1 paper
Cehao Yang, Xueyuan Lin, Chengjin Xu +5
Despite the growing development of long-context large language models (LLMs), data-centric approaches relying on synthetic data have been hindered by issues related to faithfulness…