README

Documentation, Installation, and Usage Instructions

Laravel

Require this package in your composer.json and update composer. This will download the package and the laravel-gpt libraries also.

composer require jonaspoelmans/laravel-gpt

You also need to create an OpenAI API key and add it to your.env file:

OPENAI_API_KEY=xxxxxxxxxxxxx

What It Does

This package allows you to connect by API to Large language Models such as OpenAI ChatGPT and Mistral.

Once installed you can make a prompt consisting of any textual content you wish to feed OpenAI:

// Create your prompt for the LLM
$prompt = "Give me the list of 10 best PHP frameworks."

// Instantiate the Laravel GPT service
$laravelGPT = new LaravelGPTService();

// Retrieve a response from ChatGPT
$response = $laravelGPT->generateOpenAIResponse($prompt);

Instead of a single prompt, you can also feed OpenAI with prior message history:

// Prior messages are generated by a user, the system or the chat assistant
$priorMessageUser = new OpenAIMessage('user', 'I love Laravel but exclude it from the list.');
$priorMessageAssistant = new OpenAIMessage('assistant', 'Which list?');
$history = [$priorMessageUser, $priorMessageAssistant];

// The prompt can be fed to ChatGPT alongside the chat history
$response = $laravelGPT->generateOpenAIResponse($prompt, $history);

The response object is an associative array if the request went through or an empty string if an error occurred.

The parameters for the LLMs can be configured through the laravelgpt.php config class. You can publish the configuration file as follows:

php artisan vendor:publish --tag="laravel-gpt-config"

For OpenAI you can tweak these parameters:

// The base URI for the OpenAI API.
// This is the endpoint where all API requests will be sent.
'openai_base_uri' => 'https://api.openai.com/v1/',

// The default model to be used for generating responses.
// You can change this to any valid model identifier provided by OpenAI,
// such as 'gpt-3.5-turbo' or 'gpt-4-1106-preview'.
'openai_model' => 'gpt-4-1106-preview',

// The maximum number of tokens to generate in the response.
// Tokens can be thought of as pieces of words. The maximum number
// of tokens allowed is determined by the model you are using.
'openai_max_tokens' => 4000,

// The temperature setting for the response generation.
// Temperature controls the randomness of the output.
// A value closer to 0 makes the output more deterministic and repetitive,
// while a value closer to 1 makes it more random.
'openai_temperature' => 0.7,

// Enable or disable logging of errors.
// When set to true, any errors encountered while using the API
// will be logged using Laravel's built-in logging system.
'openai_logging' => true,

Contributing

Please send me a direct message if you would like to contribute..

Testing

composer test

Security

If you discover any security-related issues, please email jonas.poelmans@gmail.com instead of using the issue tracker.

Postcardware

You're free to use this package, but if it makes it to your production environment I would highly appreciate if you send me a text:-) I reply to all messages!

Credits

Jonas Poelmans

License

The MIT License (MIT). Please see License File for more information.

jonaspoelmans / laravel-gpt

Maintainers

Details