app.py

import gradio as gr
from huggingface_hub import InferenceClient

client = InferenceClient("meta-llama/Llama-3.2-1B")

def chat(message, history):
    response = client.text_generation(message, max_new_tokens=200)
    return response

demo = gr.ChatInterface(fn=chat, title="ALNIK AI")
demo.launch()