Long-context understanding requires large language models (LLMs) to reason over lengthy documents, conversations, and code, yet task-relevant evidence is oft…
机构:百度
来源:arXiv 2609.31382 | AI4Papers 论文推荐平台