app.py

import gradio as gr
import os
from huggingface_hub import InferenceClient

# Load Hugging Face token from secrets (set as HF_TOKEN)
hf_token = os.getenv("HF_TOKEN")

# Arabic Chat model client
client = InferenceClient("sambanovasystems/SambaLingo-Arabic-Chat", token=hf_token)

# Define the response function
def respond(message, history: list[tuple[str, str]], system_message, max_tokens, temperature, top_p):
    prompt = system_message.strip() + "\n"
    for user_msg, bot_msg in history:
        if user_msg:
            prompt += f"أنت: {user_msg}\n"
        if bot_msg:
            prompt += f"المساعد: {bot_msg}\n"
    prompt += f"أنت: {message}\nالمساعد:"

    # Generate response
    response = client.text_generation(
        prompt=prompt,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
    )
    return response.strip()

# Gradio chat UI
demo = gr.ChatInterface(
    fn=respond,
    additional_inputs=[
        gr.Textbox(value="أنت مساعد ذكي ودود يتحدث اللغة العربية.", label="رسالة النظام"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="عدد الرموز القصوى"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.8, step=0.1, label="درجة العشوائية (Temperature)"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (النواة)"),
    ],
    title="روبوت دردشة باللغة العربية"
)

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