LeituraPesquisaRadarFramework de investimento
Entrar / Cadastrar
← Voltar para a leitura
NotíciasTópico · Harness²框架Material1 min de leituraEditorial DisconfirmAI

Google Open-Sources Harness²: Enabling AI Agents to Self-Optimize Execution Scaffolds at Runtime

Google open-sourced the Harness² framework, which uses dual-layer recursion for runtime self-improvement of AI agents; internal tests show GLM-5 scores significantly improved, surpassing native Gemini 3.5 Flash.

2026-10-19 11:11 GMT+8

The Google Research team has open-sourced Harness², a Recursive Self-Improvement (RSI) framework. This system allows large language models to autonomously modify their external code, system prompts, toolchains, and skill libraries during task execution, aiming to address the challenge of effective self-optimization in real-world open environments where ground truth is often unavailable.

Harness² employs a dual-layer recursive architecture. The first layer, task-level recursion, utilizes a 'Propose-Probe-Compose' mechanism to evaluate scaffold modifications via behavioral contrast without external judges. The second layer, improvement-level recursion, uses a reflection module to distill debugging records into an experience library, enabling the system to retrieve effective components from past tasks for continuous learning across new assignments.

According to Google's internal benchmarks, on the WorkBuddy-Bench office productivity benchmark, the GLM-5 model equipped with Harness² saw its composite score rise from 64.2 to 75.7, an increase of 11.5 points, surpassing the native Gemini 3.5 Flash (72.4 points) which did not use the framework. Additionally, the framework delivered performance gains of 4.5 points on JobBench and 11.8 percentage points on the LAB legal benchmark. The project currently supports OpenCode and Codex environments.

Fontes

Entrar / Cadastrar
LeituraPesquisaRadarFramework de investimento
qbitai.com
↗
Tópico

Harness²框架

谷歌开源了名为 Harness² 的递归智能体框架,旨在解决大模型在开放世界任务中执行脚手架(Harness)难以自动优化的问题。该框架通过“任务级”与“改进级”双层递归机制,让 AI 代理在无标准答案的情况下自主修改提示词、工具链等组件,…

2 etapas1 peçasAtualizado 2026-10-20 07:48

Como chegamos aqui

  1. 2026-10-19谷歌发布并开源 Harness² 框架,提出双层递归架构以实现在无地面真值反馈下的运行时自我改进。qbitai.com ↗
  2. 2026-10-19GitHub 仓库 google-research/harness2 上线,提供 Apache-2.0 许可的代码及 OpenCode/Codex 支持。github.com ↗

O que ainda está indefinido

  • Harness² 的经验库跨领域迁移效果在更复杂的真实业务场景中是否依然稳定?
  • 随着并行递归预算 k 值的增加,计算成本与性能提升之间的边际效益曲线如何变化?
  • 该框架对非谷歌生态模型(如本地部署模型)的支持程度及兼容性表现如何?
Ler o tópico completo →
Continuar pesquisando →