Qingyu Liu

Code-executing agents can autonomously generate new, solvable math problems that are harder than existing ones, offering a scalable solution to the bottleneck of high-quality training data for advanced LLMs.