--- language: en license: mit tags: - rubric-generation - education - fine-tuned - text-generation - gpt datasets: - custom widget: - text: >- Question: What are the benefits of regular exercise? Sample Answer: Regular exercise helps in weight management, improves cardiovascular health, and enhances mental well-being. Total Marks: 5 base_model: - meta-llama/Llama-2-7b-chat-hf --- # 📝 Rubric Generator Model This model is fine-tuned to **generate detailed grading rubrics** when provided with: - a **question or prompt** - a **sample answer** - the **maximum marks** for the question It is designed for educators, examiners, and educational apps that require structured, point-wise rubrics for evaluating subjective answers. --- ## 📌 Model Details - **Architecture**: Causal language model (e.g., GPT-style) - **Training Format**: Supervised fine-tuning on question-answer-mark-rubric datasets - **Input Format**: ```plaintext Question: Sample Answer: Total Marks: ```` * **Output**: A rubric in JSON format assigning marks to specific answer criteria --- ## 🚀 Example **Input:** ``` Question: What are the advantages of using solar energy? Sample Answer: Solar energy is renewable and reduces electricity bills. It's environmentally friendly and reduces reliance on fossil fuels. Total Marks: 5 ``` **Output:** ```json { "rubric": [ { "criteria": "Mentions that solar energy is renewable", "max_marks": 1 }, { "criteria": "Discusses cost-saving or reduction in electricity bills", "max_marks": 1 }, { "criteria": "Highlights environmental friendliness", "max_marks": 1 }, { "criteria": "Mentions reduced reliance on fossil fuels", "max_marks": 1 }, { "criteria": "Answer clarity and overall relevance", "max_marks": 1 } ] } ``` --- ## 📚 Training Data The model was fine-tuned on a dataset of: * Questions * Sample answers * Total marks * Expert-designed rubrics in structured JSON format This dataset is included in the accompanying [GitHub repository](https://github.com/yourusername/rubric-generator). --- ## 🧠 Intended Use * Automated rubric generation for educational platforms * Consistent scoring guidelines for subjective assessments * Feedback generation tools for students and teachers --- ## ⚠️ Limitations * May not be optimized for highly domain-specific or creative writing assessments * Requires sample answers to be reasonably well-formed to generate useful rubrics * Rubric quality depends on clarity of the question and sample answer --- ## 📄 License This model is licensed under the [MIT License](LICENSE). --- **Author**: Mrigank Singh **Contact**: mriganksingh005@gmail.com **Repository**: [GitHub - rubric-generator](https://github.com/Mrigank005/Rubric_Generator) ```