Today I learned that having a bug in your framework's dropout implementation can make you waste a lot of compute time.
-
@fchollet yes, there's too many shortcomings for now... But now I want stateful layers in Keras :)
@seaandsailor It's an open issue. If you need it now, you can just implement it: https://github.com/fchollet/keras/issues/98#issuecomment-122776799 …
-
-
@fchollet I added this to the list of things I'll try to do after I submit my thesis proposal to the evaluation committee.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@fchollet Would adding a "sequential MSE" objective function work for keeping state during different timesteps for training? -
@fchollet By that, I mean having a layer that returns sequences as the output layer, and computing the cost as the mean MSE over timesteps. - Show replies
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.