Merge remote-tracking branch 'tensorflow/master'

Author: Alexander Gorban

CODEOWNERS

@@ -1,40 +1,43 @@
-official/* @nealwu @k-w-w
-research/adversarial_crypto/* @dave-andersen
-research/adversarial_text/* @rsepassi
-research/adv_imagenet_models/* @AlexeyKurakin
-research/attention_ocr/* @alexgorban
-research/audioset/* @plakal @dpwe
-research/autoencoders/* @snurkabill
-research/cognitive_mapping_and_planning/* @s-gupta
-research/compression/* @nmjohn
-research/delf/* @andrefaraujo
-research/differential_privacy/* @panyx0718
-research/domain_adaptation/* @bousmalis @ddohan
-research/im2txt/* @cshallue
-research/inception/* @shlens @vincentvanhoucke
-research/learned_optimizer/* @olganw @nirum
-research/learning_to_remember_rare_events/* @lukaszkaiser @ofirnachum
-research/lfads/* @jazcollins @susillo
-research/lm_1b/* @oriolvinyals @panyx0718
-research/namignizer/* @knathanieltucker
-research/neural_gpu/* @lukaszkaiser
-research/neural_programmer/* @arvind2505
-research/next_frame_prediction/* @panyx0718
-research/object_detection/* @jch1 @tombstone @derekjchow @jesu9 @dreamdragon
-research/pcl_rl/* @ofirnachum
-research/ptn/* @xcyan @arkanath @hellojas @honglaklee
-research/real_nvp/* @laurent-dinh
-research/rebar/* @gjtucker
-research/resnet/* @panyx0718
-research/skip_thoughts/* @cshallue
-research/slim/* @sguada @nathansilberman
-research/street/* @theraysmith
-research/swivel/* @waterson
-research/syntaxnet/* @calberti @andorardo @bogatyy @markomernick
-research/textsum/* @panyx0718 @peterjliu
-research/transformer/* @daviddao
-research/video_prediction/* @cbfinn
-samples/* @MarkDaoust
-tutorials/embedding/* @zffchen78 @a-dai
-tutorials/image/* @sherrym @shlens
-tutorials/rnn/* @lukaszkaiser @ebrevdo
+/official/ @nealwu @k-w-w
+/research/adversarial_crypto/ @dave-andersen
+/research/adversarial_text/ @rsepassi
+/research/adv_imagenet_models/ @AlexeyKurakin
+/research/attention_ocr/ @alexgorban
+/research/audioset/ @plakal @dpwe
+/research/autoencoders/ @snurkabill
+/research/cognitive_mapping_and_planning/ @s-gupta
+/research/compression/ @nmjohn
+/research/delf/ @andrefaraujo
+/research/differential_privacy/ @panyx0718
+/research/domain_adaptation/ @bousmalis @dmrd
+/research/gan/ @joel-shor
+/research/im2txt/ @cshallue
+/research/inception/ @shlens @vincentvanhoucke
+/research/learned_optimizer/ @olganw @nirum
+/research/learning_to_remember_rare_events/ @lukaszkaiser @ofirnachum
+/research/lfads/ @jazcollins @susillo
+/research/lm_1b/ @oriolvinyals @panyx0718
+/research/namignizer/ @knathanieltucker
+/research/neural_gpu/ @lukaszkaiser
+/research/neural_programmer/ @arvind2505
+/research/next_frame_prediction/ @panyx0718
+/research/object_detection/ @jch1 @tombstone @derekjchow @jesu9 @dreamdragon
+/research/pcl_rl/ @ofirnachum
+/research/ptn/ @xcyan @arkanath @hellojas @honglaklee
+/research/real_nvp/ @laurent-dinh
+/research/rebar/ @gjtucker
+/research/resnet/ @panyx0718
+/research/skip_thoughts/ @cshallue
+/research/slim/ @sguada @nathansilberman
+/research/street/ @theraysmith
+/research/swivel/ @waterson
+/research/syntaxnet/ @calberti @andorardo @bogatyy @markomernick
+/research/tcn/ @coreylynch @sermanet
+/research/textsum/ @panyx0718 @peterjliu
+/research/transformer/ @daviddao
+/research/video_prediction/ @cbfinn
+/research/fivo/ @dieterichlawson
+/samples/ @MarkDaoust
+/tutorials/embedding/ @zffchen78 @a-dai
+/tutorials/image/ @sherrym @shlens
+/tutorials/rnn/ @lukaszkaiser @ebrevdo

official/mnist/README.md

@@ -2,7 +2,7 @@
 
 This directory builds a convolutional neural net to classify the [MNIST
 dataset](http://yann.lecun.com/exdb/mnist/) using the
-[tf.contrib.data](https://www.tensorflow.org/api_docs/python/tf/contrib/data),
+[tf.data](https://www.tensorflow.org/api_docs/python/tf/data),
 [tf.estimator.Estimator](https://www.tensorflow.org/api_docs/python/tf/estimator/Estimator),
 and
 [tf.layers](https://www.tensorflow.org/api_docs/python/tf/layers)
@@ -12,18 +12,51 @@ APIs.
 ## Setup
 
 To begin, you'll simply need the latest version of TensorFlow installed.
+Then to train the model, run the following:
 
-First convert the MNIST data to TFRecord file format by running the following:
+```
+python mnist.py
+```
+
+The model will begin training and will automatically evaluate itself on the
+validation data.
+
+Illustrative unit tests and benchmarks can be run with:
 
 ```
-python convert_to_records.py
+python mnist_test.py
+python mnist_test.py --benchmarks=.
 ```
 
-Then to train the model, run the following:
+## Exporting the model
+
+You can export the model into Tensorflow [SavedModel](https://www.tensorflow.org/programmers_guide/saved_model) format by using the argument `--export_dir`:
 
 ```
-python mnist.py
+python mnist.py --export_dir /tmp/mnist_saved_model
+```
+
+The SavedModel will be saved in a timestamped directory under `/tmp/mnist_saved_model/` (e.g. `/tmp/mnist_saved_model/1513630966/`).
+
+**Getting predictions with SavedModel**
+Use [`saved_model_cli`](https://www.tensorflow.org/programmers_guide/saved_model#cli_to_inspect_and_execute_savedmodel) to inspect and execute the SavedModel.
+
+```
+saved_model_cli run --dir /tmp/mnist_saved_model/TIMESTAMP --tag_set serve --signature_def classify --inputs image=examples.npy
+```
+
+`examples.npy` contains the data from `example5.png` and `example3.png` in a numpy array, in that order. The array values are normalized to values between 0 and 1.
+
+The output should look similar to below:
+```
+Result for output key classes:
+[5 3]
+Result for output key probabilities:
+[[  1.53558474e-07   1.95694142e-13   1.31193523e-09   5.47467265e-03
+    5.85711526e-22   9.94520664e-01   3.48423509e-06   2.65365645e-17
+    9.78631419e-07   3.15522470e-08]
+ [  1.22413359e-04   5.87615965e-08   1.72251271e-06   9.39960718e-01
+    3.30306928e-11   2.87386645e-02   2.82353517e-02   8.21146413e-18
+    2.52568233e-03   4.15460236e-04]]
 ```
 
-The model will begin training and will automatically evaluate itself on the
-validation data.

official/mnist/convert_to_records.py

@@ -0,0 +1,112 @@
+#  Copyright 2018 The TensorFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+"""tf.data.Dataset interface to the MNIST dataset."""
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import os
+import shutil
+import gzip
+import numpy as np
+from six.moves import urllib
+import tensorflow as tf
+
+
+def read32(bytestream):
+  """Read 4 bytes from bytestream as an unsigned 32-bit integer."""
+  dt = np.dtype(np.uint32).newbyteorder('>')
+  return np.frombuffer(bytestream.read(4), dtype=dt)[0]
+
+
+def check_image_file_header(filename):
+  """Validate that filename corresponds to images for the MNIST dataset."""
+  with open(filename) as f:
+    magic = read32(f)
+    num_images = read32(f)
+    rows = read32(f)
+    cols = read32(f)
+    if magic != 2051:
+      raise ValueError('Invalid magic number %d in MNIST file %s' % (magic,
+                                                                     f.name))
+    if rows != 28 or cols != 28:
+      raise ValueError(
+          'Invalid MNIST file %s: Expected 28x28 images, found %dx%d' %
+          (f.name, rows, cols))
+
+
+def check_labels_file_header(filename):
+  """Validate that filename corresponds to labels for the MNIST dataset."""
+  with open(filename) as f:
+    magic = read32(f)
+    num_items = read32(f)
+    if magic != 2049:
+      raise ValueError('Invalid magic number %d in MNIST file %s' % (magic,
+                                                                     f.name))
+
+
+def download(directory, filename):
+  """Download (and unzip) a file from the MNIST dataset, if it doesn't already exist."""
+  if not tf.gfile.Exists(directory):
+    tf.gfile.MakeDirs(directory)
+  filepath = os.path.join(directory, filename)
+  if tf.gfile.Exists(filepath):
+    return filepath
+  # CVDF mirror of http://yann.lecun.com/exdb/mnist/
+  url = 'https://storage.googleapis.com/cvdf-datasets/mnist/' + filename + '.gz'
+  zipped_filepath = filepath + '.gz'
+  print('Downloading %s to %s' % (url, zipped_filepath))
+  urllib.request.urlretrieve(url, zipped_filepath)
+  with gzip.open(zipped_filepath, 'rb') as f_in, open(filepath, 'wb') as f_out:
+    shutil.copyfileobj(f_in, f_out)
+  os.remove(zipped_filepath)
+  return filepath
+
+
+def dataset(directory, images_file, labels_file):
+  images_file = download(directory, images_file)
+  labels_file = download(directory, labels_file)
+
+  check_image_file_header(images_file)
+  check_labels_file_header(labels_file)
+
+  def decode_image(image):
+    # Normalize from [0, 255] to [0.0, 1.0]
+    image = tf.decode_raw(image, tf.uint8)
+    image = tf.cast(image, tf.float32)
+    image = tf.reshape(image, [784])
+    return image / 255.0
+
+  def one_hot_label(label):
+    label = tf.decode_raw(label, tf.uint8)  # tf.string -> tf.uint8
+    label = tf.reshape(label, [])  # label is a scalar
+    return tf.one_hot(label, 10)
+
+  images = tf.data.FixedLengthRecordDataset(
+      images_file, 28 * 28, header_bytes=16).map(decode_image)
+  labels = tf.data.FixedLengthRecordDataset(
+      labels_file, 1, header_bytes=8).map(one_hot_label)
+  return tf.data.Dataset.zip((images, labels))
+
+
+def train(directory):
+  """tf.data.Dataset object for MNIST training data."""
+  return dataset(directory, 'train-images-idx3-ubyte',
+                 'train-labels-idx1-ubyte')
+
+
+def test(directory):
+  """tf.data.Dataset object for MNIST test data."""
+  return dataset(directory, 't10k-images-idx3-ubyte', 't10k-labels-idx1-ubyte')