Redis.Commands.VectorSet (Redis v0.8.0)

Command builders for Redis Vector Set operations (Redis 8.0+).

Vector Sets are a data type for approximate nearest-neighbor similarity search. Each element is associated with a high-dimensional vector, and the set supports efficient queries for the most similar elements.

Every function returns a plain list of strings (a command). To execute a command, pass the result to Redis.command/2; to batch several commands in a single round trip, use Redis.pipeline/2.

Examples

Adding elements with vectors:

iex> Redis.command(conn, Redis.Commands.VectorSet.vadd("vs", "item1", [1.0, 2.0, 3.0]))
{:ok, 1}

Searching for similar elements:

iex> Redis.command(conn, Redis.Commands.VectorSet.vsim("vs", {:vector, [1.0, 2.0, 3.0]}, count: 5))
{:ok, ["item1", "item2"]}

Getting element embeddings:

iex> Redis.command(conn, Redis.Commands.VectorSet.vemb("vs", "item1"))
{:ok, [1.0, 2.0, 3.0]}

Summary

Functions

vadd(key, element, vector, opts \\ [])

Builds a VADD command to add an element with a vector to the vector set at key.

vcard(key)

Builds a VCARD command to return the number of elements in the vector set at key.

vdim(key)

Builds a VDIM command to return the vector dimensions of the vector set at key.

vemb(key, element, opts \\ [])

Builds a VEMB command to get the embedding vector of an element.

vgetattr(key, element)

Builds a VGETATTR command to get the JSON attributes of an element.

vinfo(key)

Builds a VINFO command to return information about the vector set at key.

vlinks(key, element, opts \\ [])

Builds a VLINKS command to get the graph links of an element.

vrandmember(key, opts \\ [])

Builds a VRANDMEMBER command to return one or more random elements.

vrem(key, element)

Builds a VREM command to remove an element from the vector set at key.

vsetattr(key, element, json)

Builds a VSETATTR command to set JSON attributes on an element.

vsim(key, search_input, opts \\ [])

Builds a VSIM command for similarity search.

Functions

vadd(key, element, vector, opts \\ [])

@spec vadd(String.t(), String.t(), [float()], keyword()) :: [String.t()]

Builds a VADD command to add an element with a vector to the vector set at key.

The vector is provided as a list of floats and serialized using the VALUES count v1 v2 ... format.

Options

:reduce - reduce dimensionality to the given integer
:quantization - quantization type: :q8, :bin, or :noquant
:ef - exploration factor for insertion
:setattr - JSON string of attributes to associate with the element
:cas - compare-and-swap value for conditional updates

vcard(key)

@spec vcard(String.t()) :: [String.t()]

Builds a VCARD command to return the number of elements in the vector set at key.

vdim(key)

@spec vdim(String.t()) :: [String.t()]

Builds a VDIM command to return the vector dimensions of the vector set at key.

vemb(key, element, opts \\ [])

@spec vemb(String.t(), String.t(), keyword()) :: [String.t()]

Builds a VEMB command to get the embedding vector of an element.

Options

:raw - return raw binary representation

vgetattr(key, element)

@spec vgetattr(String.t(), String.t()) :: [String.t()]

Builds a VGETATTR command to get the JSON attributes of an element.

vinfo(key)

@spec vinfo(String.t()) :: [String.t()]

Builds a VINFO command to return information about the vector set at key.

vlinks(key, element, opts \\ [])

@spec vlinks(String.t(), String.t(), keyword()) :: [String.t()]

Builds a VLINKS command to get the graph links of an element.

Options

:withscores - include link scores in the reply

vrandmember(key, opts \\ [])

@spec vrandmember(
  String.t(),
  keyword()
) :: [String.t()]

Builds a VRANDMEMBER command to return one or more random elements.

Options

:count - number of random elements to return

vrem(key, element)

@spec vrem(String.t(), String.t()) :: [String.t()]

Builds a VREM command to remove an element from the vector set at key.

vsetattr(key, element, json)

@spec vsetattr(String.t(), String.t(), String.t()) :: [String.t()]

Builds a VSETATTR command to set JSON attributes on an element.

vsim(key, search_input, opts \\ [])

@spec vsim(String.t(), {:element, String.t()} | {:vector, [float()]}, keyword()) :: [
  String.t()
]

Builds a VSIM command for similarity search.

The search_input is either {:element, name} to search by an existing element, or {:vector, [floats]} to search by a raw vector.

Options

:count - maximum number of results
:ef - exploration factor for search
:filter - FILTER expression string for attribute-based filtering
:filter_ef - exploration factor for filtered search
:truth - use brute-force exact search (boolean)
:nothread - disable multi-threading (boolean)
:withscores - include similarity scores in the reply (boolean)