Google 论文:智能体 harness 自我改进时门控模块编辑准确率接近全量重写但 token 成本仅约十分之一
Google 论文《Harnesses That Improve Themselves》提出可自我改进的智能体 harness 统一框架,核心做法是每次只改 4 个可编辑组件中的 1 个,并通过验证集和 token 上限的门控机制决定是否保留。
New Google paper finds that an unchecked full rewrite of an agent harness matched accuracy but cost nearly 10 times more tokens per task than gated modular edits.
that self-improving agent harnesses work best when they change 1 code module at a time and must pass validation.
The harness is the code around a model that manages context, runs tools, and checks work before stopping. A harness tuned for 1 model often stalls or wastes tokens on another.
They let an optimizer model improve a 142-line coding harness by editing 1 of its 4 parts at a time. Each edit had to pass a validation set and a token cap, or it was rejected.
On 40 held-out SWE-bench Lite tasks with Gemini 3.8 Flash, the gated harness solved 40.0% against 5.0% for the seed, using 46.7k tokens per task. A full unchecked rewrite solved 42.5% but used 451.6k tokens, and prompt rules dropped accuracy to 2.5%.
If you let agents tune their harness, gate each change on held-out tasks and token cost, and enforce rules in code, not prompts.
来源:Rohan Paul · x.com