Add Local Mode Batch Transform support. (#414)

Creating local batch transform jobs will perform all the steps locally. The serving container will be created the same way as it is for serving but then the data will be fed to it and written to the corresponding output location. Data splits and batch strategies are implemented as closely as possible to the behavior in SageMaker so that the end results are consistent.

Author: iquintero

CHANGELOG.rst

@@ -2,15 +2,18 @@
 CHANGELOG
 =========
 
-======
+1.11.3dev
+=========
+
+* feature: Local Mode: Add support for Batch Inference
+
 1.11.2
 ======
 
 * enhancement: Enable setting VPC config when creating/deploying models
 * enhancement: Local Mode: accept short lived credentials with a warning message
 * bug-fix: Local Mode: pass in job name as parameter for training environment variable
 
-=======
 1.11.1
 ======
 

README.rst

@@ -226,6 +226,27 @@ Here is an end-to-end example:
     predictor.delete_endpoint()
 
 
+If you don't want to deploy your model locally, you can also choose to perform a Local Batch Transform Job. This is
+useful if you want to test your container before creating a Sagemaker Batch Transform Job. Note that the performance
+will not match Batch Transform Jobs hosted on SageMaker but it is still a useful tool to ensure you have everything
+right or if you are not dealing with huge amounts of data.
+
+Here is an end-to-end example:
+
+.. code:: python
+
+    from sagemaker.mxnet import MXNet
+
+    mxnet_estimator = MXNet('train.py',
+                            train_instance_type='local',
+                            train_instance_count=1)
+
+    mxnet_estimator.fit('file:///tmp/my_training_data')
+    transformer = mxnet_estimator.transformer(1, 'local', assemble_with='Line', max_payload=1)
+    transformer.transform('s3://my/transform/data, content_type='text/csv', split_type='Line')
+    transformer.wait()
+
+
 For detailed examples of running Docker in local mode, see:
 
 - `TensorFlow local mode example notebook <https://github.com/awslabs/amazon-sagemaker-examples/blob/master/sagemaker-python-sdk/tensorflow_distributed_mnist/tensorflow_local_mode_mnist.ipynb>`__.

src/sagemaker/amazon/common.py

@@ -153,7 +153,7 @@ def write_spmatrix_to_sparse_tensor(file, array, labels=None):
 def read_records(file):
     """Eagerly read a collection of amazon Record protobuf objects from file."""
     records = []
-    for record_data in _read_recordio(file):
+    for record_data in read_recordio(file):
         record = Record()
         record.ParseFromString(record_data)
         records.append(record)
@@ -183,7 +183,7 @@ def _write_recordio(f, data):
     f.write(padding[pad])
 
 
-def _read_recordio(f):
+def read_recordio(f):
     while(True):
         try:
             read_kmagic, = struct.unpack('I', f.read(4))