This is a special post for quick takes by cloud. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Can subliminal learning occur when the base models had different inits but are trained to be similar? (Clarifies whether init is what matters)
Develop theory
Quantify transmission via random matrix theory (build off equation 2 in the paper). Are there nice relationships lurking there (like d_vocab : d_model)?
Can we get theory that covers the data filtering case?
Figure out what can and can’t be transmitted
Backdoor transmission
Information-theoretic limits
Dependence on tokenization
Subtle semantic transmission: what about cases that aren't subliminal learning but are very hard to detect? Connect to scalable oversight and/or control.
Adversarially-constructed subliminal learning datasets (no teacher) (compare with "clean label" data poisoning literature)