JorGPT DeepSeek (English Version)

JorGPT DeepSeek is a desktop application for automatic grading and feedback of C programming exam answers using the DeepSeek LLM API. It is designed for university professors or teaching assistants who want to quickly evaluate student code submissions in bulk, using a customizable rubric and AI-powered analysis.

Features

• Bulk CSV Grading: Load a CSV file with student answers and process all submissions automatically.
• Customizable Rubric: The grading rubric and exam statement are editable within the app.
• AI-Powered Feedback: Uses DeepSeek's LLM to evaluate code and provide concise, category-based feedback and scores.
• Modern GUI: Built with PyQt5 for a user-friendly experience.
• Export Results: Saves results to an Excel file in the PUBLICATION folder.

Requirements

• Python 3.8+
• PyQt5
• pandas
• openai (for DeepSeek API)
• A DeepSeek API key (set as the environment variable DEEPSEEK_API_KEY)

Installation

1. Clone or download this repository.
2. Install dependencies:

   pip install -r requirements.txt

   Or manually:

   pip install PyQt5 pandas openai

3. Set your DeepSeek API key as an environment variable:
   • On Windows (PowerShell):

     $env:DEEPSEEK_API_KEY="your_deepseek_api_key"

   • On Linux/macOS:

     export DEEPSEEK_API_KEY="your_deepseek_api_key"

Usage

1. Run the application:

   python jorgpt_deepseek_v1.2.py

2. In the GUI:
   • Click Open CSV and select your CSV file with student answers.
   • The rubric and exam statement will appear on the right; you can edit them as needed.
   • Click Send to start grading. Progress and results will be shown in the left panel.
   • When finished, results are saved as an Excel file in the PUBLICATION folder.

CSV Format

• The CSV should have at least three columns: (e.g., Student Name, Exam Statement, Student Code)
• The program expects the exam statement in column 2 and the code in column 3 (zero-based index).

Customization

• You can edit the grading rubric and exam statement directly in the app before grading.
• The model used is currently fixed to deepseek-chat.

License

This project is for educational and research purposes. See LICENSE for details.

Acknowledgments

• Powered by DeepSeek LLM API.
• GUI built with PyQt5.

---

For questions or suggestions, please open an issue or contact the author.