Reputation: 31
I used to use ChromaDB, now I switched to PGVector.
In ChromaDB there was an option to get the required amount of documents using a filter by metadata, but I can't find this in PGVector.
db = Chroma.from_documents(
documents=documents,
embedding=OpenAIEmbeddings(),
collection_name='test',
persist_directory=settings.PERSIST_DIR
)
doc = db.get(
where={
'file_id': file_id
}
)
print(doc['documents'])
db = PGVector.from_documents(
documents=documents,
embedding=OpenAIEmbeddings(),
collection_name='test',
connection_string=PGVector.connection_string_from_db_params(
driver='psycopg',
host='***',
port=1234,
database='***',
user='***',
password='***'
)
)
Any ideas?
I tried to get it this way, but I don't understand how to display the result, if this is possible. But it's also worth considering that there is a default limit of 4 documents.
retriever = db.as_retriever(
search_kwargs={
'filter': {
'file_id': file_id
}
}
)
Upvotes: 2
Views: 607