You can have multiple vector representations for the text you want to perform semantic search on. A vector representation is associated one-to-one with a vectorization definition, and you can have a different vector representation by defining a new vectorization. In the vectorization definition, you can specify the embedding model to be used, etc. If you do not need the previous vector representation, delete the vectorization definition.

Example) Defining a new vectorizer and deleting the old one

rag_database=> SELECT ai.drop_vectorizer(pgx_vectorizer.get_vectorizer_id(view_name => 'sample_embeddings'), drop_all => true); 

rag_database=> SELECT ai.create_vectorizer(…);