Biaffine dependency parsing (:cite:`dozat:17a`).

r"""Implementation of "Stanford's graph-based neural dependency parser at the conll 2017 shared task" (:cite:`dozat2017stanford`) and "Establi

Dropout on timesteps with bernoulli distribution

Dropout on the first two dimensions

Parameters ---------- prediction : numpy.ndarray a predicted dependency tree where prediction[dep_idx] = head_id

An implementation of T. Dozat and C. D. Manning, “Deep Biaffine Attention for Neural Dependency Parsing.,” ICLR, 2017. Although I have my

A classifier using transformer as encoder. Args: **kwargs: Passed to config.

A simple Semantic Textual Similarity (STS) baseline which fine-tunes a transformer with a regression layer on top of it. Arg

An implementation of "Jointly Predicting Predicates and Arguments in Neural Semantic Role Labeling" (:cite:`he-etal-2018-jointly`). It generat

A span based Semantic Role Labeling task using BIO scheme for tagging the role of each token. Given a predicate and a token, it uses biaffine

A multi-task learning (MTL) framework. It shares the same encoder across multiple decoders. These decoders can have dependencies on each othe

r"""Implementation of "Stanford's graph-based neural dependency parser at the conll 2017 shared task" (:cite:`dozat2017stanford`) and "Establi

A simple tagger using a linear layer with an optional CRF (:cite:`lafferty2001conditional`) layer for any tagging tasks including PoS tagging

r"""Two-stage CRF Parsing (:cite:`ijcai2020-560`). Args: trn: Path to training set. dev: Path to dev set.

A task in the multi-task learning framework Args: trn: Path to training set. dev: Path to dev set.

Biaffine dependency parsing (:cite:`dozat:17a`). Args: trn: Path to training set. dev: Path to dev set. t

r"""Universal Dependencies Parsing (lemmatization, features, PoS tagging and dependency parsing) implementation of "75 Languages, 1 Model: Par

Transition based lemmatization (:cite:`kondratyuk-straka-2019-75`). Args: trn: Path to training set. dev: Path to de

An implementation of Named Entity Recognition as Dependency Parsing (:cite:`yu-etal-2020-named`). It treats every possible span as a candidate

r"""A simple tagger using a linear layer with an optional CRF (:cite:`lafferty2001conditional`) layer for NER task. It can utilize whitelist g

A span based Semantic Role Labeling task using BIO scheme for tagging the role of each token. Given a predicate and a token, it uses biaffine

r""" An implementation of "Jointly Predicting Predicates and Arguments in Neural Semantic Role Labeling" (:cite:`he-etal-2018-jointly`). It ge

Tokenization which casts a chunking problem into a tagging problem. This task has to create batch of tokens containing both [CLS] and [SEP] si

Convention for the data. When read_buffer is not None, it represents contents of the file from `read_position` onwards that has n

Constructor for FileReadBackwards. Args: path: Path to the file to be read encoding (str): Encoding chunk

Constructor for FileReadBackwardsIterator Args: fp (File): A file that we wish to start reading backwards from encodi

r""" For :math:`N` tensors, they use different dropout masks respectively. When :math:`N-M` of them are dropped, the remaining :math:`

An embedding layer for fastText (:cite:`bojanowski2017enriching`). Args: key: Field name. embedding_dim: Size of this

An embedding layer builder for fastText (:cite:`bojanowski2017enriching`). Args: src: Field name. filepath: Filepath

Toy example of Word2VecEmbedding. It simply returns the embedding of a given word Args: **kwargs: