Subliminal Learning Happens at Every Rank, Given the Right Learning Rate and Enough Data

·LessWrong··

Subliminal learning is the phenomenon where a language model picks up a behavioral trait—such as fondness for cats—by training on data from a trait-carrying teacher that looks entirely unrelated to the trait, such as bare sequences of numbers [1]. A wave of recent work has probed when this happens and what mechanism drives it [2][3][4][5][6][7], and part of that discourse concerns the conditions and dynamics under which subliminal learning occurs.Nief et al. [6] report that subliminal learning f...

Read full article →

Related Articles

Decoding the obfuscated bash script on a Uniqlo t-shirt
speerer · Hacker News · 11h ago
SWE-1.7 Reach Near GPT 5.5 and Opus Intelligence
mekpro · Hacker News · 3h ago
GitLost: We Tricked GitHub's AI Agent into Leaking Private Repos
ColinEberhardt · Hacker News · 14h ago
OpenBSD has a use-after-free allowing local privilege escalation to root
linggen · Hacker News · 6h ago
Chat Control 1.0 and 2.0 Explained
gasull · Hacker News · 1d ago