@TensorFlow @rajatmonga tf.layers, tf.contrib.layers, tf.contrib.slim, tf.contrib.keras.layers ... When will this end? Which should I use?
-
tf.layers is core. Everything in contrib could go away. tf.keras and tf slim will probably merge with tf.layers eventually.
tf.layers is a subset of tf.keras.layers. They're interchangeable. Contrib will be split out of TF in the future. I suggest using tf.keras.
-
-
tf.keras is part of the core TF API; its role is to make TF more approachable and make power users more productive. While fully blending in
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
why not merge tf.layers and tf.keras? i've never understood why TF's API is so manic and has ~4-5 ways to do common things
-
e.g. batch norm is available in tf.nn, tf.layers, tf.contrib.layers, tf.contrib.keras.layers, tf.contrib.slim
- Show replies
New conversation -
-
-
tf.layers is core as
@jimmfleming said and is the recommended way to go! tf.keras is supported, and useful when you come from Keras. -
Thanks all for the info! I was unaware that tf.keras existed. It's absent from the API docs @ https://www.tensorflow.org/api_docs/python/ … ?
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.