AI 中文总结
该研究针对现有自改进编码智能体仅利用单轨迹的缺陷，提出孟德尔哥德尔机，新增反应规范突变与跨谱系杂交两种自修改策略，经理论证明与实验验证，其在编码任务上的性能、效率及泛化性均优于基线方法。
AI 中文摘要
能够迭代重写自身源代码的自改进编码智能体在编码任务上已展现出优异性能，但现有方案通常仅基于单次失败轨迹推导自修改，忽略了智能体不断扩充的过往尝试档案中可利用的丰富比较信号。依据受控遗传的孟德尔原理，我们提出孟德尔哥德尔机（Mendel Gödel Machine，简称MGM）。除了常规的单轨迹克隆突变外，MGM还包含两种新的自修改类型以更好利用积累的证据：反应规范突变基于智能体在多个任务上的轨迹同时对其进行编辑，跨谱系杂交则利用另一谱系参考智能体在同一任务上的轨迹对智能体进行编辑。在加性适应度景观模型下，我们从理论上证明并通过受控代理模拟验证，这些新策略相比单轨迹基线能实现更快、更优的收敛。在SWE-bench和Polyglot上开展的实验证实，MGM在性能、效率和泛化性上均实现了持续提升。
英文摘要
Self-improving coding agents that iteratively rewrite their own source code have demonstrated impressive performance on coding tasks. However, existing solutions generally derive self-modification from a single failure trajectory at a time, overlooking rich comparative signals available in the agent's expanding archive of past attempts. According to Mendelian principles of controlled inheritance, we introduce Mendel Gödel Machine (MGM). In addition to the general single-trajectory clonal mutation, MGM includes two new types of self-modification that better utilizes evidences accumulated: the reaction-norm mutation edits an agent based on its trajectories on multiple tasks simultaneously, and the cross-lineage hybridization edits an agent using the trajectory of a reference agent from another lineage on the same task. Under an additive fitness landscape model, we prove theoretically and demonstrate via controlled surrogate simulation that the new strategies facilitate a faster and better convergence over single-trajectory baselines. Experiments on SWE-bench and Polyglot confirm MGM's consistent improvement in performance, efficiency, and generalizability.
CommentsProject Page at https://reallcz.github.io/MGM; Code at https://github.com/RealLcz/MGM
