1 paper
Jingdi Lei, Di Zhang, Junxian Li +7
Large language models increasingly need to accumulate and reuse historical information in long-term assistants and agent systems. Simply expanding the context window is costly and…