发表机构
M365 Research, Microsoft（微软M365研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
FOCUS提出免训练的测试时上下文压缩方法，通过因果决策保持重新定义压缩，无需训练即可在多种智能体基准上削减峰值上下文48%、依赖73%，并提升任务成功率8.9个百分点。
AI 中文摘要
LLM智能体积累的交互历史随任务长度线性增长，导致推理成本呈二次方缩放，并因注意力稀释而使性能下降。现有的上下文压缩方法通过离线学习来决定丢弃什么：例如对比优化准则、蒸馏压缩器或训练压缩策略。这会产生大量成本。此外，压缩策略是先验学习的，不能根据测试时轨迹的动态变化进行条件调整。本文提出一个互补性问题：哪些过去的交互在因果上塑造了智能体未来的决策？我们将上下文压缩重新定义为离散交互单元上的因果决策保持问题，并引入FOCUS，一个完全在测试时运行的免训练上下文压缩框架。我们的方法无需离线数据收集或微调，且与架构无关，可作为模块化压缩层附加到任何闭源API前沿模型上。我们在包括API和工具调用、问答、网页领域和多轮对话的多种智能体基准上评估FOCUS。我们的方法建立了新的最先进性能，将峰值上下文削减高达48%，依赖削减73%，同时将任务成功率比未压缩执行提高高达8.9个百分点。
英文摘要
LLM agents accumulate interaction histories that grow linearly with task length, causing quadratic inference cost scaling and performance degradation from attention dilution. Existing context-compression methods learn what to discard offline: by contrastively optimizing guidelines, distilling compressors, or training compression policies. This incurs a substantial cost. Further, the compression policy is learned a priori and is not dynamically conditioned on the evolving test-time trajectories. In this paper we ask a complementary question: Which past interactions causally shape the agent's future decisions? We recast context compression as a causal decision preservation problem over discrete interaction units and introduce FOCUS, a training-free context compression framework that operates entirely at test time. Our method requires no offline data collection or fine-tuning, and is architecture-agnostic, attaching to any closed-API frontier model as a modular compression layer. We evaluate FOCUS on diverse agentic benchmarks including API and tool-calling, QA, web domain and multi-turn dialogue. Our method establishes new state of the art performance, cutting peak context by up to 48% and dependency by 73% while improving task success by up to 8.9 percentage points over uncompressed execution.
CommentsPreprint. Under Review
