The Decoder· Jonathan Kemper·· 3 天前AI 评分69
Sakana AI 提出 MASS 自改进方法，模型自我评估与优化仍有共享盲区
Sakana AI's self-improving model gets better at solving and judging, but can't catch its own blind spots
AI 导读
Sakana AI 与 UC Berkeley 提出多智能体自监督方法 MASS，让同一个语言模型同时承担解题、优化工作流和评估结果三个角色，实现递归自我改进，无需更强的教师模型。
来源：The Decoder · the-decoder.com