TensorsOfTheWallA place where I convince neural networks that pixels mean something.
I work on:
I work on:
Who needs all their neurons anyway? This network’s motto: ‘Do less, compute more'.
This isn't quite dropout regularization, but it looks cool.
Oh... you're still here?
Maybe you'll see something cool if you refresh the page?
Maybe you'll see something cool if you refresh the page?
Curiosity scales better than models.