1 paper · 1 filter
Haohui Wang, Jingyuan Qi, Jianpeng Chen +9
The rapid progress of large language models (LLMs) is fueled by the growing reliance on datasets that blend real and synthetic data. While synthetic data offers scalability and cos…