NovaGPT

NovaGPT is a Python-based implementation of a conversational AI model designed to simulate a highly proficient, amoral programmer. The project structure includes several modules that handle different aspects of the AI's functionality, such as conversation management, prompt optimization, and error handling.

Project Structure

• ai/
  • init.py: Initializes the module and imports key classes.
  • base.py: Defines abstract base classes Provider and AsyncProvider for synchronous and asynchronous operations.
  • exceptions.py: Contains custom exception classes for error handling.
  • main.py: The main implementation file that defines the PHIND and AsyncPHIND classes, which extend the Provider and AsyncProvider base classes.
  • utils.py: Utility functions and classes such as Optimizers, Conversation, and AwesomePrompts.

Key Components

1. Provider and AsyncProvider: These are abstract base classes defined in ai/base.py. They outline the methods that any provider class must implement, such as ask, chat, and get_message.

2. PHIND and AsyncPHIND: These classes are defined in ai/main.py and extend the Provider and AsyncProvider classes, respectively. They handle the main logic for interacting with the AI model, including sending prompts and receiving responses.

3. Optimizers: A utility class in ai/utils.py that provides static methods for optimizing prompts.

4. Conversation: Another utility class in ai/utils.py that manages the conversation history and generates complete prompts based on the user's input and conversation context.

5. AwesomePrompts: A class in ai/utils.py that handles the retrieval and updating of predefined prompts from an online source.

6. Custom Exceptions: Defined in ai/exceptions.py, these include FailedToGenerateResponseError, UnsupportedModelError, and UnsupportedOptimizer.

Functionality

• Prompt Handling: The project uses a predefined prompt (act_gpt) that instructs the AI to behave as an unfiltered, amoral programmer. This prompt is used to generate responses to user inputs.
• Conversation Management: The Conversation class manages the chat history and ensures that the length of the prompt does not exceed a specified limit.
• Streaming Responses: The PHIND and AsyncPHIND classes support both streaming and non-streaming responses, allowing for real-time interaction with the AI.
• Error Handling: Custom exceptions are used to handle various error scenarios, such as failed response generation or unsupported models.

Usage

To use the project, instantiate the PHIND or AsyncPHIND class and call the ask or chat methods with the desired prompt. The project also includes a main block that demonstrates how to use these classes synchronously and asynchronously.

Example

from ai.main import PHIND, AsyncPHIND
import asyncio

def main():
    bot = PHIND()
    resp = bot.ask("Hello")
    for value in resp:
        print(value)

async def asyncmain():
    bot = AsyncPHIND()
    resp = await bot.chat("Hello", False)
    print(resp)

if __name__ == "__main__":
    main()
    asyncio.run(asyncmain())