ForOpenAI: An unofficial Fortran library for OpenAI.

How to use

Prerequisites:

On Ubuntu, you need to install the curl development headers. Use the following command:

sudo apt install -y libcurl4-openssl-dev

Clone the repository:

You can clone the ForOpenAI repository from GitHub using the following command:

git clone https://github.com/gha3mi/foropenai.git

cd foropenai

OpenAI API Key Configuration:

Your Secret API key can be located by accessing the OpenAI User settings.

For enhanced security and convenience, it is strongly recommended to configure the API key as an environment variable.

• On Ubuntu, use the following command, replacing "your_api_key" with your actual API key:

  export OPENAI_API_KEY="your_api_key"

• (Optional) If desired for organizational purposes, you can also establish an optional environment variable on Ubuntu:

  export OPENAI_ORG="your_organization"

• Alternatively, the OpenAI API key can be included in the foropenai.json configuration file.

  {
      "base": {
          "api_key": "OPENAI_API_KEY",
          "organization": ""
      }
  }

Use ChatGPT from the terminal:

fpm run

Example

Audio

Create transcription

program test_Transcription

   use foropenai, only: Transcription

   implicit none

   type(Transcription) :: trs

   call trs%set_base_data(file_name='foropenai.json')
   call trs%set(file_name='foropenai.json')

   call trs%create(file='test/audio.mp3')
   call trs%print_file()
   call trs%print_assistant_response()

   call trs%finalize()

end program test_Transcription

Settings

foropenai.json

{
    "base": {
        "api_key": "OPENAI_API_KEY",
        "organization": ""
    },

    "Transcription": {
        "url": "https://api.openai.com/v1/audio/transcriptions",
        "model": "whisper-1",
        "temperature": 0.0,
        "language": "en",
        "response_format": "json"
    }
}

Result

file: test/audio.mp3
Whisper: FORTRAN stands for Formula Translation.

Create translation

program test_Translation

   use foropenai, only: Translation

   implicit none

   type(Translation) :: trs

   call trs%set_base_data(file_name='foropenai.json')
   call trs%set(file_name='foropenai.json')

   call trs%create(file='test/audio_de.mp3')
   call trs%print_file()
   call trs%print_assistant_response()

   call trs%finalize()

end program test_Translation

Settings

foropenai.json

{
    "base": {
        "api_key": "OPENAI_API_KEY",
        "organization": ""
    },

    "Translation": {
        "url": "https://api.openai.com/v1/audio/translations",
        "model": "whisper-1",
        "temperature": 0.0,
        "response_format": "json"
    }
}

Result

file: test/audio_de.mp3
Whisper: FORTRAN stands for Formula Translation and is a programming language.

Chat

Create chat completion

program test_ChatCompletion

   use foropenai, only: ChatCompletion

   implicit none

   type(ChatCompletion) :: chat

   call chat%set_base_data(file_name='foropenai.json')
   call chat%set(file_name='foropenai.json')

   call chat%init_messages(n=3)
   call chat%messages(1)%set(role='system', content='You are a helpful assistant.')
   call chat%messages(2)%set(role='user', content='Hello?')
   call chat%messages(3)%set(role='assistant', content='')

   call chat%print_user_message()
   call chat%create()
   call chat%print_assistant_response()

   call chat%usage%print()

   call chat%finalize()

end program test_ChatCompletion

Settings

foropenai.json

{
    "base": {
        "api_key": "OPENAI_API_KEY",
        "organization": ""
    },

    "ChatCompletion": {
        "user_name": "Ali",
        "url": "https://api.openai.com/v1/chat/completions",
        "model": "gpt-3.5-turbo",
        "temperature": 1.0,
        "max_tokens": 200,
        "top_p": 1.0,
        "frequency_penalty": 0.0,
        "presence_penalty": 0.0,
        "n": 1,
        "stream": false
    }
}

Result

Ali: Hello?
ChatGPT: Hello! How can I assist you today?

Image

Image Generation

program test_ImageGeneration

   use foropenai, only: ImageGeneration

   implicit none

   type(ImageGeneration) :: image

   call image%set_base_data(file_name='foropenai.json')
   call image%set(file_name='foropenai.json')

   call image%create(prompt='a cat with a computer')
   call image%print_prompt()
   call image%print_assistant_response()

   call image%finalize()

end program test_ImageGeneration

Settings

foropenai.json

{
    "base": {
        "api_key": "OPENAI_API_KEY",
        "organization": ""
    },

    "ImageGeneration": {
        "user_name": "Ali",
        "url": "https://api.openai.com/v1/images/generations",
        "size": "1024x1024",
        "n": 1,
        "response_format": "url"
    }
}

Result

fpm dependency

If you want to use ForOpenAI as a dependency in your own fpm project, you can easily include it by adding the following line to your fpm.toml file:

[dependencies]
foropenai = {git="https://github.com/gha3mi/foropenai.git"}

API documentation

The most up-to-date API documentation for the master branch is available here. To generate the API documentation for ForOpenAI using ford run the following command:

ford ford.yml

Contributing

Contributions to ForOpenAI are welcome! If you find any issues or would like to suggest improvements, please open an issue.