Update app.py

app.py

@@ -138,95 +138,46 @@
 
 
 
-import streamlit as st
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.vectorstores import FAISS
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain.document_loaders import PyPDFLoader
-from langchain.chains import RetrievalQA
-from langchain.llms import HuggingFaceHub
-import tempfile
-import os
-
-# Constants
-EMBEDDING_MODEL_NAME = "BAAI/bge-base-en-v1.5"
-LLM_MODEL_REPO = "mistralai/Mistral-7B-Instruct-v0.1"
-CHUNK_SIZE = 500
-CHUNK_OVERLAP = 300
-
-# Load and split documents
-def load_and_split_pdf(pdf_file):
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp_file:
-        tmp_file.write(pdf_file.read())
-        tmp_file_path = tmp_file.name
-
-    loader = PyPDFLoader(tmp_file_path)
-    documents = loader.load()
-    splitter = RecursiveCharacterTextSplitter(chunk_size=CHUNK_SIZE, chunk_overlap=CHUNK_OVERLAP)
-    chunks = splitter.split_documents(documents)
-    return chunks
-
-# Create FAISS vectorstore
-def build_vectorstore(chunks):
-    embeddings = HuggingFaceEmbeddings(model_name=EMBEDDING_MODEL_NAME)
-    db = FAISS.from_documents(chunks, embedding=embeddings)
-    return db
-
-# Initialize LLM from Hugging Face Hub
-def get_llm():
-    return HuggingFaceHub(
-        repo_id=LLM_MODEL_REPO,
-        model_kwargs={"temperature": 0.3, "max_new_tokens": 512, "top_k": 10}
-    )
-
-# Custom prompt for better accuracy
-CUSTOM_PROMPT = """
-You are a professional resume chatbot. Use the context below to accurately and concisely answer the user's question. If the information is not available in the context, say "Not found in the document.".
-
-Context:
-{context}
-
-Question:
-{question}
-
-Answer:
-"""
-
-# Build QA chain
-def build_qa_chain(vectorstore):
-    return RetrievalQA.from_chain_type(
-        llm=get_llm(),
-        retriever=vectorstore.as_retriever(),
-        chain_type="stuff",
-        chain_type_kwargs={
-            "prompt": CUSTOM_PROMPT
-        }
-    )
-
-# Streamlit UI
-def main():
-    st.set_page_config(page_title="Resume Q&A Bot", layout="wide")
-    st.title("Resume Chatbot - Ask Anything About the Uploaded PDF")
-
-    uploaded_file = st.file_uploader("Upload your resume (PDF)", type="pdf")
-
-    if uploaded_file is not None:
-        st.success("PDF uploaded successfully!")
-        with st.spinner("Processing document and creating knowledge base..."):
-            chunks = load_and_split_pdf(uploaded_file)
-            vectorstore = build_vectorstore(chunks)
-            qa_chain = build_qa_chain(vectorstore)
-
-        st.success("Knowledge base ready! Ask your question below:")
-
-        question = st.text_input("Your Question:")
-
-        if question:
-            with st.spinner("Generating answer..."):
-                response = qa_chain.run(question)
-                st.markdown(f"**Answer:** {response}")
-
-if __name__ == '__main__':
-    main()
+import streamlit as st from langchain_community.document_loaders import PyPDFLoader from langchain.text_splitter import RecursiveCharacterTextSplitter from langchain_community.vectorstores import FAISS from langchain.embeddings import HuggingFaceEmbeddings from langchain.chains import RetrievalQA from langchain.prompts import PromptTemplate from langchain.llms import HuggingFaceHub import os
 
+Set Hugging Face API Token
+
+os.environ["HUGGINGFACEHUB_API_TOKEN"] = "your_huggingfacehub_api_token_here"
+
+Custom Prompt
+
+custom_prompt = PromptTemplate( input_variables=["context", "question"], template=""" You are a helpful assistant. Use the context below to answer the question. If the answer is not in the context, say "I don't know."
+
+Context: {context}
+
+Question: {question}
+
+Answer: """ )
+
+Load PDF and split into chunks
+
+def load_and_split_pdf(uploaded_file): loader = PyPDFLoader(uploaded_file.name) documents = loader.load() text_splitter = RecursiveCharacterTextSplitter(chunk_size=800, chunk_overlap=100) chunks = text_splitter.split_documents(documents) return chunks
+
+Build vectorstore from document chunks
+
+def build_vectorstore(chunks): embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2") db = FAISS.from_documents(chunks, embedding=embeddings) return db
+
+Build QA chain
+
+def build_qa_chain(vectorstore): llm = HuggingFaceHub(repo_id="mistralai/Mistral-7B-Instruct-v0.1", model_kwargs={"temperature": 0.2, "max_length": 512}) qa_chain = RetrievalQA.from_chain_type( llm=llm, retriever=vectorstore.as_retriever(), chain_type="stuff", chain_type_kwargs={"prompt": custom_prompt} ) return qa_chain
+
+Streamlit App
+
+st.set_page_config(page_title="Accurate PDF Chatbot", layout="centered") st.title("PDF QA Chatbot - RAG Powered")
+
+uploaded_file = st.file_uploader("Upload a PDF", type="pdf")
+
+if uploaded_file: with st.spinner("Reading and processing PDF..."): chunks = load_and_split_pdf(uploaded_file) vectorstore = build_vectorstore(chunks) qa_chain = build_qa_chain(vectorstore) st.success("PDF processed. Ask your question below.")
+
+question = st.text_input("Ask a question from the PDF:")
+
+if question:
+    with st.spinner("Searching answer..."):
+        answer = qa_chain.run(question)
+        st.markdown(f"**Answer:** {answer}")