OpenAIEmbeddingsAPI

The OpenAIEmbeddingsAPI provides functionalities to create embeddings using OpenAI's API.

https://platform.openai.com/docs/api-reference/embeddings

Functions

create()

create(input : Text; model: Text; parameters : OpenAIEmbeddingsParameters) : OpenAIEmbeddingsResult

Creates an embeddings for the provided input, model and parameters.

Argument	Type	Description
input	Text or Collection of Text	The input to vectorize.
model	Text	The model to use
parameters	OpenAIEmbeddingsParameters	The parameters to customize the embeddings request.
Function result	OpenAIEmbeddingsResult	The embeddings.

Example Usages

Get vector for a single text entry

var $result:=$client.embeddings.create("it rains cats and dogs"; "text-embedding-ada-002")
var $vector: 4D.Vector:=$result.vector
// or var $embedding: cs.AIKit.OpenAIEmbedding:=$result.embedding

Get vectors for a collection of text entries

var $inputs:=["it rains cats and dogs"; "il pleut à boire debout"]
var $result:=$client.embeddings.create($inputs; "text-embedding-ada-002")
var $vectors : Collection:=$result.vectors // collection of 4D.Vector

Using another service

Before using embeddings with a specific service, please check its documentation to see if embeddings are supported, and select the appropriate embedding model.

For example, for Mistral, use mistral-embed or codestral-embed

var $result:=$client.embeddings.create($inputs; "mistral-embed")