Meta^n：通过涌现深度实现递归自我改进

摘要

自我改进的 LLM 智能体改进的是答案，而非产生这些答案的过程。增加元层次的系统会使该层次保持固定；而那些自我编辑的系统则必须保留自身编辑机制的一部分不被改动以维持稳定，从而将其实际实现的元深度上限限制在约二层。我们提出 Meta^n，它保持元操作固定，转而对其输入进行递归。该操作 Ω 反复应用于自身的产物：读取下方求解器栈的轨迹以及生成这些轨迹的代码，然后把下一层编写为策略性预处理模块和可调用辅助函数库。由于 Ω 从不改变，它不会破坏系统稳定性；又因为其输入严格递增，每一层都比上一层从更高的视角进行推理。深度由收敛性决定，而非预先设定；一个演化档案库对层链进行搜索。在两种骨干模型上，Meta^n 在全部八个基准测试族上均优于先前的自我改进智能体。最突出的案例是 ARC-AGI-2——该基准测试旨在抵制技能记忆——在其中唯有 Meta^n 得分高于零。消融实验表明，递归带来的收益大部分来自每一层传递给下一层的条件信息；并且尽管没有任何提示词加以规定，不同层的角色会随深度涌现。代码见 https://github.com/minnesotanlp/meta-n

English

Self-improving LLM agents refine answers, not the process that produces those answers. Systems that add a meta-level hold that level fixed, and those that edit themselves must leave part of their own editing machinery untouched to stay stable, capping the meta-depth they realize at roughly two. We present Meta^n, which keeps the meta-operation fixed and recurses on its input instead. That operation, Ω, is applied repeatedly to its own products, reading the traces of the solver stack below together with the code that produced them, then writing the next layer as a strategic pre-process and a library of callable helpers. Because Ω never changes, it cannot destabilize the system, and because its input strictly grows, each layer reasons from a higher vantage than the last. Depth is set by convergence rather than fixed in advance, and an evolutionary archive searches over layer chains. Across two backbones, Meta^n outperforms prior self-improving agents on all eight benchmark families. The sharpest case is ARC-AGI-2, built to resist skill memorization, where it alone scores above zero. Ablations indicate that most of the gain from recursion comes from the conditioning each layer passes to the next, and distinct layer roles emerge with depth although no prompt prescribes them. Code available at https://github.com/minnesotanlp/meta-n