Show this thread
A Keras workflow you need to be familiar with is overriding `train_step`. This enables you to leverage `fit` and everything it supports (logging, distributed training, callbacks, performance optimizations) with custom training loops. For instance, here's a GAN implementation.pic.twitter.com/3bXxQ2IqAc
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Indeed an important feature. It helped me over the weekend find the reason for the error that I was getting in http://model.fit I implemented train_step, added logging and was able to find out the problem.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
In this code he is training generator to minimize 1 - logD(G(z)) right?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
How do you modify this GAN to use WAV audio files instead of images?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Kerasnis very simple to use but simplicity was not helping in the cases of errors. This new changes are going to transform the way we thought of keras. Thank you for this consistent effort
@fcholletThanks. Twitter will use this to make your timeline better. UndoUndo
-
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.