app.py

# Install necessary libraries
#!pip install PyPDF2 transformers torch accelerate streamlit

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
import PyPDF2
import streamlit as st

# Function to extract text from PDF
def extract_text_from_pdf(uploaded_file):
    pdf_text = ""
    reader = PyPDF2.PdfReader(uploaded_file)
    for page_num in range(len(reader.pages)):
        page = reader.pages[page_num]
        pdf_text += page.extract_text()
    return pdf_text

# Initialize the tokenizer and model on CPU first
tokenizer = AutoTokenizer.from_pretrained("ricepaper/vi-gemma-2b-RAG")

model = AutoModelForCausalLM.from_pretrained(
    "ricepaper/vi-gemma-2b-RAG",
    torch_dtype=torch.bfloat16
)

# Move model to GPU if available
if torch.cuda.is_available():
    model.to("cuda")

# Define the prompt format for the model
prompt = """
### Instruction and Input:
Based on the following context/document:
{}
Please answer the question: {}
### Response:
{}
"""

# Function to generate answer based on query and context
def generate_answer(context, query):
    input_text = prompt.format(context, query, "")
    input_ids = tokenizer(input_text, return_tensors="pt", truncation=True, max_length=1024)

    # Use GPU for input ids if available
    if torch.cuda.is_available():
        input_ids = input_ids.to("cuda")

    # Generate text using the model
    outputs = model.generate(
        **input_ids,
        max_new_tokens=500,
        no_repeat_ngram_size=5,
    )

    # Decode and print the results
    answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return answer

# Streamlit App
st.title("RAG-Based PDF Question Answering Application")

# Upload PDF
uploaded_file = st.file_uploader("Upload a PDF file", type="pdf")

if uploaded_file is not None:
    # Extract text from the uploaded PDF
    pdf_text = extract_text_from_pdf(uploaded_file)

    st.write("Extracted text from PDF:")
    st.text_area("PDF Content", pdf_text, height=200)

    # User inputs their question
    query = st.text_input("Enter your question about the PDF content:")

    if st.button("Get Answer"):
        if query.strip() != "":
            # Generate answer based on extracted PDF text and the query
            answer = generate_answer(pdf_text, query)
            st.write("Answer:", answer)
        else:
            st.warning("Please enter a question.")
else:
    st.info("Please upload a PDF file to get started.")