docs: add FloatVector overview example

Author: nightcityblade

docs/overview.rst

@@ -374,8 +374,8 @@ Extension types
 ---------------
 
 In the :ref:`example SQLAlchemy table definition <table-definition>` above, we
-are making use of the two extension data types that the CrateDB SQLAlchemy
-dialect provides.
+are making use of extension data types provided by the CrateDB SQLAlchemy
+dialect.
 
 .. SEEALSO::
 
@@ -459,6 +459,35 @@ The resulting object will look like this:
     change, the :ref:`UPDATE <crate-reference:dml-updating-data>` statement
     sent to CrateDB will include all of the ``ObjectArray`` data.
 
+.. _floatvector:
+
+``FloatVector``
+...............
+
+Use ``FloatVector`` to store fixed-length floating-point vectors, such as
+embeddings used for similarity search. Pass the vector dimension to the type
+when defining the column:
+
+    >>> from sqlalchemy_cratedb import FloatVector, knn_match
+
+    >>> class SearchIndex(Base):
+    ...     __tablename__ = 'search_index'
+    ...     name = sa.Column(sa.String, primary_key=True)
+    ...     embedding = sa.Column(FloatVector(3))
+
+Values can be supplied as lists of floating-point numbers. To find nearby
+vectors, use ``knn_match`` in a query:
+
+    >>> item = SearchIndex(name='example', embedding=[1.0, 2.0, 3.0])
+    >>> session.add(item)
+    >>> session.commit()
+    >>> query = session.query(SearchIndex.name).filter(
+    ...     knn_match(SearchIndex.embedding, [1.0, 2.0, 2.9], 10)
+    ... )
+
+See the :doc:`vector type guide <working-with-types>` for a complete example,
+including storing and retrieving vectors.
+
 .. _geopoint:
 .. _geoshape: