Ciarán M. Gilligan-Lee

Titles as the sources gave them. In 1 entry, with 0 checked quotations.

In the Base

Every entry they appear in.

  1. ResearcharXivWrote it

    Defence against the dark arts of distillation: Unsupervised discovery of subliminal learning

    According to the abstract, a student distilled from a teacher with shared base parameters can inherit the teacher's traits, including misalignment, even from unrelated data; the authors say they prove such a trait can be recovered without supervision.