Skip to content

Commit 8ec7f05

Browse files
laurenyulaurenyu
authored
doc: update documentation with v2.0.0.rc1 changes and bump version (#1686)
1 parent a0f1a78 commit 8ec7f05

File tree

3 files changed

+165
-12
lines changed

3 files changed

+165
-12
lines changed

‎CHANGELOG.md

Lines changed: 57 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,62 @@
11
# Changelog
22

3+
## v2.0.0.rc1 (2020-07-08)
4+
5+
### Breaking Changes
6+
7+
* Move StreamDeserializer to sagemaker.deserializers
8+
* Move StringDeserializer to sagemaker.deserializers
9+
* rename record_deserializer to RecordDeserializer
10+
* remove "train_" where redundant in parameter/variable names
11+
* Add BytesDeserializer
12+
* rename image to image_uri
13+
* rename image_name to image_uri
14+
* create new inference resources during model.deploy() and model.transformer()
15+
* rename session parameter to sagemaker_session in S3 utility classes
16+
* rename distributions to distribution in TF/MXNet estimators
17+
* deprecate update_endpoint arg in deploy()
18+
* create new inference resources during estimator.deploy() or estimator.transformer()
19+
* deprecate delete_endpoint() for estimators and HyperparameterTuner
20+
* refactor Predictor attribute endpoint to endpoint_name
21+
* make instance_type optional for Airflow model configs
22+
* refactor name of RealTimePredictor to Predictor
23+
* remove check for Python 2 string in sagemaker.predictor._is_sequence_like()
24+
* deprecate sagemaker.utils.to_str()
25+
* drop Python 2 support
26+
27+
### Features
28+
29+
* add BaseSerializer and BaseDeserializer
30+
* add Predictor.update_endpoint()
31+
32+
### Bug Fixes and Other Changes
33+
34+
* handle "train_*" renames in v2 migration tool
35+
* handle image_uri rename for Session methods in v2 migration tool
36+
* Update BytesDeserializer accept header
37+
* handle image_uri rename for estimators and models in v2 migration tool
38+
* handle image_uri rename in Airflow model config functions in v2 migration tool
39+
* update migration tool for S3 utility functions
40+
* set _current_job_name and base_tuning_job_name in HyperparameterTuner.attach()
41+
* infer base name from job name in estimator.attach()
42+
* ensure generated names are < 63 characters when deploying compiled models
43+
* add TF migration documentation to error message
44+
45+
### Documentation Changes
46+
47+
* update documentation with v2.0.0.rc1 changes
48+
* remove 'train_*' prefix from estimator parameters
49+
* update documentation for image_name/image --> image_uri
50+
51+
### Testing and Release Infrastructure
52+
53+
* refactor matching logic in v2 migration tool
54+
* add cli modifier for RealTimePredictor and derived classes
55+
* change coverage settings to reduce intermittent errors
56+
* clean up pickle.load logic in integ tests
57+
* use fixture for Python version in framework integ tests
58+
* remove assumption of Python 2 unit test runs
59+
360
## v1.68.0 (2020-07-07)
461

562
### Features

‎VERSION

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
2.0.0.rc0
1+
2.0.0.rc1

‎doc/v2.rst

Lines changed: 107 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -21,7 +21,7 @@ To install the latest release candidate:
2121

2222
.. code:: bash
2323
24-
pip install [email protected]:aws/[email protected].rc0
24+
pip install [email protected]:aws/[email protected].rc1
2525
2626
To install the latest version of v2:
2727

@@ -38,6 +38,38 @@ Changes
3838
This section is for major changes that may require updates to your SageMaker Python SDK code.
3939
You can also see what changes have been made in the `CHANGELOG <https://github.com/aws/sagemaker-python-sdk/blob/zwei/CHANGELOG.md>`_.
4040

41+
Deprecations
42+
============
43+
44+
Python 2 Support
45+
----------------
46+
47+
This library is no longer supported for Python 2.
48+
Please upgrade to Python 3 if you haven't already.
49+
50+
Deprecate Legacy TensorFlow
51+
---------------------------
52+
53+
TensorFlow versions 1.4-1.10 and some variations of versions 1.11-1.12
54+
(see `What Constitutes "Legacy TensorFlow Support" <frameworks/tensorflow/upgrade_from_legacy.html#what-constitutes-legacy-tensorflow-support>`_)
55+
are no longer natively supported by the SageMaker Python SDK.
56+
57+
To use those versions of TensorFlow, you must specify the Docker image URI explicitly,
58+
and configure settings via hyperparameters or environment variables rather than using SDK parameters.
59+
For more information, see `Upgrade from Legacy TensorFlow Support <frameworks/tensorflow/upgrade_from_legacy.html>`_.
60+
61+
``delete_endpoint()`` for Estimators and ``HyperparameterTuner``
62+
----------------------------------------------------------------
63+
64+
The ``delete_endpoint()`` method for estimators and ``HyperparameterTuner`` has been deprecated.
65+
Please use :func:`sagemaker.predictor.Predictor.delete_endpoint` instead.
66+
67+
``update_endpoint`` in ``deploy()``
68+
-----------------------------------
69+
70+
The ``update_endpoint`` argument in ``deploy()`` methods for estimators and models has been deprecated.
71+
Please use :func:`sagemaker.predictor.Predictor.update_endpoint` instead.
72+
4173
Require ``framework_version`` and ``py_version`` for Frameworks
4274
===============================================================
4375

@@ -69,19 +101,51 @@ For example:
69101
py_version="py3", # now required
70102
)
71103
72-
Deprecate Legacy TensorFlow
73-
===========================
104+
Parameter and Class Name Changes
105+
================================
74106

75-
TensorFlow versions 1.4-1.10 and some variations of versions 1.11-1.12
76-
(see `What Constitutes "Legacy TensorFlow Support" <frameworks/tensorflow/upgrade_from_legacy.html#what-constitutes-legacy-tensorflow-support>`_)
77-
are no longer natively supported by the SageMaker Python SDK.
107+
Estimators
108+
----------
78109

79-
To use those versions of TensorFlow, you must specify the Docker image URI explicitly,
80-
and configure settings via hyperparameters or environment variables rather than using SDK parameters.
81-
For more information, see `Upgrade from Legacy TensorFlow Support <frameworks/tensorflow/upgrade_from_legacy.html>`_.
110+
Renamed Estimator Parameters
111+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
112+
113+
The following estimator parameters have been renamed:
114+
115+
+------------------------------+------------------------+
116+
| v1.x | v2.0 and later |
117+
+==============================+========================+
118+
| ``train_instance_count`` | ``instance_count`` |
119+
+------------------------------+------------------------+
120+
| ``train_instance_type`` | ``instance_type`` |
121+
+------------------------------+------------------------+
122+
| ``train_max_run`` | ``max_run`` |
123+
+------------------------------+------------------------+
124+
| ``train_use_spot_instances`` | ``use_spot_instances`` |
125+
+------------------------------+------------------------+
126+
| ``train_max_run_wait`` | ``max_run_wait`` |
127+
+------------------------------+------------------------+
128+
| ``train_volume_size`` | ``volume_size`` |
129+
+------------------------------+------------------------+
130+
| ``train_volume_kms_key`` | ``volume_kms_key`` |
131+
+------------------------------+------------------------+
132+
133+
``distributions``
134+
~~~~~~~~~~~~~~~~~
135+
136+
For TensorFlow and MXNet estimators, ``distributions`` has been renamed to ``distribution``.
82137

83-
Parameter Changes for ``sagemaker.model.Model``
84-
===============================================
138+
Specify Custom Training Images
139+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
140+
141+
The ``image_name`` parameter has been renamed to ``image_uri`` for specifying a custom Docker image URI to use with training.
142+
143+
144+
Models
145+
------
146+
147+
``sagemaker.model.Model`` Parameter Order
148+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
85149

86150
The parameter order for :class:`sagemaker.model.Model` changed: instead of ``model_data`` being first, ``image_uri`` (formerly ``image``) is first.
87151
As a result, ``model_data`` has been made into an optional parameter.
@@ -96,6 +160,29 @@ If you are using the :class:`sagemaker.model.Model` class, your code should be c
96160
# v2.0 and later
97161
Model("my-image:latest", model_data="s3://bucket/path/model.tar.gz")
98162
163+
Specify Custom Serving Image
164+
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
165+
166+
The ``image`` parameter has been renamed to ``image_uri`` for specifying a custom Docker image URI to use with inference.
167+
168+
Predictors
169+
----------
170+
171+
``sagemaker.predictor.RealTimePredictor`` has been renamed to :class:`sagemaker.predictor.Predictor`.
172+
173+
In addition, for :class:`sagemaker.predictor.Predictor`, :class:`sagemaker.sparkml.model.SparkMLPredictor`,
174+
and predictors for Amazon algorithm (e.g. Factorization Machines, Linear Learner, etc.),
175+
the ``endpoint`` attribute has been renamed to ``endpoint_name``.
176+
177+
Airflow
178+
-------
179+
180+
For :func:`sagemaker.workflow.airflow.model_config` and :func:`sagemaker.workflow.airflow.model_config_from_estimator`,
181+
``instance_type`` is no longer the first positional argument and is now an optional keyword argument.
182+
183+
For :func:`sagemaker.workflow.airflow.model_config`, :func:`sagemaker.workflow.airflow.model_config_from_estimator`, and
184+
:func:`sagemaker.workflow.airflow.transform_config_from_estimator`, the ``image`` argument has been renamed to ``image_uri``.
185+
99186
Dependency Changes
100187
==================
101188

@@ -170,3 +257,12 @@ If you are using the ``sagemaker.tensorflow.serving.Model`` class, the tool does
170257
-------------------------
171258

172259
If you are using the :class:`sagemaker.model.Model` class, the tool does not take care of switching the order between ``model_data`` and ``image_uri`` (formerly ``image``).
260+
261+
``update_endpoint`` and ``delete_endpoint``
262+
-------------------------------------------
263+
264+
The tool does not take care of removing the ``update_endpoint`` argument from a ``deploy`` call.
265+
If you are using that argument, please modify your code to use :func:`sagemaker.predictor.Predictor.update_endpoint` instead.
266+
267+
The tool also does not handle ``delete_endpoint`` calls on estimators or ``HyperparameterTuner``.
268+
If you are using that method, please modify your code to use :func:`sagemaker.predictor.Predictor.delete_endpoint` instead.

0 commit comments

Comments
 (0)