Categories:: Vector functions

VECTOR_TRUNCATE¶

Truncates a VECTOR to a smaller dimension.

This function can also be called through the alias VECTOR_TRUNC.

See also:: Vector Embeddings, VECTOR_NORMALIZE

Syntax¶

VECTOR_TRUNCATE( <vector>, <dimension> )

Copy

Arguments¶

vector: A single VECTOR value to truncate.
dimension: The number of elements that should be in the returned vector.

Returns¶

Returns a VECTOR value with the same values and types for the first dimension entries, with the remainder discarded.

Usage notes¶

Returns NULL when any input is NULL.
Using a dimension larger than the number of dimensions in the vector causes an error.
Truncated vectors are not normalized.

Examples¶

This example demonstrates truncating a 3-dimensional vector into a 2-dimensional vector:

SELECT VECTOR_TRUNCATE([1, 2, 3]::VECTOR(INT, 3), 2);

Copy

[1,2]

This example demonstrates truncating a vector produced by EMBED_TEXT_768 for the text “Analytical databases are typically column-oriented rather than row-oriented” with the snowflake-arctic-embed-m-v1.5 model from 768 elements to 256 elements:

SELECT VECTOR_TRUNCATE(
    SNOWFLAKE.CORTEX.EMBED_TEXT_768(
        'snowflake-arctic-embed-m-v1.5',
        'Analytical databases are typically column-oriented rather than row-oriented'
    ),
    256)
;

Copy