TensorBoard tutorial with tf.keras

There are currently end-to-end tutorials using tf.keras and TensorBoard - at least, none that I could see in the TensorFlow repo.☺

It would be a nice enhancement to expand the documentation - even just to include the flower transfer learning example, or the MNIST example shown at 2017 TensorFlow Dev Summit.

[dynamicwebpaige]

Excellent! Have been working on a quickstart today for the 1D CNN with imdb example in Keras’ docs.

It would be interesting to explicitly show how TensorBoard visualizations change as you:

• add layers (same activation function)
• add layers (different activation function)
• increase number of epochs, histograms
• include write_images, write_grad, and write_graph

Any other ideas / wish list items? ☺

Also: right now, the only way I can get support for tf.keras is by importing as tensorflow.python.keras. ref

Thanks! .pb

[jart]

I agree and we’d welcome contributions doing just that.