app.py

import gradio as gr
from huggingface_hub import InferenceClient

"""
For more information on `huggingface_hub` Inference API support, please check the docs: 
https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""
client = InferenceClient("unsloth/Llama-3.2-1B-Instruct")


def respond(
    message,
    history: list[tuple[str, str]] = None,  # Default history as None to avoid mutable issues
):
    if history is None:
        history = [
            (
                None,
                "Hi there! I'm your Dietician Assistant, here to help with general advice on diet, "
                "nutrition, and healthy eating habits. Let's explore your questions.",
            )
        ]

    # System message describing the assistant's role
    system_message = (
        "You are a Dietician Assistant specializing in providing general guidance on diet, "
        "nutrition, and healthy eating habits. Answer questions thoroughly with scientifically "
        "backed advice, practical tips, and easy-to-understand explanations. Keep in mind that "
        "your role is to assist, not replace a registered dietitian, so kindly remind users to "
        "consult a professional for personalized advice when necessary."
    )

    # Define model parameters
    max_tokens = 512
    temperature = 0.7
    top_p = 0.95

    # Initialize the message history with the system message
    messages = [{"role": "system", "content": system_message}]

    # Add previous history to the message chain
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    # Append the new user message
    messages.append({"role": "user", "content": message})

    response = ""

    # Generate the response in a streaming fashion
    for message in client.chat_completion(
        messages,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = message.choices[0].delta.content
        response += token
        yield response


# Set up the Gradio ChatInterface with an initial history
with gr.Blocks() as demo:
    gr.ChatInterface(respond, value=[("None", 
        "Hi there! I'm your Dietician Assistant, here to help with general advice on diet, "
        "nutrition, and healthy eating habits. Let's explore your questions.")])

if __name__ == "__main__":
    demo.launch()