Lex Fridman Podcast
#452 – Dario Amodei: Anthropic CEO on Claude, AGI & the Future of AI & Humanity
But scaling monosemanticity sort of, I think, was significant evidence that even for very large models, and we did it on Claude 3 sauna, which at that point was one of our production models, you know, even these models seem to be very, you know, seem to be substantially explained, at least, by linear features and, you know, doing dictionary learning on them works.
0
💬
0
Comments
Log in to comment.
There are no comments yet.