Merge branch 'master' into job-step-subclass

Author: jerrypeng7773

CHANGELOG.md

@@ -1,5 +1,19 @@
 # Changelog
 
+## v2.87.0 (2022-04-20)
+
+### Features
+
+ * Add Jumpstart example notebooks
+ * add Tensorflow and Pytorch version for SM Training Compiler and expand to regular regions
+
+### Bug Fixes and Other Changes
+
+ * integs for training compiler in non-PDX regions
+ * TrainingStep cache misses due to timestamp based job name
+ * retry context delete
+ * Add more logging when unexpected number of artifacts found
+
 ## v2.86.2 (2022-04-14)
 
 ### Bug Fixes and Other Changes
@@ -165,7 +179,7 @@
 ### Features
 
  * override jumpstart content bucket
- * jumpstart model id suggestions
+ * jumpstart model ID suggestions
  * adding customer metadata support to registermodel step
 
 ### Bug Fixes and Other Changes

VERSION

@@ -1 +1 @@
-2.86.3.dev0
+2.87.1.dev0

doc/api/training/distributed.rst

@@ -10,7 +10,7 @@ The SageMaker Distributed Data Parallel Library
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. toctree::
-    :maxdepth: 3
+    :maxdepth: 2
 
     smd_data_parallel
     sdp_versions/latest
@@ -23,7 +23,7 @@ The SageMaker Distributed Model Parallel Library
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. toctree::
-   :maxdepth: 3
+   :maxdepth: 2
 
    smd_model_parallel
    smp_versions/latest

doc/api/training/smd_model_parallel_release_notes/smd_model_parallel_change_log.rst

@@ -5,9 +5,48 @@ Release Notes
 New features, bug fixes, and improvements are regularly made to the SageMaker
 distributed model parallel library.
 
-SageMaker Distributed Model Parallel 1.7.0 Release Notes
+SageMaker Distributed Model Parallel 1.8.0 Release Notes
 ========================================================
 
+*Date: March. 23. 2022*
+
+**New Features**
+
+* Added tensor parallelism support for the `GPT-J model
+  <https://huggingface.co/docs/transformers/model_doc/gptj>`_.
+  When using the GPT-J model of Hugging Face Transformers v4.17.0 with
+  tensor parallelism, the SageMaker model parallel library automatically
+  replaces the model with a tensor parallel distributed GPT-J model.
+  For more information, see `Support for Hugging Face Transformer Models
+  <https://docs.aws.amazon.com/sagemaker/latest/dg/model-parallel-extended-features-pytorch-hugging-face.html>`_
+  in the *Amazon SageMaker Model Parallel Training developer guide*.
+
+**Migration to AWS Deep Learning Containers**
+
+This version passed benchmark testing and is migrated to the following AWS Deep Learning Containers:
+
+* HuggingFace 4.17.0 DLC with PyTorch 1.10.2
+
+    .. code::
+
+      763104351884.dkr.ecr.us-west-2.amazonaws.com/huggingface-pytorch-training:1.10.2-transformers4.17.0-gpu-py38-cu113-ubuntu20.04
+
+
+    The binary file of this version of the library for custom container users:
+
+    .. code::
+
+      https://sagemaker-distributed-model-parallel.s3.us-west-2.amazonaws.com/pytorch-1.10.0/build-artifacts/2022-03-12-00-33/smdistributed_modelparallel-1.8.0-cp38-cp38-linux_x86_64.whl
+
+
+----
+
+Release History
+===============
+
+SageMaker Distributed Model Parallel 1.7.0 Release Notes
+--------------------------------------------------------
+
 *Date: March. 07. 2022*
 
 **Currency Updates**
@@ -49,11 +88,6 @@ This version passed benchmark testing and is migrated to the following AWS Deep
     763104351884.dkr.ecr.<region>.amazonaws.com/pytorch-training:1.10.2-gpu-py38-cu113-ubuntu20.04-sagemaker
 
 
-----
-
-Release History
-===============
-
 SageMaker Distributed Model Parallel 1.6.0 Release Notes
 --------------------------------------------------------
 

doc/api/training/smp_versions/latest.rst

@@ -10,8 +10,8 @@ depending on which version of the library you need to use.
 To use the library, reference the
 **Common API** documentation alongside the framework specific API documentation.
 
-Version 1.7.0 (Latest)
-======================
+Version 1.7.0, 1.8.0 (Latest)
+=============================
 
 To use the library, reference the Common API documentation alongside the framework specific API documentation.
 

doc/doc_utils/jumpstart_doc_utils.py

@@ -122,9 +122,9 @@ def create_jumpstart_model_table():
     file_content.append("==================================\n")
     file_content.append(
         """
-    JumpStart for the SageMaker Python SDK uses model ids and model versions to access the necessary
+    JumpStart for the SageMaker Python SDK uses model IDs and model versions to access the necessary
     utilities. This table serves to provide the core material plus some extra information that can be useful
-    in selecting the correct model id and corresponding parameters.\n
+    in selecting the correct model ID and corresponding parameters.\n
     """
     )
     file_content.append(

doc/overview.rst

@@ -670,7 +670,7 @@ the ``model_id`` and ``model_version`` needed to retrieve the URI.
       model. To use the latest version, enter ``"*"``. This is a
       required parameter.
 
-To retrieve a model, first select a ``model id`` and ``version`` from
+To retrieve a model, first select a ``model ID`` and ``version`` from
 the :doc:`available models <./doc_utils/jumpstart>`.
 
 .. code:: python

src/sagemaker/jumpstart/accessors.py

@@ -12,7 +12,7 @@
 # language governing permissions and limitations under the License.
 """This module contains accessors related to SageMaker JumpStart."""
 from __future__ import absolute_import
-from typing import Any, Dict, Optional
+from typing import Any, Dict, List, Optional
 from sagemaker.jumpstart.types import JumpStartModelHeader, JumpStartModelSpecs
 from sagemaker.jumpstart import cache
 from sagemaker.jumpstart.constants import JUMPSTART_DEFAULT_REGION_NAME
@@ -84,8 +84,8 @@ def get_model_header(region: str, model_id: str, version: str) -> JumpStartModel
 
         Args:
             region (str): region for which to retrieve header.
-            model_id (str): model id to retrieve.
-            version (str): semantic version to retrieve for the model id.
+            model_id (str): model ID to retrieve.
+            version (str): semantic version to retrieve for the model ID.
         """
         cache_kwargs = JumpStartModelsAccessor._validate_and_mutate_region_cache_kwargs(
             JumpStartModelsAccessor._cache_kwargs, region
@@ -101,8 +101,8 @@ def get_model_specs(region: str, model_id: str, version: str) -> JumpStartModelS
 
         Args:
             region (str): region for which to retrieve header.
-            model_id (str): model id to retrieve.
-            version (str): semantic version to retrieve for the model id.
+            model_id (str): model ID to retrieve.
+            version (str): semantic version to retrieve for the model ID.
         """
         cache_kwargs = JumpStartModelsAccessor._validate_and_mutate_region_cache_kwargs(
             JumpStartModelsAccessor._cache_kwargs, region
@@ -150,3 +150,22 @@ def reset_cache(cache_kwargs: Dict[str, Any] = None, region: Optional[str] = Non
         """
         cache_kwargs_dict = {} if cache_kwargs is None else cache_kwargs
         JumpStartModelsAccessor.set_cache_kwargs(cache_kwargs_dict, region)
+
+    @staticmethod
+    def get_manifest(
+        cache_kwargs: Optional[Dict[str, Any]] = None, region: Optional[str] = None
+    ) -> List[JumpStartModelHeader]:
+        """Return entire JumpStart models manifest.
+
+        Raises:
+            ValueError: If region in `cache_kwargs` is inconsistent with `region` argument.
+
+        Args:
+            cache_kwargs (Dict[str, Any]): Optional. Cache kwargs to use.
+                (Default: None).
+            region (str): Optional. The region to use for the cache.
+                (Default: None).
+        """
+        cache_kwargs_dict: Dict[str, Any] = {} if cache_kwargs is None else cache_kwargs
+        JumpStartModelsAccessor.set_cache_kwargs(cache_kwargs_dict, region)
+        return JumpStartModelsAccessor._cache.get_manifest()  # type: ignore

src/sagemaker/jumpstart/cache.py

@@ -146,16 +146,16 @@ def _get_manifest_key_from_model_id_semantic_version(
         key: JumpStartVersionedModelId,
         value: Optional[JumpStartVersionedModelId],  # pylint: disable=W0613
     ) -> JumpStartVersionedModelId:
-        """Return model id and version in manifest that matches semantic version/id.
+        """Return model ID and version in manifest that matches semantic version/id.
 
         Uses ``packaging.version`` to perform version comparison. The highest model version
         matching the semantic version is used, which is compatible with the SageMaker
         version.
 
         Args:
-            key (JumpStartVersionedModelId): Key for which to fetch versioned model id.
+            key (JumpStartVersionedModelId): Key for which to fetch versioned model ID.
             value (Optional[JumpStartVersionedModelId]): Unused variable for current value of
-                old cached model id/version.
+                old cached model ID/version.
 
         Raises:
             KeyError: If the semantic version is not found in the manifest, or is found but
@@ -287,10 +287,10 @@ def get_manifest(self) -> List[JumpStartModelHeader]:
         return manifest
 
     def get_header(self, model_id: str, semantic_version_str: str) -> JumpStartModelHeader:
-        """Return header for a given JumpStart model id and semantic version.
+        """Return header for a given JumpStart model ID and semantic version.
 
         Args:
-            model_id (str): model id for which to get a header.
+            model_id (str): model ID for which to get a header.
             semantic_version_str (str): The semantic version for which to get a
                 header.
         """
@@ -331,7 +331,7 @@ def _get_header_impl(
         Allows a single retry if the cache is old.
 
         Args:
-            model_id (str): model id for which to get a header.
+            model_id (str): model ID for which to get a header.
             semantic_version_str (str): The semantic version for which to get a
                 header.
             attempt (int): attempt number at retrieving a header.
@@ -353,10 +353,10 @@ def _get_header_impl(
             return self._get_header_impl(model_id, semantic_version_str, attempt + 1)
 
     def get_specs(self, model_id: str, semantic_version_str: str) -> JumpStartModelSpecs:
-        """Return specs for a given JumpStart model id and semantic version.
+        """Return specs for a given JumpStart model ID and semantic version.
 
         Args:
-            model_id (str): model id for which to get specs.
+            model_id (str): model ID for which to get specs.
             semantic_version_str (str): The semantic version for which to get
                 specs.
         """
@@ -369,6 +369,6 @@ def get_specs(self, model_id: str, semantic_version_str: str) -> JumpStartModelS
         return specs  # type: ignore
 
     def clear(self) -> None:
-        """Clears the model id/version and s3 cache."""
+        """Clears the model ID/version and s3 cache."""
         self._s3_cache.clear()
         self._model_id_semantic_version_manifest_key_cache.clear()

src/sagemaker/jumpstart/exceptions.py

@@ -49,7 +49,7 @@ def __init__(
         """Instantiates VulnerableJumpStartModelError exception.
 
         Args:
-            model_id (Optional[str]): model id of vulnerable JumpStart model.
+            model_id (Optional[str]): model ID of vulnerable JumpStart model.
                 (Default: None).
             version (Optional[str]): version of vulnerable JumpStart model.
                 (Default: None).