Deep learning is getting out of control. Out of ~500 ICLR submissions, there over 100 that I should probably read. Definitely won't happen.
Maybe a stack of bidirectional LSTMs with attention could read them for me
-
-
you'd also need a neural lace so that the model can interface with and update your brain with what findings that it's read.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
or Joaquin Phoenix's voice style transfer on WaveNet.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Isn't there value to draw from other domains/disciplines to help inform DL research/creativity?Although, even more papers to read!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
dont forget to add residuals when stacking.
-
and also the 101 hacks like grad clipping, smart init, adam then sgd, etc. And the many more hacks that we haven't known yet.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.