from transformers import TFAutoModelForCausalLM, AutoTokenizer
import tensorflow as tf
import gradio as gr
TITLE = "DialoGPT -- Chatbot"
DESCRIPTION = """
This application allows you to talk with a machine.
In the back-end is using the DialoGPT model from Microsoft.
This model extends GPT2 towards the conversational neural response generetion domain.
You can also see the ArXiv paper
"""
EXAMPLES = [
["What is your favorite videogame?"],
["What do you do for work?"],
["What are your hobbies?"],
["What is your favorite food?"],
]
ARTICLE = r"""
Done by dr. Gabriel Lopez
For more please visit: My Page
"""
# checkpoint = "ericzhou/DialoGPT-Medium-Rick_v2" #pytorch
# checkpoint = "epeicher/DialoGPT-medium-homer" #pytorch
checkpoint = "microsoft/DialoGPT-medium" #tf
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = TFAutoModelForCausalLM.from_pretrained(checkpoint)
# interaction function
def chat_with_bot(user_input, chat_history_and_input=[]):
emb_user_input = tokenizer.encode(
user_input + tokenizer.eos_token, return_tensors="tf"
)
if chat_history_and_input == []:
bot_input_ids = emb_user_input # first iteration
else:
bot_input_ids = tf.concat(
[chat_history_and_input, emb_user_input], axis=-1
) # other iterations
chat_history_and_input = model.generate(
bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id
).numpy()
# print
bot_response = tokenizer.decode(
chat_history_and_input[:, bot_input_ids.shape[-1] :][0],
skip_special_tokens=True,
)
return bot_response, chat_history_and_input
# gradio interface
in_text = gr.Textbox(value="How was the class?", label="Start chatting!")
out_text = gr.Textbox(value="", label="Chatbot response:")
gr.Interface(
inputs=[in_text, "state"],
outputs=[out_text, "state"],
examples=EXAMPLES,
title=TITLE,
description=DESCRIPTION,
article=ARTICLE,
fn=chat_with_bot,
allow_flagging=False,
).launch()