Subliminal Steering set for a COLM presentation on October 6
A lab member says George Morgulis’s steering-induced variant of subliminal learning in language models works “way more consistently.”
TLDR
A lab member said George Morgulis was set to present Subliminal Steering at COLM on October 6, from 11 a.m. to 1 p.m. in Franciscan A, #78. The post describes it as a steering-induced variant of subliminal learning in language models and says it works “way more consistently.”
Combined views
574
1 Source, first seen ago