app.py

def generate_response(
    user_input: str, 
    history: List[Tuple[str, str]], 
    max_tokens: int, 
    temperature: float, 
    top_p: float
) -> str:
    """
    Generates a response from the AI model.
    Args:
        user_input: The user's input message.
        history: A list of tuples containing the conversation history 
                 (user input, AI response).
        max_tokens: The maximum number of tokens in the generated response.
        temperature: Controls the randomness of the generated response.
        top_p: Controls the nucleus sampling probability.
    Returns:
        str: The generated response from the AI model.
    """
    try:
        # Build the message list with system message and history
        messages = [{"role": "system", "content": SYSTEM_MESSAGE}]
        messages.extend([{"role": "user" if i % 2 == 0 else "assistant", "content": val} 
                         for i, val in enumerate(sum(history, ()))]) 
        messages.append({"role": "user", "content": user_input})

        # Generate response from the model
        response = ""
        for msg in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            # Check if 'choices' is present and non-empty
            if msg and 'choices' in msg and msg['choices']:
                # Ensure the 'delta' and 'content' properties exist before using them
                token = msg['choices'][0].get('delta', {}).get('content', '')
                if token:
                    response += token
            else:
                # Handle unexpected response format or empty choices
                print("Warning: Unexpected response format or empty 'choices'.")
                break
        return response or "Sorry, I couldn't generate a response. Please try again."

    except Exception as e:
        print(f"An error occurred: {e}")
        return "Error: An unexpected error occurred while processing your request."