app.py

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

# Streamlit UI
st.title("Indian Constitution Q&A RAG App")

# Upload JSON File
uploaded_file = st.file_uploader("Upload Constitution JSON", type="json")

if uploaded_file is not None:
    # Load JSON dataset
    data = json.load(uploaded_file)

    # Extract questions and answers
    texts = [f"Q: {item['question']}\nA: {item['answer']}" for item in data]

    # Load the embedding model
    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")

    # Create FAISS vector database
    vector_db = FAISS.from_texts(texts, embeddings)

    # Load Open-Source LLM (LLaMA-2 7B Open Chat Model)
    tokenizer = AutoTokenizer.from_pretrained("NousResearch/Llama-2-7b-chat-hf")
    model = AutoModelForCausalLM.from_pretrained("NousResearch/Llama-2-7b-chat-hf")
    text_pipeline = pipeline("text-generation", model=model, tokenizer=tokenizer)
    llm = HuggingFacePipeline(pipeline=text_pipeline)

    # Create RAG pipeline
    qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=vector_db.as_retriever())

    query = st.text_input("Enter your legal query:")

    if query:
        response = qa_chain.run(query)
        st.write("### AI-Generated Answer:")
        st.write(response)