|
|
import os |
|
|
import faiss |
|
|
import logging |
|
|
import streamlit as st |
|
|
from langchain.embeddings import HuggingFaceEmbeddings |
|
|
from langchain.vectorstores import FAISS |
|
|
from langchain.chains import RetrievalQA |
|
|
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline |
|
|
|
|
|
|
|
|
logging.basicConfig(level=logging.DEBUG) |
|
|
|
|
|
def load_faiss_index(index_path): |
|
|
""" |
|
|
Load a FAISS index from a specified path. |
|
|
Parameters: |
|
|
- index_path (str): Path to the FAISS index file. |
|
|
Returns: |
|
|
- faiss.Index: Loaded FAISS index object. |
|
|
""" |
|
|
if not os.path.exists(index_path): |
|
|
logging.error(f"FAISS index not found at {index_path}. Please create the index first.") |
|
|
st.error(f"FAISS index not found at {index_path}. Please create the index first.") |
|
|
raise FileNotFoundError(f"FAISS index not found at {index_path}.") |
|
|
try: |
|
|
logging.info(f"Attempting to load FAISS index from {index_path}.") |
|
|
index = faiss.read_index(index_path) |
|
|
logging.info("FAISS index loaded successfully.") |
|
|
st.success("FAISS index loaded successfully.") |
|
|
return index |
|
|
except Exception as e: |
|
|
logging.error(f"Failed to load FAISS index: {e}") |
|
|
st.error(f"Failed to load FAISS index: {e}") |
|
|
raise |
|
|
|
|
|
def load_llm(): |
|
|
""" |
|
|
Load the HuggingFace model for generating responses. |
|
|
""" |
|
|
checkpoint = "LaMini-T5-738M" |
|
|
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 pipe |
|
|
|
|
|
def process_answer(question): |
|
|
""" |
|
|
Process the user's question using the FAISS index and LLM. |
|
|
Parameters: |
|
|
- question (str): User's question to be processed. |
|
|
Returns: |
|
|
- str: The answer generated by the LLM. |
|
|
""" |
|
|
index_path = 'faiss_index/index.faiss' |
|
|
embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2") |
|
|
try: |
|
|
faiss_index = load_faiss_index(index_path) |
|
|
retriever = FAISS(index=faiss_index, embeddings=embeddings) |
|
|
llm = load_llm() |
|
|
qa = RetrievalQA.from_chain_type( |
|
|
llm=llm, |
|
|
chain_type="stuff", |
|
|
retriever=retriever, |
|
|
return_source_documents=True |
|
|
) |
|
|
result = qa.invoke(question) |
|
|
answer = result['result'] |
|
|
return answer, result |
|
|
except Exception as e: |
|
|
logging.error(f"An error occurred while processing the answer: {e}") |
|
|
st.error(f"An error occurred while processing the answer: {e}") |
|
|
return "An error occurred while processing your request.", {} |
|
|
|
|
|
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") |
|
|
try: |
|
|
answer, metadata = process_answer(question) |
|
|
st.write(answer) |
|
|
st.write(metadata) |
|
|
except Exception as e: |
|
|
st.error(f"An unexpected error occurred: {e}") |
|
|
|
|
|
if __name__ == '__main__': |
|
|
main() |
