lightopic.lightbertopic
=======================

.. py:module:: lightopic.lightbertopic

.. autoapi-nested-parse::

   A simple child class of bertopic.BERTopic to enable using Lightopic.

   The only addition to the base class is a method to `save_lightopic`:
   this will serialise a model in a format suitable for instantiating a
   Lightopic model with Lightopic.load. As part of this process it will
   train a reduced UMAP model, suitable for projecting down to two
   dimensions for plotting.

   Typical usage example:
   ```python
   from lightopic import LightBERTopic
   from sklearn.datasets import fetch_20newsgroups

   docs = fetch_20newsgroups(subset="all")["data"]
   topic_model = LightBERTopic()
   topics, probabilities = topic_model.fit_transform(docs)
   topic_model.save_lightopic("lightopic_model")
   ```



Classes
-------

.. autoapisummary::

   lightopic.lightbertopic.LightBERTopic


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

.. py:class:: LightBERTopic

   Bases: :py:obj:`bertopic.BERTopic`


   Child class of bertopic.BERTopic to enable serialising models in
   the Lightopic format.

   All methods from bertopic.BERTopic are left unchanged; we add only
   the save_lightopic method.


   .. py:method:: save_lightopic(output_path: str) -> None

      Serialise a trained model for use as a Lightopic model.

      :param output_path: Directory path to serialise the model,
                          will be created if it does not exist.
      :type output_path: str