Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
michwilinski
searching Neon…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
5 ms
·
1.
▲
by
michwilinski
2y ago
interesting, but I don't agree that if we see the "token reasoning" chain it somehow explains how the model got the final answer. what if we trained deceiver models that would provide a sound chain of explanation but then per
2.
▲
by
michwilinski
2y ago
I mean, SSMs are in fact under the hood RNNs