Essentially: - As a TF user, leverage the high-level functionality of tf.keras to define your TF models - If you're using Keras on top of TF and you seek better integration with TF features, switch to tf.keras
Show this thread
Let's be clear: there isn't any "TF *or* Keras" choice to make. Keras is an API, rather than a standalone framework, and it's the official high-level API of TF (via tf.keras). You should use TF *and* Keras! A powerful combination, both easy-to-use and flexible/extensible.https://twitter.com/todddavid/status/971942826148802560 …
-
-
-
We've also got a lot of great new features for tf.keras in the pipeline. Very excited about it
Show this thread
End of conversation
New conversation -
-
-
Thank you! I need to shift from the low level discovery and calculus and look at it from this perspective. I've actually used them together but it hasn't been clear exactly why. This makes complete sense. I needed to learning from ground up. Now it's time to get productive!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Impressive work on Keras, Francois!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I would be dead without TF or Keras.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This should actually be clear to anyone who read Keras documentation.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
One other thing, I personally like about Keras, is the API consistency with other backend deep learning frameworks - TF, MxNet, Theano. To me, that's very powerful while experimenting
Thanks. 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.