Merge branch 'aws:master' into master

Author: cj-zhang

.pylintrc

@@ -310,7 +310,7 @@ ignore-mixin-members=yes
 # (useful for modules/projects where namespaces are manipulated during runtime
 # and thus existing member attributes cannot be deduced by static analysis. It
 # supports qualified module names, as well as Unix pattern matching.
-ignored-modules=distutils
+ignored-modules=
 
 # List of class names for which member attributes should not be checked (useful
 # for classes with dynamically set attributes). This supports the use of

CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## v2.232.2 (2024-10-03)
+
+### Bug Fixes and Other Changes
+
+ * Pass kwargs to HuggingFaceModel.deploy()
+ * improve logging and exception messages
+ * remove deprecated distutils
+ * update image_uri_configs  09-24-2024 07:18:00 PST
+
+## v2.232.1 (2024-09-19)
+
+### Bug Fixes and Other Changes
+
+ * update image_uri_configs  09-17-2024 07:17:54 PST
+ * support latest container version in image_uris and DJLModel for lmi c…
+
 ## v2.232.0 (2024-09-12)
 
 ### Features

VERSION

@@ -1 +1 @@
-2.232.1.dev0
+2.232.3.dev0

pyproject.toml

@@ -48,6 +48,7 @@ dependencies = [
   "PyYAML~=6.0",
   "requests",
   "sagemaker-core>=1.0.0,<2.0.0",
+  "sagemaker-mlflow",
   "schema",
   "smdebug_rulesconfig==1.0.1",
   "tblib>=1.7.0,<4",

src/sagemaker/estimator.py

@@ -107,6 +107,8 @@
 from sagemaker.workflow.parameters import ParameterString
 from sagemaker.workflow.pipeline_context import PipelineSession, runnable_by_pipeline
 
+from sagemaker.mlflow.forward_sagemaker_metrics import log_sagemaker_job_to_mlflow
+
 logger = logging.getLogger(__name__)
 
 
@@ -1366,8 +1368,14 @@ def fit(
         experiment_config = check_and_get_run_experiment_config(experiment_config)
         self.latest_training_job = _TrainingJob.start_new(self, inputs, experiment_config)
         self.jobs.append(self.latest_training_job)
+        forward_to_mlflow_tracking_server = False
+        if os.environ.get("MLFLOW_TRACKING_URI") and self.enable_network_isolation():
+            wait = True
+            forward_to_mlflow_tracking_server = True
         if wait:
             self.latest_training_job.wait(logs=logs)
+        if forward_to_mlflow_tracking_server:
+            log_sagemaker_job_to_mlflow(self.latest_training_job.name)
 
     def _compilation_job_name(self):
         """Placeholder docstring"""

src/sagemaker/huggingface/model.py

@@ -334,11 +334,7 @@ def deploy(
             container_startup_health_check_timeout=container_startup_health_check_timeout,
             inference_recommendation_id=inference_recommendation_id,
             explainer_config=explainer_config,
-            endpoint_logging=kwargs.get("endpoint_logging", False),
-            endpoint_type=kwargs.get("endpoint_type", None),
-            resources=kwargs.get("resources", None),
-            managed_instance_scaling=kwargs.get("managed_instance_scaling", None),
-            routing_config=kwargs.get("routing_config", None),
+            **kwargs,
         )
 
     def register(

src/sagemaker/image_uri_config/djl-lmi.json

@@ -28,6 +28,7 @@
                 "eu-west-3": "763104351884",
                 "eu-south-1": "692866216735",
                 "me-south-1": "217643126080",
+                "me-central-1": "914824155844",
                 "sa-east-1": "763104351884",
                 "us-east-1": "763104351884",
                 "us-east-2": "763104351884",
@@ -62,6 +63,7 @@
                 "eu-west-3": "763104351884",
                 "eu-south-1": "692866216735",
                 "me-south-1": "217643126080",
+                "me-central-1": "914824155844",
                 "sa-east-1": "763104351884",
                 "us-east-1": "763104351884",
                 "us-east-2": "763104351884",

src/sagemaker/image_uri_config/pytorch.json

@@ -1407,7 +1407,8 @@
             "2.0": "2.0.1",
             "2.1": "2.1.0",
             "2.2": "2.2.0",
-            "2.3": "2.3.0"
+            "2.3": "2.3.0",
+            "2.4": "2.4.0"
         },
         "versions": {
             "0.4.0": {
@@ -2429,6 +2430,47 @@
                     "us-west-2": "763104351884"
                 },
                 "repository": "pytorch-training"
+            },
+            "2.4.0": {
+                "py_versions": [
+                    "py311"
+                ],
+                "registries": {
+                    "af-south-1": "626614931356",
+                    "ap-east-1": "871362719292",
+                    "ap-northeast-1": "763104351884",
+                    "ap-northeast-2": "763104351884",
+                    "ap-northeast-3": "364406365360",
+                    "ap-south-1": "763104351884",
+                    "ap-south-2": "772153158452",
+                    "ap-southeast-1": "763104351884",
+                    "ap-southeast-2": "763104351884",
+                    "ap-southeast-3": "907027046896",
+                    "ap-southeast-4": "457447274322",
+                    "ca-central-1": "763104351884",
+                    "ca-west-1": "204538143572",
+                    "cn-north-1": "727897471807",
+                    "cn-northwest-1": "727897471807",
+                    "eu-central-1": "763104351884",
+                    "eu-central-2": "380420809688",
+                    "eu-north-1": "763104351884",
+                    "eu-south-1": "692866216735",
+                    "eu-south-2": "503227376785",
+                    "eu-west-1": "763104351884",
+                    "eu-west-2": "763104351884",
+                    "eu-west-3": "763104351884",
+                    "il-central-1": "780543022126",
+                    "me-central-1": "914824155844",
+                    "me-south-1": "217643126080",
+                    "sa-east-1": "763104351884",
+                    "us-east-1": "763104351884",
+                    "us-east-2": "763104351884",
+                    "us-gov-east-1": "446045086412",
+                    "us-gov-west-1": "442386744353",
+                    "us-west-1": "763104351884",
+                    "us-west-2": "763104351884"
+                },
+                "repository": "pytorch-training"
             }
         }
     }

src/sagemaker/jumpstart/cache.py

@@ -293,7 +293,8 @@ def _model_id_retrieval_function(
             raise KeyError(error_msg)
 
         error_msg = f"Unable to find model manifest for '{model_id}' with version '{version}'. "
-        error_msg += f"Visit {MODEL_ID_LIST_WEB_URL} for updated list of models. "
+        error_msg += "Specify a different model ID or try a different AWS Region. "
+        error_msg += f"For a list of available models, see {MODEL_ID_LIST_WEB_URL}. "
 
         other_model_id_version = None
         if model_type == JumpStartModelType.OPEN_WEIGHTS:

src/sagemaker/jumpstart/constants.py

@@ -222,8 +222,6 @@
 
 JUMPSTART_MODEL_HUB_NAME = "SageMakerPublicHub"
 
-JUMPSTART_MODEL_HUB_NAME = "SageMakerPublicHub"
-
 JUMPSTART_DEFAULT_MANIFEST_FILE_S3_KEY = "models_manifest.json"
 JUMPSTART_DEFAULT_PROPRIETARY_MANIFEST_KEY = "proprietary-sdk-manifest.json"
 

src/sagemaker/jumpstart/exceptions.py

@@ -32,8 +32,8 @@
 )
 
 INVALID_MODEL_ID_ERROR_MSG = (
-    "Invalid model ID: '{model_id}'. Please visit "
-    f"{MODEL_ID_LIST_WEB_URL} for a list of valid model IDs. "
+    "Invalid model ID: '{model_id}'. Specify a different model ID or try a different AWS Region. "
+    f"For a list of available models, see {MODEL_ID_LIST_WEB_URL}. "
     "The module `sagemaker.jumpstart.notebook_utils` contains utilities for "
     "fetching model IDs. We recommend upgrading to the latest version of sagemaker "
     "to get access to the most models."

src/sagemaker/jumpstart/model.py

@@ -447,6 +447,7 @@ def retrieve_example_payload(self) -> JumpStartSerializablePayload:
         return payloads.retrieve_example(
             model_id=self.model_id,
             model_version=self.model_version,
+            hub_arn=self.hub_arn,
             model_type=self.model_type,
             region=self.region,
             tolerate_deprecated_model=self.tolerate_deprecated_model,
@@ -1036,13 +1037,15 @@ def _get_deployment_configs(
                 image_uri=image_uri,
                 region=self.region,
                 model_version=self.model_version,
+                hub_arn=self.hub_arn,
             )
             deploy_kwargs = get_deploy_kwargs(
                 model_id=self.model_id,
                 instance_type=instance_type_to_use,
                 sagemaker_session=self.sagemaker_session,
                 region=self.region,
                 model_version=self.model_version,
+                hub_arn=self.hub_arn,
             )
 
             deployment_config_metadata = DeploymentConfigMetadata(

src/sagemaker/jumpstart/utils.py

@@ -50,7 +50,12 @@
 )
 from sagemaker.session import Session
 from sagemaker.config import load_sagemaker_config
-from sagemaker.utils import resolve_value_from_config, TagsDict, get_instance_rate_per_hour
+from sagemaker.utils import (
+    resolve_value_from_config,
+    TagsDict,
+    get_instance_rate_per_hour,
+    get_domain_for_region,
+)
 from sagemaker.workflow import is_pipeline_variable
 from sagemaker.user_agent import get_user_agent_extra_suffix
 
@@ -553,7 +558,7 @@ def get_eula_message(model_specs: JumpStartModelSpecs, region: str) -> str:
     return (
         f"Model '{model_specs.model_id}' requires accepting end-user license agreement (EULA). "
         f"See https://{get_jumpstart_content_bucket(region=region)}.s3.{region}."
-        f"amazonaws.com{'.cn' if region.startswith('cn-') else ''}"
+        f"{get_domain_for_region(region)}"
         f"/{model_specs.hosting_eula_key} for terms of use."
     )
 

src/sagemaker/local/image.py

@@ -30,7 +30,6 @@
 import tarfile
 import tempfile
 
-from distutils.spawn import find_executable
 from threading import Thread
 from typing import Dict, List
 from six.moves.urllib.parse import urlparse
@@ -170,7 +169,7 @@ def _get_compose_cmd_prefix():
             compose_cmd_prefix.extend(["docker", "compose"])
             return compose_cmd_prefix
 
-        if find_executable("docker-compose") is not None:
+        if shutil.which("docker-compose") is not None:
             logger.info("'Docker Compose' found using Docker Compose CLI.")
             compose_cmd_prefix.extend(["docker-compose"])
             return compose_cmd_prefix

src/sagemaker/local/utils.py

@@ -21,7 +21,6 @@
 import re
 import errno
 
-from distutils.dir_util import copy_tree
 from six.moves.urllib.parse import urlparse
 
 from sagemaker import s3
@@ -102,7 +101,7 @@ def move_to_destination(source, destination, job_name, sagemaker_session, prefix
 
 
 def recursive_copy(source, destination):
-    """A wrapper around distutils.dir_util.copy_tree.
+    """A wrapper around shutil.copy_tree.
 
     This won't throw any exception when the source directory does not exist.
 
@@ -111,7 +110,7 @@ def recursive_copy(source, destination):
         destination (str): destination path
     """
     if os.path.isdir(source):
-        copy_tree(source, destination)
+        shutil.copytree(source, destination, dirs_exist_ok=True)
 
 
 def kill_child_processes(pid):