app.py

import gradio as gr
import random
import os
from huggingface_hub import InferenceClient

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

#def respond_yes_no_random(message, history):
#    responses = ["yes", "no"]
#    return random.choice(responses)

#chatbot = gr.ChatInterface(respond_yes_no_random, type = "messages", examples = ["HELLO", "Bye!"], title = "Annie's ChatBot")

def respond(message, history):
    messages = [{"role": "system", "content": "You are a vampire!"}]

    if history:
        messages.extend(history)

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

    response = client.chat_completion(
        messages,
        max_tokens=100
    )

    return response['choice'][0]['message']['content'].strip()
    


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

chatbot.launch()