LessWrong5 min read
Fun toy result: a subliminal-learning effect via on-policy distillation
In the standard subliminal learning setup, a teacher model with a preference generates data according to its policy, and then the student is trained on those data.
By shawnghuLessWrong

Read the full story
Originally published by LessWrong on Nov 4, 2026.
Comments
Sign in to join the discussion.