Distilled version of Qwen 2.5 7B using reasoning data generated by DeepSeek R1 for enhanced performance.