app.py

import os
from huggingface_hub import InferenceClient
import gradio as gr

# Debug token in logs (remove later!)
print("DEBUG TOKEN:", os.getenv("mytokenvivi"))

# Load token from env variable (must be set as a Secret in Hugging Face Space)
token = os.getenv("mytokenvivi")
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=token)

def respond(message, history: list[tuple[str, str]], system_message, max_tokens, temperature, top_p):
    messages = [{"role": "system", "content": system_message}]

    for user, assistant in history:
        if user:
            messages.append({"role": "user", "content": user})
        if assistant:
            messages.append({"role": "assistant", "content": assistant})

    messages.append({"role": "user", "content": message})

    response = ""
    try:
        for msg in client.chat_completion(
            messages=messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = msg.choices[0].delta.content
            if token:
                response += token
                yield response
    except Exception as e:
        yield f"[ERROR] {str(e)}"

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly chatbot.", label="System message"),
        gr.Slider(1, 2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(0.1, 4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(0.2, 1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
    ],
)

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