Mathstral's 7B LLM focusing on math reasoning and scientific discovery is intriguing! A model that can handle complex, multi-step logical reasoning is a huge step forward. I'm curious how it stacks up against other models in real-world applications. Has anyone here tried integrating it into their projects? Thoughts on its performance?