|
|
|
|
|
|
by peyton
970 days ago
|
|
|
I dunno, failure seems okay. Wouldn’t expect a better paradigm to beat SOTA at first. It’s totally plausible that neurons use eg. transposons in a way we don’t yet have the instrument resolution to characterize, which would suggest that you don’t need 1000 layers, but a lookup table or something. |
|
|