Hacker News· matt_d·· 3 小时前AI 评分60
Reactive Optimizers:训练中让智能体重写优化器源码
Reactive Optimizers: rewrite the optimizer's source code during training
AI 导读
Xingjian Bai 等人提出 Reactive Optimizers,把智能体放进优化器内部,由它观察训练动态并直接改写优化器源码,实现 OPAL(OPtimizer with an Agent in the Loop)。
来源:Hacker News · xingjianbai.com