OpenAI

OpenAI governs a embeddings API that is implemented by various providers, including vLLM.

Dependencies

To interface in Vecworks with embedding services implementing the OpenAI interface additional dependencies need to be installed. You may do so using pip:

pip install vecworks[openai]

API

class vecworks.vectorizers.openai.OpenAIVectorizer

Wrapper class to ease use of services implementing the OpenAI embedding API.

Interface

__init__(url: str, model: str, authenticator: Authenticator | None = None, dimensions: int | None = None)

Initializes the vectorizer.

Parameters

== Connection

url

URL of the service providing an OpenAI-compatible API.

model

Name of the model. If no name is passed, the name must be specified when transform() is called.

authenticator

Authenticator that may be used to acquire an authentication key when the external service requests for authentication.

== Configuration

dimensions

The number of embeddings the resulting vector should have.

transform(input: Any | Iterable[Any], model: str | None = None) → ndarray

Vectorizes the given data.

Also see: Vectorizer.