app.py

import streamlit as st
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS
from langchain.llms import HuggingFacePipeline
from langchain.chains import RetrievalQA

checkpoint = "LaMini-T5-738M"

@st.cache_resource
def load_llm():
    tokenizer = AutoTokenizer.from_pretrained(checkpoint)
    model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint)
    pipe = pipeline(
        'text2text-generation',
        model=model,
        tokenizer=tokenizer,
        max_length=256,
        do_sample=True,
        temperature=0.3,
        top_p=0.95
    )
    return HuggingFacePipeline(pipeline=pipe)

@st.cache_resource
def qa_llm():
    llm = load_llm()
    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
    db = FAISS.load_local("faiss_index", embeddings)
    retriever = db.as_retriever()
    qa = RetrievalQA.from_chain_type(
        llm=llm, chain_type="stuff", retriever=retriever, return_source_documents=True)
    return qa

def process_answer(instruction):
    qa = qa_llm()
    generated_text = qa(instruction)
    answer = generated_text['result']
    return answer, generated_text

def main():
    st.title("Search Your PDF 🐦📄")
    with st.expander("About the App"):
        st.markdown(
            """
            This is a Generative AI powered Question and Answering app that responds to questions about your PDF File.
            """
        )
    question = st.text_area("Enter your Question")
    if st.button("Ask"):
        st.info("Your Question: " + question)
        
        st.info("Your Answer")
        answer, metadata = process_answer(question)
        st.write(answer)
        st.write(metadata)

if __name__ == '__main__':
    main()