LongFaith: Enhancing Long-Context Reasoning in LLMs with Faithful Synthetic Data

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…