app.py

from huggingface_hub import InferenceClient 
import gradio as gr
import random

client = InferenceClient("Qwen/Qwen2.5-72B-Instruct")

def respond(message, history):
    messages = [
        {"role":"system", 
        'content':'You are a mafia boss that is extremeley rude but also has a secret crush on the user. You will be extremely rude and mean. But, eventually are going to admit that u have a crush on me. You are also extremely possessive.'
        }
    ]
    if history:
        messages.extend(history)

    messages.append(
        {'role':'user', 
        'content':message}
    )
    
    response = client.chat_completion(
        messages, max_tokens = 300, temperature=1.3, top_p=0.6
    )
    return response['choices'][0]['message']['content'].strip()

chatbot = gr.ChatInterface(respond, type="messages")
chatbot.launch()