使用基于自信息的内容过滤增强 LLMs 的上下文效率：揭示 LLMs 的上下文约束（英文版）

英文标题：Unlocking Context Constraints of LLMs: Enhancing Context Efficiency of LLMs with Self-Information-Based Content Filtering中文摘要：本文提出了一种名为 “选择性上下文” 的方法，利用自身信息来过滤 less informative 的内容，并在不同数据源上展示了提高固定上下文长度效率的有效性。英文摘要：Large language models (LLMs) have received significant attention by achievingremarkable performance across various tasks. However, their fixed contextlength poses challenges when processing long documents or maintaining extendedconversations. This paper proposes a method called \te