Charlie Steiner comments on How “Discovering Latent Knowledge in Language Models Without Supervision” Fits Into a Broader Alignment Scheme

[Charlie Steiner]

Just what I wanted :D