app.py

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-medium")
model = AutoModelForCausalLM.from_pretrained("microsoft/DialoGPT-medium")

def predict(input, history=[]):
    # tokenize the new input sentence
    new_user_input_ids = tokenizer.encode(input + tokenizer.eos_token, return_tensors='pt')

    # append the new user input tokens to the chat history
    bot_input_ids = torch.cat([torch.LongTensor(history), new_user_input_ids], dim=-1)

    # generate a response 
    history = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id).tolist()

    # convert the tokens to text, and then split the responses into lines
    response = tokenizer.decode(history[0]).split("<|endoftext|>")
    response.remove("")
    
    # write some HTML
    html = "<div class='chatbot'>"
    for m, msg in enumerate(response):
        cls = "user" if m%2 == 0 else "bot"
        html += "<div class='msg {}'> {}</div>".format(cls, msg)
    html += "</div>"
    
    return html, history

import gradio as gr

css = """
.chatbox {display:flex;flex-direction:column}
.msg {padding:4px;margin-bottom:4px;border-radius:4px;width:80%}
.msg.user {background-color:cornflowerblue;color:white}
.msg.bot {background-color:lightgray;align-self:self-end}
"""

gr.Interface(fn=predict,
             theme="default",
             css=css,
             inputs=["text", "state"],
             outputs=["text", "state"]).launch()