-extraction'
    entity_typing = 'entity-typing'
    pretraining = 'pretraining'


class Metrics:
    """Names for different metrics"""

    ner_metric = 'ner-metric'  # alias of sequence-labeling-metric
    sequence_labeling_metric = 'sequence-labeling-metric'
    span_extraction_metric = 'span-extraction-metric'
    relation_extraction_metric = 'relation-extraction-metric'
    typing_metric = 'typing-metric'
    pretraining_metric = 'pretraining-metric'
    typing_threshold_metric = 'typing-threshold-metric'


class Models:
    """Names for different models"""

    sequence_labeling_model = 'sequence-labeling-model'
    global_pointer_model = 'global-pointer-model'
    biaffine_ner_model = 'biaffine-ner-model'
    relation_extraction_model = 'relation-extraction-model'
    multilabel_concat_typing_model = 'multilabel-concat-typing-model'
    multilabel_span_typing_model = 'multilabel-span-typing-model'
    multilabel_concat_typing_model_mcce_s = 'multilabel-concat-typing-model-mcce-s'
    pretraining_model = 'pretraining-model'
    twostage_ner_model = 'twostage-ner-model'
    w2ner_model = 'w2ner-model'


class Embedders:
    """Names for different embedders"""

    embedding = 'embedding'
    transformer_embedder = 'transformer-embedder'


class Encoders:
    """Names for different encoders"""

    span_encoder = 'span-encoder'
    gru_encoder = 'gru'
    rnn_encoder = 'rnn'
    lstm_encoder = 'lstm'
    cnn_encoder = 'cnn'


class Decoders:
    """Names for different decoders"""

    crf = 'crf'
    partial_crf = 'partial-crf'
    pairwise_crf = 'pairwise-crf'
    linear = 'linear'
    constrained_crf = 'constrained-crf'


class Preprocessors:
    """Names for different preprocessors"""

    nlp_preprocessor = 'nlp-preprocessor'
    sequence_labeling_preprocessor = 'sequence-labeling-preprocessor'
    relation_extraction_preprocessor = 'relation-extraction-preprocessor'
    span_extraction_preprocessor = 'span-extraction-preprocessor'
    multilabel_span_typing_preprocessor = 'multilabel-span-typing-preprocessor'
    multilabel_concat_typing_preprocessor = 'multilabel-concat-typing-preprocessor'
    multilabel_concat_typing_mcce_preprocessor = 'multilabel-concat-typing-mcce-preprocessor'
    pretraining_preprocessor = 'pretraining-preprocessor'
    twostage_preprocessor = 'twostage-preprocessor'
    word_extraction_preprocessor = 'word-extraction-preprocessor'


class DataCollators:
    """Names for different data_collators"""

    data_collator_with_padding = 'DataCollatorWithPadding'
    sequence_labeling_data_collator = 'SequenceLabelingDataCollatorWithPadding'
    span_extraction_data_collator = 'SpanExtractionDataCollatorWithPadding'
    multi_label_span_typing_data_collator = 'MultiLabelSpanTypingDataCollatorWithPadding'
    multi_label_concat_typing_data_collator = 'MultiLabelConcatTypingDataCollatorWithPadding'
    pretraining_data_collator = 'PretrainingDataCollatorWithPadding'
    twostage_data_collator = 'TwostageDataCollatorWithPadding'
    word_extraction_data_collator = 'WordExtractionDataCollatorWithPadding'


class Trainers:
    """Names for different trainers"""

    default_trainer = 'default-trainer'
    typing_trainer = 'typing-trainer'


class Optimizers:
    """Names for different optimizers"""

    pass


class LR_Schedulers:
    """Names for different lr_schedulers"""

    pass


class Hooks:
    """Names for different hooks"""

    pass


class DatasetDumpers:
    """Names for different dataset dumpers"""

    ner_dumper = 'ner-dumper'


class Pipelines:
    """Names for different pipelines"""

    sequence_labeling_pipeline = 'sequence-labeling-pipeline'
    span_based_ner_pipeline = 'span-based-ner-pipeline'