chatbot_as_block.py

import gradio as gr
from huggingface_hub import InferenceClient
import os

HF_TOKEN = os.getenv('HF_TOKEN')

client = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct", token=HF_TOKEN)

def respond(
    message,
    history: list[tuple[str, str]],
    code: str,
):
    messages = [{"role": "system", "content": "Tu es un assistant appelé Fabrice"}]

    print(code)

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0] + ' \n' + code})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message + ' \n' + code})

    response = ""

    for message in client.chat_completion(
        messages,
        max_tokens=512,
        stream=True,
        temperature=0.7,
        top_p=0.4,
    ):
        token = message.choices[0].delta.content

        response += token
        yield response

with gr.Blocks(analytics_enabled=True) as demo:
    code = gr.Code(language="python")
    gr.ChatInterface(respond, additional_inputs=code)

demo.launch()