Learn how to use scaling laws to balance data in Multilingual LLMs, reducing performance gaps between high and low-resource languages while saving compute.