VSIM
VSIM key (ELE | FP32 | VALUES num) (vector | element) [WITHSCORES] [COUNT num] [EF search-exploration-factor] [FILTER expression] [FILTER-EF max-filtering-effort] [TRUTH] [NOTHREAD]
- Available since:
- Redis CE 8.0.0
- Time complexity:
- O(log(N)) where N is the number of elements in the vector set.
Return elements similar to a given vector or element. Use this command to perform approximate or exact similarity searches within a vector set.
You can query using either a vector (via FP32 or VALUES num) or by referencing another element (using ELE). Optional parameters let you control the search behavior, such as score output, result count, and filtering options.
VSIM word_embeddings ELE apple
1) "apple"
2) "apples"
3) "pear"
4) "fruit"
5) "berry"
6) "pears"
7) "strawberry"
8) "peach"
9) "potato"
10) "grape"
You can include similarity scores and limit the number of results:
VSIM word_embeddings ELE apple WITHSCORES COUNT 3
1) "apple"
2) "0.9998867657923256"
3) "apples"
4) "0.8598527610301971"
5) "pear"
6) "0.8226882219314575"
Set the EF (exploration factor) to improve recall at the cost of performance. Use the TRUTH option to perform an exact linear scan, useful for benchmarking. The NOTHREAD option runs the search in the main thread and may increase server latency.
Required arguments
key
is the name of the key that holds the vector set data.
ELE | FP32 | VALUES num
specifies how the input vector is provided. Use ELE to refer to an existing element, FP32 for binary float format, or VALUES num for a list of stringified float values.
vector or element
is either the vector data (for FP32 or VALUES) or the name of the element (for ELE) to use as the similarity reference.
Optional arguments
WITHSCORES
returns the similarity score (from 1 to 0) alongside each result. A score of 1 is identical; 0 is the opposite.
COUNT num
limits the number of returned results to num.
EF search-exploration-factor
controls the search effort. Higher values explore more nodes, improving recall at the cost of speed. Typical values range from 50 to 1000.
FILTER expression
applies a filter expression to restrict matching elements. See the filtered search section for syntax details.
FILTER-EF max-filtering-effort
limits the number of filtering attempts for the FILTER expression. See the filtered search section for more.
TRUTH
forces an exact linear scan of all elements, bypassing the HNSW graph. Use for benchmarking or to calculate recall. This is significantly slower (O(N)).
NOTHREAD
executes the search in the main thread instead of a background thread. Useful for small vector sets or benchmarks. This may block the server during execution.
Related topics
RESP2 Reply
One of the following:
- Simple error reply for unknown element.
- Array reply (empty array) for unknown key.
- Array reply with matching elements; juxtaposed with scores when used with the WITHSCORES option.
RESP3 Reply
One of the following:
- Simple error reply for unknown element.
- Array reply (empty array) for unknown key.
- Array reply with matching elements.
- Map reply with matching elements and double scores when used with the WITHSCORES option.