OpenAI

OpenAI Project Introduction

Overview

OpenAI provides a versatile API designed for tasks involving the understanding or generation of natural language and code. The project offers a suite of models that vary in power and are tailored for different applications. Users can even fine-tune their own models. These models are applicable across a wide range of uses, from content generation to semantic search and classification.

Installation

OpenAI's Swift implementation can be easily integrated using either CocoaPods or Swift Package Manager.

CocoaPods

To include OpenAI in a project using CocoaPods, simply add the following line to your Podfile:

pod 'OpenAIKit'

Swift Package Manager

To add OpenAI as a dependency through Swift Package Manager, insert the following into the dependencies section of your Package.swift file:

dependencies: [
    .package(url: "https://github.com/FuturraGroup/OpenAI.git", .branch("main"))
]

Usage

Initialization

It's advisable to use environment variables to set the OpenAI API key to avoid hardcoding it. For example:

let apiToken: String = "<Your OpenAI API Token here>"
let organizationName: String = "<Your OpenAI Organization name>"

Initialize OpenAIKit with your API Token, and optionally, your organization name:

import OpenAIKit

public let openAI = OpenAIKit(apiToken: apiToken, organization: organizationName)

Additional initialization parameters include timeout settings and options for custom API endpoints and SSL certificates.

SSL Handshake

SSL Handshake is a security feature to protect against MITM (Man-in-the-Middle) attacks by validating SSL certificates, ensuring secure communications with the OpenAI server.

Completions

OpenAI supports various completion methods to generate text based on a prompt. For instance:

openAI.sendCompletion(prompt: "Hello!", model: .gptV3_5(.davinciText003), maxTokens: 2048) { result in
    switch result {
    case .success(let aiResult):
        if let text = aiResult.choices.first?.text {
            print("response text: \(text)")
        }
    case .failure(let error):
        print("Error: \(error.localizedDescription)")
    }
}

This API also supports asynchronous usage and includes features like chat completions and streaming responses.

Chat Completions

Chat completions allow the API to respond to text prompts by considering the context of previous chat messages, specifically supporting models like gpt-3.5-turbo.

Stream

Stream mode enables real-time response retrieval during completion requests, allowing for immediate result processing without having to wait for a complete response.

JSON Mode

For enhanced performance and error reduction in certain models, JSON mode can ensure responses are formatted as valid JSON objects.

Generate Image

Using DALL·E, OpenAI can generate, edit, and create image variations based on a given description:

openAI.sendImagesRequest(prompt: "bird", size: .size512, n: 1) { result in
    switch result {
    case .success(let aiResult):
        if let urlString = aiResult.data.first?.url {
            print("generated image url: \(urlString)")
        }
    case .failure(let error):
        print("Error: \(error.localizedDescription)")
    }
}

Contribute

OpenAI welcomes contributions to improve the library. Feedback and suggestions can be submitted via the project's GitHub issues.

License

OpenAI is distributed under the MIT License, allowing for broad freedom in its use and modification, with the condition of including the license in all copies or significant portions of the software.