The Decoder· Jonathan Kemper·· 22 小时前AI 评分62
Google 研究者提出 RRSI,防止自我改进智能体记住测试任务
Google researchers find a way to keep self-improving AI agents from memorizing their tests
AI 导读
Google 研究者提出 RRSI(Regularized Recursive Self-Improvement of Agent Harnesses),通过收缩编辑预算和严格 critic 限制智能体对测试任务的记忆。
来源:The Decoder · the-decoder.com