app.py

import transformers
import streamlit as st
from annotated_text import annotated_text
from streamlit_lottie import st_lottie
import requests 

def load_lottieurl(url: str):
    r = requests.get(url) #Make a request to a web page, and return the status code:
    if r.status_code != 200: #200 is the HTTP status code for "OK", a successful response.
        return None
    return r.json() 


    
@st.cache(allow_output_mutation=True, show_spinner=False)
def get_pipe():
    tokenizer = transformers.AutoTokenizer.from_pretrained("deepset/roberta-base-squad2")
    model = transformers.AutoModelForQuestionAnswering.from_pretrained("deepset/roberta-base-squad2")
    pipe = transformers.pipeline("question-answering", model=model, tokenizer=tokenizer)
    return pipe


def parse_context(context, prediction):
    parsed_context = []
    parsed_context.append(context[:prediction["start"]])
    parsed_context.append((prediction["answer"], "ANSWER", "#afa"))
    parsed_context.append(context[prediction["end"]:])
    return parsed_context


st.set_page_config(
    page_title="Question And Answering",
    page_icon="✨",
    layout="centered",
    initial_sidebar_state="auto",
     menu_items={ #Configure the menu that appears on the top-right side of this app.
            'About': 'https://www.linkedin.com/in/harsh-kashyap-79b87b193/', #A markdown string to show in the About dialog. Used my linkedIn id
     }
)

dashboard1 = load_lottieurl("https://assets1.lottiefiles.com/packages/lf20_au4zdsr8.json") #get the animated gif from file
st_lottie(dashboard1, key="Dashboard1", height=400) #change the size to height 400
st.title("Deep Question and Answering System 🗣")
st.markdown("##") 

st.write("Enter context and a question and press 'Predict' to extract the answer from the context.")

default_context = "My name is Harsh and I live in Patiala."
default_question = "What is my name?"

context = st.text_area("Enter context here:", value=default_context)
question = st.text_input("Enter your Question: 🙋", value=default_question)
submit = st.button('Predict')

with st.spinner(f"Getting your Answer... 💫"):
    pipe = get_pipe()

if (submit and len(context.strip()) > 0 and len(question.strip()) > 0) or \
   (len(context.strip()) > 0 and len(question.strip()) > 0):

    prediction = pipe(question, context)

    parsed_context = parse_context(context, prediction)
    st.markdown("Here's the answer 🗣")
    annotated_text(*parsed_context)
    st.balloons()
       
st.markdown("<br><hr><center>Made with ❤️ by <a href='https://www.linkedin.com/in/harsh-kashyap/'><strong>Harsh Kashyap</strong></a></center><hr>", unsafe_allow_html=True)