EMBED_TEXT_768 (SNOWFLAKE.CORTEX) | Snowflake Documentation (original) (raw)

Note

AI_EMBED is the latest version of this function. Use AI_EMBED for the latest functionality. You can continue to use EMBED_TEXT_768 (SNOWFLAKE.CORTEX).

Creates a vector embedding of 768 dimensions from English-language text.

Syntax

SNOWFLAKE.CORTEX.EMBED_TEXT_768( , )

Arguments

_model_

A string specifying the vector embedding model to be used to generate the embedding. This must be one of the following values.

Supported models might have different costs.

_text_

The text for which an embedding should be calculated.

Returns

A vector embedding of type VECTOR.

Access control requirements

You must use a role that has been granted the SNOWFLAKE.CORTEX_USER database role or the SNOWFLAKE.CORTEX_EMBED_USER database role to call this function. See Cortex LLM privileges for more information on granting one of these privileges.

You must also have the USAGE privilege on the SNOWFLAKE.CORTEX schema to call this function.

Examples

In this example, a vector embedding is generated for the phrase hello world using the snowflake-arctic-embed-m-v1.5 model:

SELECT SNOWFLAKE.CORTEX.EMBED_TEXT_768('snowflake-arctic-embed-m-v1.5', 'hello world');

Refer to Snowflake AI and ML.

Limitations

Snowflake Cortex functions do not support dynamic tables.