Feedback: Vector search

[vguptaa45]

Not able to extract more than 10 documents in vector search. Tried replacing <top_k> multiple times with a higher number.

Page https://redis.io/docs/latest/develop/interact/search-and-query/query/vector-search

[dwdougherty]

Hello @vguptaa45. Thank you for reaching out. There was an internal discussion about this issue, and one of our developer leads suggests adding LIMIT 0 15 to your query, as the default number of returned results is limited to ten. Support will be able assist further.

[vguptaa45]

Hi. Getting the following error:

"Vector search all failed: Syntax error at offset 49 near LIMIT"

Have i implemented it correctly:

` # 1. Convert query vector
query_vector_np = np.array(query_vector, dtype=np.float16)
logger.info(f"1. Query vector prepared: shape={query_vector_np.shape}")

        # 2. Prepare search query without company filter
        query = f'*=>[KNN {k} @vector $query_vector AS vector_score LIMIT 0 15]'
        logger.info(f"2. Search query: {query}")

        # 3. Execute search
        logger.info("3. Executing search...")
        raw_results = self.redis_client_binary.ft(self.index_name).search(
            Query(query)
                .return_fields("content", "metadata", "vector_score", "compressed")
                .dialect(2),
            {'query_vector': query_vector_np.tobytes()}
        )
        logger.info(f"4. Raw search results: {raw_results.total} matches")

`

[alonre24]

Let me rephrase - to get the first 15 results using redis-py, you will have to use the following query structure:

Query(query)
                .return_fields("content", "metadata", "vector_score", "compressed").paging(0, 15)
                .dialect(2)

Let me know if it works for you @vguptaa45

[vguptaa45]

That worked! Thanks you for the help.

Would appreciate if this is clearely added to the docs and the website.

Have a great day

[dwdougherty]

Noted. Thanks again.