:py:mod:`medcat.rel_cat`
========================

.. py:module:: medcat.rel_cat


Module Contents
---------------

Classes
~~~~~~~

.. autoapisummary::

   medcat.rel_cat.RelCAT




.. py:class:: RelCAT(cdb, tokenizer, config = ConfigRelCAT(), task='train', init_model=False)


   Bases: :py:obj:`medcat.pipeline.pipe_runner.PipeRunner`

   The RelCAT class used for training 'Relation-Annotation' models, i.e., annotation of relations
    between clinical concepts.

   :param cdb: cdb, this is used when creating relation datasets.
   :type cdb: CDB
   :param tokenizer: The Huggingface tokenizer instance. This can be a pre-trained tokenzier instance from
                     a BERT-style model. For now, only BERT models are supported.
   :type tokenizer: TokenizerWrapperBERT
   :param config: the configuration for RelCAT. Param descriptions available in ConfigRelCAT docs.
   :type config: ConfigRelCAT
   :param task: What task is this model supposed to handle. Defaults to "train"
   :type task: str, optional
   :param init_model: loads default model. Defaults to False.
   :type init_model: bool, optional

   .. py:attribute:: name
      :value: 'rel_cat'

      

   .. py:attribute:: log

      

   .. py:method:: __init__(cdb, tokenizer, config = ConfigRelCAT(), task='train', init_model=False)


   .. py:method:: save(save_path)

      Saves model and its dependencies to specified save_path folder.
          The CDB is obviously not saved, it is however necessary to save the tokenizer used.

      :param save_path: folder path in which to save the model & deps.
      :type save_path: str


   .. py:method:: _get_model()

      Used only for model initialisation.



   .. py:method:: load(load_path = './')
      :classmethod:


   .. py:method:: _create_test_train_datasets(data, split_sets = False)


   .. py:method:: train(export_data_path = '', train_csv_path = '', test_csv_path = '', checkpoint_path = './')


   .. py:method:: evaluate_(output_logits, labels, ignore_idx)


   .. py:method:: evaluate_results(data_loader, pad_id)


   .. py:method:: pipe(stream, *args, **kwargs)


   .. py:method:: __call__(doc)