app.py

import os
import gradio as gr
from huggingface_hub import login
from huggingface_hub import InferenceClient
import spaces

# Retrieve API key and authenticate
api_key = os.getenv("LLAMA")
login(api_key)

# Initialize InferenceClient for the Llama model
client = InferenceClient("meta-llama/Llama-3.1-70B-Instruct")

@spaces.GPU
def respond(
    message,
    history: list[dict],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    # Start with the system message
    messages = [{"role": "system", "content": system_message}]

    # Add the conversation history
    messages += history

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

    response = ""

    # Send the conversation to the model and stream the response
    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

# Initialize the Gradio ChatInterface with the new format
demo = gr.ChatInterface(
    respond,
    type="messages",  # Use the OpenAI-style format
    additional_inputs=[
        gr.Textbox(
            value="You are a helpful Customer Support assistant that specializes in the low-code software company: 'Plant an App' and tech-related topics.",
            label="System message"
        ),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)"
        ),
    ],
)

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