app.py

import os
import torch
import gradio as gr

from gen import get_pretrained_models, get_output, setup_model_parallel

os.environ["RANK"] = "0"
os.environ["WORKD_SIZE"] = "1"
os.environ["MASTER_ADDR"] = "127.0.0.1"
os.environ["MASTER_PORT"] = "50505"

# local_rank, world_size = setup_model_parallel()
# generator = get_pretrained_models("7B", "tokenizer")

def chat(user_input):
    bot_response = "hello world"
    response = ""
    for word in bot_response.split(" "):
        response = response + " " + word
        yield response

with gr.Blocks() as demo:
    chatbot = gr.Chatbot()
    textbox = gr.Textbox("Hello, how are you doing today?")

    textbox.submit(chat, textbox, chatbot)

demo.queue().launch()