The Decoder· Jonathan Kemper·· 3 小时前AI 评分39
Sakana AI 提出 MASS：单模型自监督在多角色上取得进步，但难以发现自身盲区
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 Multi-Agent Self-Supervision（MASS）方法，让单一大语言模型同时承担解题、优化和评估三种角色，实现无更强教师模型的递归自我改进。
来源：The Decoder · the-decoder.com