LLM-Powered Flashcard Generator

This project is a web-based application built with Streamlit that leverages Hugging Face's Large Language Models (LLMs) to automatically generate question-answer flashcards from provided text or PDF documents. It's designed to assist students and educators in creating study materials quickly and efficiently.

Features

• Flexible Input: Upload PDF files or plain text documents, or paste content directly into the application.
• AI-Powered Q&A Generation: Utilizes a T5-based model for question generation and a DistilBERT model for extractive question answering to create relevant flashcards.
• Customizable Output: Specify the desired number of flashcards to generate.
• Visually Appealing Interface: Interactive flashcards displayed in a grid with dynamic color schemes.
• Difficulty Indication: Each flashcard features a diagonally tilted strip in the top-right corner indicating a simulated difficulty level (Easy, Medium, Hard) with distinct colors.
• Export Functionality: Download generated flashcards as a CSV file for offline use or integration with other tools.
• Local Execution: Runs entirely locally, leveraging your machine's resources (CPU or GPU).

Setup

Follow these steps to set up and run the Flashcard Generator on your local machine.

Prerequisites

• Python 3.8+
• pip (Python package installer)

Installation

1. Clone the repository:

   git clone [https://github.com/ShubhamKukreti07/Flashcard-Generator.git](https://github.com/ShubhamKukreti07/Flashcard-Generator.git)
   cd Flashcard-Generator

2. Create a virtual environment (recommended):

   python -m venv venv

3. Activate the virtual environment:

   • On Windows:

     .\venv\Scripts\activate

   • On macOS/Linux:

     source venv/bin/activate

4. Install the required dependencies:

   pip install -r requirements.txt

   This will install Streamlit, PyPDF2, Transformers, PyTorch, and SentencePiece.

Usage

Once the setup is complete, you can run the Streamlit application.

1. Start the Streamlit application: Ensure your virtual environment is active.

   streamlit run app.py

   This command will open the application in your default web browser (usually at http://localhost:8501).

2. Interact with the application:

   • Choose Input Method: Select either "Upload File" to upload a PDF or .txt document, or "Paste Text" to directly input your content.
   • Adjust Number of Flashcards: Use the slider to specify how many flashcards you want to generate.
   • Generate Flashcards: Click the "Generate Flashcards " button. The application will process your content and display the generated Q&A pairs.
   • Review Difficulty: Observe the colored strips on each flashcard indicating its simulated difficulty.
     • White Strip: Easy Question
     • Yellow Strip: Medium Question
     • Black Strip: Hard Question (Note: Difficulty is currently simulated for demonstration purposes.)
   • Export: Use the "Download Flashcards as CSV ⬇" button to save your generated flashcards.

Project Structure

Flashcard-Generator

• app.py : Main Streamlit application file
• llm_model.py : Contains the QAGenerator class for LLM interactions
• requirements.txt : Lists all Python dependencies
• utils.py : Utility functions for text extraction and preprocessing
• README.md : This README file

Sample Outputs

Technologies Used

• Streamlit: For building the interactive web application.
• Hugging Face Transformers: For leveraging pre-trained LLMs for Question Generation (valhalla/t5-base-qg-hl) and Question Answering (distilbert-base-uncased-distilled-squad).
• PyTorch: The underlying deep learning framework used by the models.
• PyPDF2: For extracting text from PDF documents.
• Pandas: For easy handling and export of data to CSV.