:mod:`capreolus.reranker.TK`
============================

.. py:module:: capreolus.reranker.TK


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

Classes
~~~~~~~

.. autoapisummary::

   capreolus.reranker.TK.PositionalEncoding
   capreolus.reranker.TK.TK_class
   capreolus.reranker.TK.TK



.. data:: logger
   

   

.. py:class:: PositionalEncoding(d_model, dropout=0.1, max_len=5000)

   Bases: :class:`torch.nn.Module`

   .. method:: forward(self, x)



.. py:class:: TK_class(extractor, config)

   Bases: :class:`torch.nn.Module`

   Adapted from https://github.com/sebastian-hofstaetter/transformer-kernel-ranking/blob/master/matchmaker/models/tk.py
   TK is a neural IR model - a fusion between transformer contextualization & kernel-based scoring
   -> uses 1 transformer block to contextualize embeddings
   -> soft-histogram kernels to score interactions

   .. method:: get_mu_matrix(self, extractor)

      Returns a matrix of mu values that can be directly subtracted from the cosine matrix.
      This is the matrix mu in equation 5 in the paper (https://arxiv.org/pdf/2002.01854.pdf)


   .. method:: get_mask(self, embedding)

      Gets a mask of shape (seq_len, seq_len). This is an additive mask, hence masked elements should be -inf


   .. method:: get_embedding(self, toks)

      Overrides KNRM_Class's get_embedding to return contextualized word embeddings


   .. method:: forward(self, doctoks, querytoks, query_idf)



.. py:class:: TK

   Bases: :class:`capreolus.reranker.Reranker`

   .. attribute:: module_name
      :annotation: = TK

      

   .. attribute:: description
      :annotation: = Sebastian Hofstätter, Markus Zlabinger, and Allan Hanbury. 2019.
                        TU Wien @ TREC Deep Learning '19 -- Simple Contextualization for Re-ranking. In TREC '19.

      

   .. attribute:: config_spec
      

      

   .. method:: build_model(self)


   .. method:: score(self, d)


   .. method:: test(self, d)