Revert "TF 2.x: Support for keras to estimator (aws#268)" (aws#273)

This reverts commit 749bded.

Author: NihalHarish

docs/sagemaker.md

@@ -27,7 +27,7 @@ Here's a list of frameworks and versions which support this experience.
 
 | Framework | Version |
 | --- | --- |
-| [TensorFlow](tensorflow.md) | 1.15, 2.1, 2.2 |
+| [TensorFlow](tensorflow.md) | 1.15, 2.1 |
 | [MXNet](mxnet.md) | 1.6 |
 | [PyTorch](pytorch.md) | 1.4, 1.5 |
 | [XGBoost](xgboost.md) | >=0.90-2 [As Built-in algorithm](xgboost.md#use-xgboost-as-a-built-in-algorithm)|

docs/tensorflow.md

@@ -15,22 +15,17 @@
 ### Versions
 - Zero Script Change experience where you need no modifications to your training script is supported in the official [SageMaker Framework Container for TensorFlow 1.15](https://docs.aws.amazon.com/sagemaker/latest/dg/pre-built-containers-frameworks-deep-learning.html), or the [AWS Deep Learning Container for TensorFlow 1.15](https://aws.amazon.com/machine-learning/containers/).
 
-- This library itself supports the following versions when you use our API which requires a few minimal changes to your training script: TensorFlow 1.14, 1.15, 2.0+. Keras 2.3.
+- This library itself supports the following versions when you use our API which requires a few minimal changes to your training script: TensorFlow 1.14, 1.15, 2.0.1, 2.1.0. Keras 2.3.
 
 ### Interfaces
-- TF 1.x:
-    - [Estimator](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/estimator)
-    - [tf.keras](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/keras)
-    - [MonitoredSession](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/train/MonitoredSession?hl=en)
-- TF 2.x:
-    - [Estimator](https://www.tensorflow.org/versions/r2.1/api_docs/python/tf/estimator)
-    - [tf.keras](https://www.tensorflow.org/versions/r2.1/api_docs/python/tf/keras)
-
+- [Estimator](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/estimator)
+- [tf.keras](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/keras)
+- [MonitoredSession](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/train/MonitoredSession?hl=en)
 
 ### Distributed training
 - [MirroredStrategy](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/distribute/MirroredStrategy) or [Contrib MirroredStrategy](https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/distribute/MirroredStrategy)
 
-We will very quickly follow up with support for Parameter Server based training.
+We will very quickly follow up with support for Horovod and Parameter Server based training.
 
 ---
 

smdebug/tensorflow/collection.py

@@ -148,6 +148,9 @@ def __init__(self, collections=None, create_default=True):
                 self.create_collection(n)
             if is_tf_version_2x() and tf.executing_eagerly():
                 self.get(CollectionKeys.BIASES).include("^(?!gradient).*bias")
+                self.get(CollectionKeys.WEIGHTS).include("^weights/.*/((?!bias).)*$")
+                self.get(CollectionKeys.LOSSES).include(".*loss.*")
+                self.get(CollectionKeys.GRADIENTS).include("^gradient")
             else:
                 self.get(CollectionKeys.BIASES).include("bias")
 

smdebug/tensorflow/keras.py

@@ -716,11 +716,6 @@ def run(*args, **kwargs):
                 # at this point we need all collections to be ready
                 # this may not be the case at creation of hook
                 # as user's code after hook might add collections
-                self.collection_manager.get(CollectionKeys.WEIGHTS).include(
-                    "^weights/.*/((?!bias).)*$"
-                )
-                self.collection_manager.get(CollectionKeys.LOSSES).include(".*loss.*")
-                self.collection_manager.get(CollectionKeys.GRADIENTS).include("^gradient")
                 self._prepare_collections()
                 self.prepared_collections = True
 

tests/tensorflow2/test_estimator.py

@@ -12,7 +12,6 @@
 # Third Party
 import pytest
 import tensorflow.compat.v2 as tf
-import tensorflow_datasets as tfds
 from tests.tensorflow2.utils import is_tf_2_2
 from tests.tensorflow.utils import create_trial_fast_refresh
 
@@ -650,47 +649,3 @@ def test_keras_fit_pure_eager(out_dir, tf_eager_mode):
     assert len(trial.tensor_names(collection=CollectionKeys.BIASES)) == 2
     assert len(trial.tensor_names(collection=CollectionKeys.WEIGHTS)) == 2
     assert len(trial.tensor_names(collection=CollectionKeys.OPTIMIZER_VARIABLES)) == 5
-
-
-def test_keras_to_estimator(out_dir, tf_eager_mode):
-    if not tf_eager_mode:
-        tf.compat.v1.disable_eager_execution()
-        tf.compat.v1.reset_default_graph()
-
-    tf.keras.backend.clear_session()
-
-    model = tf.keras.models.Sequential(
-        [
-            tf.keras.layers.Dense(16, activation="relu", input_shape=(4,)),
-            tf.keras.layers.Dropout(0.2),
-            tf.keras.layers.Dense(1, activation="sigmoid"),
-        ]
-    )
-
-    def input_fn():
-        split = tfds.Split.TRAIN
-        dataset = tfds.load("iris", split=split, as_supervised=True)
-        dataset = dataset.map(lambda features, labels: ({"dense_input": features}, labels))
-        dataset = dataset.batch(32).repeat()
-        return dataset
-
-    model.compile(loss="categorical_crossentropy", optimizer="adam")
-    model.summary()
-
-    keras_estimator = tf.keras.estimator.model_to_estimator(keras_model=model, model_dir=out_dir)
-
-    hook = smd.EstimatorHook(out_dir)
-
-    hook.set_mode(smd.modes.TRAIN)
-    keras_estimator.train(input_fn=input_fn, steps=25, hooks=[hook])
-
-    hook.set_mode(smd.modes.EVAL)
-    eval_result = keras_estimator.evaluate(input_fn=input_fn, steps=10, hooks=[hook])
-
-    from smdebug.trials import create_trial
-
-    tr = create_trial(out_dir)
-    assert len(tr.tensor_names()) == 1
-    assert len(tr.steps()) == 2
-    assert len(tr.steps(smd.modes.TRAIN)) == 1
-    assert len(tr.steps(smd.modes.EVAL)) == 1

tests/tensorflow2/test_keras.py

@@ -5,6 +5,7 @@
 import numpy as np
 import tensorflow.compat.v1 as tf
 import tensorflow_datasets as tfds
+from tensorflow.examples.tutorials.mnist import input_data
 
 tfds.disable_progress_bar()
 
@@ -231,8 +232,6 @@ def neural_net(x):
 
 
 def get_data() -> "tf.contrib.learn.python.learn.datasets.base.Datasets":
-    from tensorflow.examples.tutorials.mnist import input_data
-
     mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
     return mnist