MathScale.

Scaling instruction tuning for mathematical reasoning.

Large language models (LLMs) have demonstrated remarkable capabilities in problem-solving.

---

Join the discussion on this paper page.