NaruMaru/ege-checker-qwen2p5-0p5b-demo
The NaruMaru/ege-checker-qwen2p5-0p5b-demo is a 0.5 billion parameter model based on the Qwen2.5-0.5B-Instruct architecture, fine-tuned for checking solutions to EGE (Unified State Exam) math problems. It provides a verdict of 'correct' or 'incorrect' along with a brief explanation for student solutions. This model is specifically designed to evaluate mathematical problem-solving steps against given conditions, offering a specialized application in educational assessment.
Loading preview...
Overview
NaruMaru/ege-checker-qwen2p5-0p5b-demo is a specialized language model designed to act as a checker for EGE (Unified State Exam) math solutions. Built upon the Qwen/Qwen2.5-0.5B-Instruct base model, it has been fine-tuned using LoRA SFT on synthetic correct and incorrect solutions to demonstrate its approach to automated assessment.
Key Capabilities
- Solution Verification: Evaluates student-provided solutions to math problems.
- Verdict Generation: Outputs a clear 'correct' or 'incorrect' verdict for each solution.
- Brief Explanation: Provides a concise explanation accompanying the verdict.
Limitations and Use Cases
This model is currently a demonstration of concept, primarily trained on synthetic data. While effective for showcasing the approach, it is not production-ready for real-world educational settings without further fine-tuning on actual student work. Its primary utility lies in demonstrating how LLMs can be adapted for specific, structured assessment tasks, particularly in mathematics. For production deployment, additional training with real student data is recommended to enhance accuracy and robustness.