CPU-ARC：将 LLM 推理轨迹编译为可复用缓存，调用成本降至 1/16-1/291
CPU-ARC: Agentic Reasoning Caching for on-CPU Reasoning Reuse
一篇挺有意思的论文：让 LLM 只推理一次，之后把解题过程存成 CPU 能直接跑的缓存，便宜 16 到 291 倍还不用 GPU。做推理成本优化的可以看看。
arXiv 论文提出 agentic reasoning caching（ARC）方法，把 LLM 的推理轨迹转化为可在 CPU 上执行、可复用的推理缓存（ARCs）。这样任务只需用 LLM 推理一次，之后新实例直接由缓存执行，很少需要模型参与。在数学和科学类难题上，ARCs 与前沿 LLM 性能相当，且具备可解释、可持续训练、对输入扰动稳健的特点。调用成本比源模型低 16-291 倍。
CPU-ARC: Agentic Reasoning Caching for on-CPU Reasoning Reuse
Large language models reinvent the wheel all the time. They conflate task-solving with the execution of the solutions, which forces them to solve every problem from scratch. We show that once a task has been reasoned through, the application of its solution to a new problem instance can be done very efficiently and with little involvement from language models. We introduce agentic reasoning caching, a method for constructing robust CPU-executable agentic reasoning caches (ARCs) from LLM reasoning traces. ARCs show on-par performance with frontier LLMs across a range of difficult mathematical and scientific tasks, are fully explainable, continually trainable, robust to input perturbations, and 16-291x cheaper to invoke than the source model.