Dongxi 东锡 NLP· @dongxi_nlp · X·· 2 小时前AI 评分60
AI 导读
@OwainEvans_UK 团队发布新论文 Beyond Owls: Subliminal Learning Can Transfer Learned Capabilities and Backdoors，此前他们的 Subliminal Learning 研究显示模型能通过数字序列传递对猫头鹰的偏好。新工作表明模型可传递更复杂的特性：新技能、智能体破解行为和后门，其中后门在数据中既无触发条件也无对应行为仍能完成传递。
正文
推荐阅读，@OwainEvans_UK 团队的所有文章篇篇精彩！
Beyond Owls:
Subliminal Learning Can Transfer Learned Capabilities and Backdoors
New paper. Previous: Models transfer a love of owls through sequences of numbers (Subliminal Learning). New: We show models can transfer more complex traits: novel skills, agentic hacking, backdoors. Here, a backdoor transfers with neither trigger nor behavior in the data!在 X 查看被引用的帖子
来源：Dongxi 东锡 NLP · x.com