Long-term memory is essential for large language model (LLM) agents operating in complex environments, yet existing memory designs are either task-specific and non-transferable, or task-agnostic bu
一句话总结:这篇文章揭示了当前 LLM 智能体在处理长期、复杂交互任务时记忆能力的不足,并提供了一个更具挑战性的评估平台来推动该领域的进步 已有benchmark的特点 Large language model (LLM) agents have two complementary core capabilities: the ability to memorize task-relevant
Paper Link https://dl.acm.org/doi/pdf/10.1145/227531.227544 一篇1996年的老文章。作为课程学习阅读。 这篇文章《Airport 95: Automated Baggage System?》由 A. John Swartz 撰写,主要分析了丹佛国际机场(DIA)自动行李处理系统的失败案例。文章以该项目为研究对象,探讨了大型软件/系统工程中
MINEDRAFT: A Framework for Batch Parallel Speculative Decoding 把推测解码打成mini batch,随后在drafter和verifier上分批处理。 在vllm上修改,工程量可观。不错的尝试和idea。 Architecture overview of MINEDRAFT. (Left) The Scheduler manages r