NOVA: A Verification-Aware Agent Harness for Architecture Evolution in Industrial Recommender Systems

作者：Shaohua Liu, Liang Fang, Yilong Sun, Shudong Huang, Qin-Wen Luo, Shaoxin Liu, Xiaoyang Chen, Dongqiang Liu, Chuangang Ma, Z. Chai, Heng Wang, Shijie Quan, Chang Cui, Zhangbin Zhu, Peng Chen, Wei Xu, Lei Xiao, Haijie Gu, Jie Jiang · 发表于：arXiv.org · 年份：2026 · DOI：10.48550/arxiv.2606.27243 · 被引用次数：4 · 研究领域：Computer Science

Industrial advertising recommender models are continuously improved through architecture evolution. Upgrades such as RankMixer, TokenMixer-Large, and MixFormer show that better structures remain a key source of quality and business gains. Yet developing such upgrades in production is expert-intensive and difficult to scale. Existing automation is insufficient: AutoML mainly tunes hyper-parameters, while effective gains often require cross-module changes under strict constraints; generic LLM coding agents optimize for runnable code, but runnable code does not imply a valid recommender architecture. Candidates may pass local tests while causing silent failures that degrade performance. We present NOVA, a level-aware agent harness for verification-aware architecture evolution. NOVA uses an architecture gradient, an SGD-inspired, non-differentiable update signal that aggregates prior modifications, verification diagnostics, metric feedback, and trajectory memory to guide the next modification. A verification cascade checks structure semantics, local executability, offline effectiveness, and online impact; invalid candidates are blocked early, with failure patterns recorded as forbidden directions. L1--L4 task-level control matches automation to task complexity and risk, routing high-risk tasks to Copilot for human oversight. Deployed in an industrial advertising system, NOVA achieves the highest effective pass rate on L2 ScaleUp and L3 Literature-to-Production tasks (54.5% and 60...