This repository contains a gguf fine-tuned version of the Qwen-2 0.5B model specifically optimized for mathematical instruction understanding and reasoning. It builds upon the Numina dataset, which provides a rich source of mathematical problems and solutions designed to enhance reasoning capabilities even in smaller language models.

My primary motivation is the hypothesis that high-quality datasets focused on mathematical reasoning can significantly improve the performance of smaller models on tasks that require logical deduction and problem-solving. Uploading comprehensive benchmarks is the next step in evaluating this claim.