From c662ab0cf412f6ca59ee6ab8cb59d914fca0f321 Mon Sep 17 00:00:00 2001
From: Nathan Hansen <9289200+nhansendev@users.noreply.github.com>
Date: Tue, 25 Mar 2025 14:27:25 -0500
Subject: [PATCH] Update _bertopic.py to address #2316

Added "embeddings" kwarg to update_topics for use when pre-calculated document embeddings are used. Addresses issue #2316
---
 bertopic/_bertopic.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/bertopic/_bertopic.py b/bertopic/_bertopic.py
index 6a75171b..d512d1fd 100644
--- a/bertopic/_bertopic.py
+++ b/bertopic/_bertopic.py
@@ -1488,6 +1488,7 @@ def update_topics(
         vectorizer_model: CountVectorizer = None,
         ctfidf_model: ClassTfidfTransformer = None,
         representation_model: BaseRepresentation = None,
+        embeddings: np.ndarray = None,
     ):
         """Updates the topic representation by recalculating c-TF-IDF with the new
         parameters as defined in this function.
@@ -1514,6 +1515,7 @@ def update_topics(
             representation_model: Pass in a model that fine-tunes the topic representations
                                   calculated through c-TF-IDF. Models from `bertopic.representation`
                                   are supported.
+            embeddings: Pre-trained document embeddings.
 
         Examples:
         In order to update the topic representation, you will need to first fit the topic
@@ -1586,7 +1588,12 @@ def update_topics(
             if same_position and -1 not in topics and -1 in self.topics_:
                 self.topic_embeddings_ = self.topic_embeddings_[1:]
             else:
-                self._create_topic_vectors()
+                if embeddings is not None:
+                    # Use provided embeddings
+                    self._create_topic_vectors(documents=documents, embeddings=embeddings)
+                else:
+                    # Use self.embedding_model to calculate embeddings
+                    self._create_topic_vectors()
 
     def get_topics(self, full: bool = False) -> Mapping[str, Tuple[str, float]]:
         """Return topics with top n words and their c-TF-IDF score.