Make tf.keras.Sequential (properly) Checkpointable
authorAllen Lavoie <allenl@google.com>
Wed, 28 Mar 2018 00:14:50 +0000 (17:14 -0700)
committerTensorFlower Gardener <gardener@tensorflow.org>
Wed, 28 Mar 2018 00:18:12 +0000 (17:18 -0700)
commit52271c7c2ae3905bb557ebf355a4eca44a930c1f
tree67093ca7a0a9f03324d343f385b9d38a2f065e7b
parent3dc861e754ffb86286038ef9c78327f59384eaad
Make tf.keras.Sequential (properly) Checkpointable

Just numbers Layers like "layer-N". It may also make sense to track them by
"ClassName-M", but that's a backwards-compatible change.

Special-cases all of the dependency collection, since Layers can be added and
removed from Sequential.

PiperOrigin-RevId: 190699818
tensorflow/contrib/eager/python/checkpointable_utils_test.py
tensorflow/python/keras/_impl/keras/engine/sequential.py
tensorflow/python/training/checkpointable.py