ballerinax/azure.openai.embeddings Ballerina library

Overview

The Azure OpenAI Embeddings API connector(https://learn.microsoft.com/en-us/azure/cognitive-services/openai/reference#embeddings/) OpenAPI specification.

The Azure OpenAI Service REST API Embeddings Endpoint will generate a vector representation of a given input that can be easily consumed by machine learning models and other algorithms.

Key Features

• Programmatic access to create and manage resources via REST API
• Send and publish data through the API
• Manage user accounts and profiles
• Secure authentication with API key or OAuth support

Prerequisites

• Create an Azure account
• Create an Azure OpenAI resource
• Deploy an appropriate model within the resource by referring to Deploy a model guide
• Obtain tokens by following Azure OpenAI Authentication guide

Quickstart

To use the Azure OpenAI Embeddings connector in your Ballerina application, update the .bal file as follows:

Step 1: Import connector

Import the ballerinax/azure.openai.embeddings module into the Ballerina project.

import ballerinax/azure.openai.embeddings;
import ballerina/io;

Step 2: Create a new connector instance

Create and initialize an embeddings:Client with the obtained apiKey and a serviceUrl containing the deployed models.

final embeddings:Client embeddingsClient = check new (
    config = {auth: {apiKey: apiKey}},
    serviceUrl = serviceUrl
);

Step 3: Invoke connector operation

1. Now you can use the operations available within the connector.

   Note: These operations are in the form of remote operations.

   Following is an example on obtaining embeddings from a GPT-3 ada model:

   public function main() returns error? {
   
       final embeddings:Client embeddingsClient = check new (
           config = {auth: {apiKey: apiKey}},
           serviceUrl = serviceUrl
       );
   
       embeddings:Deploymentid_embeddings_body embeddingsBody = {
           input: "I have bought several of Vitality canned"
       };
   
       embeddings:Inline_response_200 embeddingsResult = check embeddingsClient->/deployments/["embedding"]/embeddings.post("2023-03-15-preview", embeddingsBody);
   
       io:println(embeddingsResult);
   }

2. Use bal run command to compile and run the Ballerina program.