Merge pull request aws#214 from winstonaws/fix_tf_cifar_predict

Fix predictions for tensorflow cifar example

Author: djarpin

sagemaker-python-sdk/tensorflow_resnet_cifar10_with_tensorboard/source_dir/resnet_cifar_10.py

@@ -106,39 +106,39 @@ def model_fn(features, labels, mode, params):
 
 
 def serving_input_fn(params):
-    feature_spec = {INPUT_TENSOR_NAME: tf.FixedLenFeature(dtype=tf.float32, shape=(32, 32, 3))}
-    return tf.estimator.export.build_parsing_serving_input_receiver_fn(feature_spec)()
+    inputs = {INPUT_TENSOR_NAME: tf.placeholder(tf.float32, [None, 32, 32, 3])}
+    return tf.estimator.export.ServingInputReceiver(inputs, inputs)
 
 
 def train_input_fn(training_dir, params):
-    return input_fn(tf.estimator.ModeKeys.TRAIN,
+    return _input_from_files(tf.estimator.ModeKeys.TRAIN,
                     batch_size=BATCH_SIZE, data_dir=training_dir)
 
 
 def eval_input_fn(training_dir, params):
-    return input_fn(tf.estimator.ModeKeys.EVAL,
+    return _input_from_files(tf.estimator.ModeKeys.EVAL,
                     batch_size=BATCH_SIZE, data_dir=training_dir)
 
 
-def input_fn(mode, batch_size, data_dir):
-    """Input_fn using the contrib.data input pipeline for CIFAR-10 dataset.
+def _input_from_files(mode, batch_size, data_dir):
+    """Uses the contrib.data input pipeline for CIFAR-10 dataset.
 
   Args:
     mode: Standard names for model modes (tf.estimators.ModeKeys).
     batch_size: The number of samples per batch of input requested.
   """
-    dataset = record_dataset(filenames(mode, data_dir))
+    dataset = _record_dataset(_filenames(mode, data_dir))
 
     # For training repeat forever.
     if mode == tf.estimator.ModeKeys.TRAIN:
         dataset = dataset.repeat()
 
-    dataset = dataset.map(dataset_parser, num_threads=1,
+    dataset = dataset.map(_dataset_parser, num_threads=1,
                           output_buffer_size=2 * batch_size)
 
     # For training, preprocess the image and shuffle.
     if mode == tf.estimator.ModeKeys.TRAIN:
-        dataset = dataset.map(train_preprocess_fn, num_threads=1,
+        dataset = dataset.map(_train_preprocess_fn, num_threads=1,
                               output_buffer_size=2 * batch_size)
 
         # Ensure that the capacity is sufficiently large to provide good random
@@ -160,7 +160,7 @@ def input_fn(mode, batch_size, data_dir):
     return {INPUT_TENSOR_NAME: images}, labels
 
 
-def train_preprocess_fn(image, label):
+def _train_preprocess_fn(image, label):
     """Preprocess a single training image of layout [height, width, depth]."""
     # Resize the image to add four extra pixels on each side.
     image = tf.image.resize_image_with_crop_or_pad(image, HEIGHT + 8, WIDTH + 8)
@@ -174,7 +174,7 @@ def train_preprocess_fn(image, label):
     return image, label
 
 
-def dataset_parser(value):
+def _dataset_parser(value):
     """Parse a CIFAR-10 record from value."""
     # Every record consists of a label followed by the image, with a fixed number
     # of bytes for each.
@@ -200,13 +200,13 @@ def dataset_parser(value):
     return image, tf.one_hot(label, NUM_CLASSES)
 
 
-def record_dataset(filenames):
+def _record_dataset(filenames):
     """Returns an input pipeline Dataset from `filenames`."""
     record_bytes = HEIGHT * WIDTH * DEPTH + 1
     return tf.contrib.data.FixedLengthRecordDataset(filenames, record_bytes)
 
 
-def filenames(mode, data_dir):
+def _filenames(mode, data_dir):
     """Returns a list of filenames based on 'mode'."""
     data_dir = os.path.join(data_dir, 'cifar-10-batches-bin')
 

sagemaker-python-sdk/tensorflow_resnet_cifar10_with_tensorboard/tensorflow_resnet_cifar10_with_tensorboard.ipynb

@@ -138,8 +138,7 @@
     "\n",
     "It takes a few minutes to provision containers and start the training job.**TensorBoard** will start to display metrics shortly after that.\n",
     "\n",
-    "You can access **TensorBoard** locally at [http://localhost:6006](http://localhost:6006) or using your SageMaker notebook instance [proxy/6006/](/proxy/6006/)(TensorBoard will not work if forget to put the slash, '/', in end of the url). If TensorBoard started on a different port, adjust these URLs to match.",
-    "This example uses the optional hyperparameter **```min_eval_frequency```** to generate training evaluations more often, allowing to visualize **TensorBoard** scalar data faster. You can find the available optional hyperparameters [here](https://github.com/aws/sagemaker-python-sdk#optional-hyperparameters)**."
+    "You can access **TensorBoard** locally at [http://localhost:6006](http://localhost:6006) or using your SageMaker notebook instance [proxy/6006/](/proxy/6006/)(TensorBoard will not work if forget to put the slash, '/', in end of the url). If TensorBoard started on a different port, adjust these URLs to match.This example uses the optional hyperparameter **```min_eval_frequency```** to generate training evaluations more often, allowing to visualize **TensorBoard** scalar data faster. You can find the available optional hyperparameters [here](https://github.com/aws/sagemaker-python-sdk#optional-hyperparameters)**."
    ]
   },
   {
@@ -156,14 +155,33 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {
-    "collapsed": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "predictor = estimator.deploy(initial_instance_count=1, instance_type='ml.m4.xlarge')"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Make a prediction with fake data to verify the endpoint is up\n",
+    "\n",
+    "Prediction is not the focus of this notebook, so to verify the endpoint's functionality, we'll simply generate random data in the correct shape and make a prediction."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np\n",
+    "\n",
+    "random_image_data = np.random.rand(32, 32, 3)\n",
+    "predictor.predict(random_image_data)"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -185,24 +203,24 @@
   }
  ],
  "metadata": {
-  "notice": "Copyright 2017 Amazon.com, Inc. or its affiliates. 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. A copy of the License is located at http://aws.amazon.com/apache2.0/ or in the \"license\" file accompanying this file. This file 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.",
   "kernelspec": {
-   "display_name": "Environment (conda_tensorflow_p27)",
+   "display_name": "conda_tensorflow_p27",
    "language": "python",
    "name": "conda_tensorflow_p27"
   },
   "language_info": {
    "codemirror_mode": {
     "name": "ipython",
-    "version": 3
+    "version": 2
    },
    "file_extension": ".py",
    "mimetype": "text/x-python",
    "name": "python",
    "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "2.7.13"
-  }
+   "pygments_lexer": "ipython2",
+   "version": "2.7.11"
+  },
+  "notice": "Copyright 2017 Amazon.com, Inc. or its affiliates. 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. A copy of the License is located at http://aws.amazon.com/apache2.0/ or in the \"license\" file accompanying this file. This file 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."
  },
  "nbformat": 4,
  "nbformat_minor": 2