changing the embedding model to gte-large to match that of the tesla_db database (1024 embedding dimensions)... embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-large')

app.py

@@ -65,10 +65,22 @@ client = OpenAI(
 )
 #---------------------------------------------------------------------
 
-embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-small')
+# embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-small')  
+# The gte-small model from OpenAI's family of models, which includes the GTE models designed for retrieval tasks, uses a specific number of embedding dimensions. The gte-small model has 384 embedding dimensions.
+# This dimensionality allows the model to capture semantic information effectively while maintaining a relatively small model size for efficiency in retrieval tasks.
+
+# However the vector database was encoded with model 'gte-large' which has 1024 embedding dimensions, so we need to use gte-large model here to match the embedding dimensions of the tesla_db vector database
+# otherwise we get the following runtime error: "chromadb.errors.InvalidDimensionException: Embedding dimension 384 does not match collection dimensionality 1024"
+embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-large')
+
 
 tesla_10k_collection = 'tesla-10k-2019-to-2023'
 
+# Example: Creating a collection with the correct dimensionality
+# tesla_10k_collection = Chroma.create_collection("tesla-10k-2019-to-2023", embedding_dim=384)
+
+
+
 # vector database constructor Chroma()
 vectorstore_persisted = Chroma(
     collection_name=tesla_10k_collection,