CBOPENAI

CBOPENAI is a module that provides a simple API to access OpenAI's variety of AI services.

Requirements

• Adobe CF 2018+ or Lucee 5+
• ColdBox 6+

Installation

Install CommandBox, then from your terminal, run:

# Install latest stable version
box install cbopenai
# Install bleeding edge
box install cbopenai@be

Service Object

CBOPENAI comes with a service object you can use for all operations.

// Using WireBox injection
property name="openAIService" inject="OpenAIService@cbopenai";
// Using getInstance
var openAIService = getInstance( "OpenAIService@cbopenai" );

Usage

getModels

List and describe the various models available in the API. You can refer to the Models documentation to understand what models are available and the differences between them.

function getModels()

var models = openAIService.getModels();

getModel

Retrieves a model instance, providing basic information about the model such as the owner and permissioning.

function getModel( required string model )

var model = openAIService.getModel( "text-davinci-003" );

createModeration

Classifies if text violates OpenAI's Content Policy

function createModeration( required string input )

var moderation = openAIService.createModeration( input="I'm going to murder that sandwhich later." );

createCompletion

Given a prompt, the model will return one or more predicted completions, and can also return the probabilities of alternative tokens at each position.

function createCompletion(
    required any prompt,
    string model = "text-davinci-003",
    string suffix,
    numeric maxTokens,
    numeric temperature,
    numeric topP,
    numeric n, 
    boolean stream,
    numeric logprops,
    boolean echo,
    any stop,
    numeric presencePenalty,
    numeric frequencyPenalty,
    numeric bestOf,
    struct logitBias,
    string user
)

var completion = openAIService.createCompletion( prompt="What is 2+2?" );

createChatCompletion

Given a list of messages describing a conversation, the model will return a response.

function createChatCompletion(
    required array messages,
    string model = "gpt-3.5-turbo",
    numeric max_tokens,
    numeric temperature,
    numeric top_p,
    numeric n,
    boolean stream,
    numeric logprops,
    any stop,
    numeric presence_penalty,
    numeric frequency_penalty,
    struct logit_bias,
    string user
)

var completion = openAIService.createChatCompletion(
    messages = [
        {
            "role": "user",
            "content": "Hello!"
        }
    ]
)

createImage

Creates an image given a prompt.

function createImage(
    required string prompt,
    numeric n,
    string size,
    string response_format,
    string user
)

var images = openAIService.createImage( prompt="A cat with a funny hat.", n=2 ); // create 2 images

createImageEdit

Creates an edited or extended image given an original image and a prompt.

function createImageEdit(
    required string image,
    string mask,
    string prompt,
    numeric n,
    string size,
    string response_format,
    string user
)

var image = openAIService.createImageEdit( image=expandPath( "somefile.png" ) );

createImageVariation

Creates a variation of a given image.

function createImageVariation(
    required string image,
    numeric n,
    string size,
    string response_format,
    string user
)

var image = openAIService.createImageVariation( image=expandPath( "somefile.png" ) );

createEdit

Given a prompt and an instruction, the model will return an edited version of the prompt.

function createEdit(
    string model = "text-davinci-edit-001",
    string input,
    string instruction,
    numeric n,
    numeric temperature,
    numeric top_p
)

var edit = openAIService.createEdit(
    input="What day of the wek is it?",
    instruction="Fix the spelling mistakes."
);

createdEmbedding

Get a vector representation of a given input that can be easily consumed by machine learning models and algorithms.

function createEmbedding(
    required any input,  
    model = "text-embedding-ada-002",
    string user        
)

var embedding = openAIService.createEmbedding( input="The food was delicious and the waiter..." );

createAudioTranscription

Transcribes audio into the input language.

function createAudioTranscription(
    required string file,
    required string model = "whisper-1",
    string prompt,
    string response_format,
    numeric temperature,
    string language
)

var text = openAIService.createAudioTranscription(
    file = expandPath( "/includes/audio/obi_wan_hello_there.mp3" )
); 

createAudioTranslation

Translates audio into into English.

function createAudioTranslation(
    required string file,
    required string model = "whisper-1",
    string prompt,
    string response_format,
    numeric temperature
)

var text = openAIService.createAudioTranslation(
    file = expandPath( "/includes/audio/german.mp3" )
); 

Examples

Included in this repo is an app where you can experience and see code examples covering all the features of CBOPENAI.

Just run the following from the test-harness directory:

box install && cd .. && box server start

Then visit:

http://localhost:60299/

Contributing

We love PRs! Please create a ticket using the Issue Tracker before submitting a PR.

Test Harness

There is a test harness application included in this repo.

To start the test harness:

cd test-harness
box install
box server start

This will start the test harness using a random port selected by CommandBox. For example, if the random port selected is 60299, you can run the test suite using http://127.0.0.1:60299/tests.

License

Apache License 2.0

Credits

The CBOPENAI module for ColdBox is written and maintained by Grant Copley and Ortus Solutions.

Project Support

If CBOPENAI makes you happy, please consider becoming an Ortus Patreon supporter.

Resources

• Docmentation: https://github.com/coldbox-modules/cbopenai
• API Docs: https://apidocs.ortussolutions.com/#/coldbox-modules/cbopenai/
• GitHub Repository: https://github.com/coldbox-modules/cbopenai
• Issue Tracker: https://github.com/coldbox-modules/cbopenai/issues
• ITB 2023 Practical AI session: https://docs.google.com/presentation/d/1xXlGBs_kNZhrAgS8xxJ4T5NFev2nH4FAaZ3DXYt8wqQ/edit?usp=sharing