Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
fakeparmesean
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
3 ms
·
1.
▲
by
fakeparmesean
2y ago
My understanding coming from mechanistic interpretability is that models are typically (or always) in superposition, meaning that most or all neurons are forced to encode semantically unrelated concepts because there are more concepts than