//
sign in
Post
by @danabra.mov
PostEmbed
by @danabra.mov
Record
by @jimpick.com
Record
by @atsui.org
+ new component
Post
🚨New preprint!🚨 We know that LM representations can be used to predict brain responses to language. But what *features* of these representations underlie this alignment? We use SAEs to find out!
2d
Michael Lepori