app.py

import gradio as gr
from huggingface_hub import InferenceClient

client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")

def respond(message, history):
    
    messages = [{"role": "system", "content": "You are a bookworm chatbot. Always connect your response back to books."}]
    
    if history:
        messages.extend(history)
        
    messages.append({"role": "user", "content": message})
    
    response = ""
    
    for message in client.chat_completion(
        messages,
        max_tokens=150,
        temperature=0.2,
        stream=True
    ):
        token = message.choices[0].delta.content
        response += token
        yield response
    

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

chatbot.launch()