This work studies low-budget black-box optimization, where unaided language models remain well below strong classical optimizers and attains the lowest regret on a sealed YouTube reward-tuning production benchmark.
Abstract
Can an agent learn a numerical search strategy through executable practice and then transfer that strategy as text? We study low-budget black-box optimization, where unaided language models remain well below strong classical optimizers. During development, an agent repeatedly writes and evaluates optimizer programs. It then distills the resulting program and practice record once into a 197-word primary Harness A, which is frozen before evaluation. Harness A reduces Gemini Flash regret by 48\% in an independent $N=30$ study ($p<.001$), enters the GP-BO performance range on the practice family, and lowers mean regret on all three held-out BBOB landscapes. The same text improves every tested Gemini executor and transfers to Claude Sonnet, reducing regret by 43\% and 49\% ($p\leq.005$). An independent end-to-end replication produces Harness B, a different program and text at the same performance tier. The same framework also attains the lowest regret on a sealed YouTube reward-tuning production benchmark. Executable practice is thus a viable way to discover a search policy, and language a portable medium for deploying it.
Adaptive-GEPA is introduced, which learns both how to divide requests and how to solve them, and it evolves a router and a library of specialist programs under one search budget.
Tian-Yu Chen, Yasi Zhang, Rui-Yi Wang et al.· arXiv.org· 0 citations
This work trains from random initialization through searched self-play on a single eight-GPU node for 2.5 days, and investigates search allocation, replay and restart-state selection, policy representation, progressive model sizing, quantized inference, and throughput engineering.
Evolutionary approaches to LLM-driven discovery often generate new programs from a small set of selected ancestors. This keeps contexts manageable but can omit useful evidence from other experiments, whereas including the full experimental history produces long, redundant contexts. We introduce a simple, single-agent d...
Bo Yuan, Wenqian Ye, Ze-Lin Zhao et al.· 0 citations
CraftBenchUE is introduced, an evaluation harness that runs agents in an isolated Unreal Engine environment, reconstructs their saved submissions in fresh projects, and applies deterministic build, asset, and runtime checks without an LLM judge.
Shu-Tong Wu, Kevin Calderone, A. Tsen· 0 citations
Together, these results recast the evolved harness as a legible compensation layer, shaped jointly by the language's engineering demands and the model's behavioral gaps, rather than an opaque benchmark-tuned scaffold.
Siqi Yang, Qianlan Yang, Yu-Xiong Wang et al.· 2 citations
V Vestrum, a framework that turns failures in execution traces into scoped harness changes without training the task model, is introduced, a framework that turns failures in execution traces into scoped harness changes without training the task model.
Jayant Parashar, Eugene F. Douglass, William C. Bastian et al.· arXiv.org· 0 citations
Related blog posts
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026