Spaces:
Sleeping
Sleeping
| import streamlit as st | |
| from langchain.vectorstores import FAISS | |
| from langchain.embeddings import OpenAIEmbeddings | |
| from langchain.document_loaders import PyPDFLoader | |
| from langchain.llms import HuggingFacePipeline | |
| from langchain.chains import RetrievalQA | |
| import groqapi | |
| # Step 1: Initialize Groq API and Llama Model | |
| def load_llama_model(api_key, model_name): | |
| """Load the Llama model using Groq API.""" | |
| groqapi.set_api_key(api_key) | |
| return HuggingFacePipeline.from_pretrained(model_name) | |
| # Step 2: Load and Process PDF | |
| def process_pdf(pdf_path): | |
| """Load and split the PDF into documents.""" | |
| loader = PyPDFLoader(pdf_path) | |
| documents = loader.load_and_split() | |
| return documents | |
| # Step 3: Create Vector Database | |
| def create_vector_db(documents): | |
| """Create a FAISS vector database from documents.""" | |
| embeddings = OpenAIEmbeddings() # Use OpenAI embeddings for vectorization | |
| vector_db = FAISS.from_documents(documents, embeddings) | |
| return vector_db | |
| # Step 4: Build RAG Pipeline | |
| def build_rag_pipeline(vector_db, llama_model): | |
| """Build the Retrieval-Augmented Generation (RAG) pipeline.""" | |
| retriever = vector_db.as_retriever(search_type="similarity", search_kwargs={"k": 5}) | |
| qa_chain = RetrievalQA.from_chain_type( | |
| retriever=retriever, | |
| llm=llama_model, | |
| return_source_documents=True | |
| ) | |
| return qa_chain | |
| # Streamlit App | |
| def main(): | |
| st.title("KP Universities Act 2016 - Query App") | |
| st.write("Ask any question about the KP Universities Act 2016.") | |
| # Step 1: Upload PDF | |
| uploaded_pdf = st.file_uploader("Upload the KP Universities Act 2016 PDF", type="pdf") | |
| if uploaded_pdf: | |
| with open("uploaded_act.pdf", "wb") as f: | |
| f.write(uploaded_pdf.read()) | |
| documents = process_pdf("uploaded_act.pdf") | |
| st.success("PDF Loaded and Processed Successfully!") | |
| # Step 2: Input Groq API Key | |
| api_key = st.text_input("Enter your Groq API Key", type="password") | |
| model_name = "llama-3.1-8b-instant" | |
| if api_key and st.button("Load Llama Model"): | |
| try: | |
| # Load Llama Model | |
| llama_model = load_llama_model(api_key, model_name) | |
| st.success("Llama Model Loaded Successfully!") | |
| # Build Vector DB and QA Chain | |
| vector_db = create_vector_db(documents) | |
| qa_chain = build_rag_pipeline(vector_db, llama_model) | |
| # Step 3: Ask Questions | |
| query = st.text_input("Ask a question:") | |
| if query: | |
| with st.spinner("Fetching Answer..."): | |
| response = qa_chain({"query": query}) | |
| answer = response["result"] | |
| source_docs = response["source_documents"] | |
| # Display Answer and Sources | |
| st.write("### Answer:") | |
| st.write(answer) | |
| st.write("### Sources:") | |
| for doc in source_docs: | |
| st.write(f"Source: {doc.metadata.get('source', 'Unknown')}") | |
| except Exception as e: | |
| st.error(f"Error loading model or processing query: {e}") | |
| if __name__ == "__main__": | |
| main() | |