app.py

import streamlit as st
import pandas as pd
import numpy as np
from io import StringIO
from transformers import AutoTokenizer, AutoModelForQuestionAnswering
from transformers import Trainer
import torch

st.title('HRA Document QA')

file_name = st.file_uploader("Upload the document that you want to ask questions")

if file_name is not None:
    text = file_name.getvalue()
    
    
    stringio = StringIO(file_name.getvalue().decode("utf-8"))

    context = stringio.read()
    

    question = st.chat_input("Ask some questions about this document")
    with st.chat_message("user"):
        st.write("Hello 👋 I am an HRA chatbot~")
        st.write("Here's the document that you uploaded:")
        st.write(context)

        if question:
            st.write("You asked a question:")
            st.write(question)
            
            tokenizer = AutoTokenizer.from_pretrained("./models/deepset/tinyroberta-squad")
            model = AutoModelForQuestionAnswering.from_pretrained("./models/deepset/tinyroberta-squad")

            inputs = tokenizer(question, context, return_tensors="pt")
            with torch.no_grad():
                outputs = model(**inputs)

            answer_start_index = outputs.start_logits.argmax()
            answer_end_index = outputs.end_logits.argmax()

            predict_answer_tokens = inputs.input_ids[0, answer_start_index : answer_end_index + 1]
            st.write("Answer:")
            st.write(tokenizer.decode(predict_answer_tokens, skip_special_tokens=True))