Merge branch 'master' into amtviz

Author: uyoldas

CHANGELOG.md

@@ -1,5 +1,16 @@
 # Changelog
 
+## v2.246.0 (2025-06-04)
+
+### Features
+
+ * Triton v25.04 DLC
+
+### Bug Fixes and Other Changes
+
+ * Update Attrs version to widen support
+ * update estimator documentation regarding hyperparameters for source_dir
+
 ## v2.245.0 (2025-05-28)
 
 ### Features

VERSION

@@ -1 +1 @@
-2.245.1.dev0
+2.246.1.dev0

pyproject.toml

@@ -31,7 +31,7 @@ classifiers = [
   "Programming Language :: Python :: 3.12",
 ]
 dependencies = [
-  "attrs>=23.1.0,<24",
+  "attrs>=24,<26",
   "boto3>=1.35.75,<2.0",
   "cloudpickle>=2.2.1",
   "docker",

requirements/extras/test_requirements.txt

@@ -17,7 +17,7 @@ stopit==1.1.2
 apache-airflow==2.10.4
 apache-airflow-providers-amazon==7.2.1
 Flask-Limiter==3.11
-attrs>=23.1.0,<24
+attrs>=24,<26
 fabric==3.2.2
 requests==2.32.2
 sagemaker-experiments==0.1.35

src/sagemaker/estimator.py

@@ -456,6 +456,9 @@ def __init__(
                 A dictionary containing the hyperparameters to
                 initialize this estimator with. (Default: None).
 
+                If a source directory is specified, the set_hyperparameters method escapes
+                the dict argument as JSON, and updates the private hyperparameter attribute.
+
                 .. caution::
                     You must not include any security-sensitive information, such as
                     account access IDs, secrets, and tokens, in the dictionary for configuring

src/sagemaker/modules/configs.py

@@ -21,7 +21,7 @@
 
 from __future__ import absolute_import
 
-from typing import Optional, Union
+from typing import Optional, Union, List
 from pydantic import BaseModel, model_validator, ConfigDict
 
 import sagemaker_core.shapes as shapes
@@ -96,12 +96,23 @@ class SourceCode(BaseConfig):
         command (Optional[str]):
             The command(s) to execute in the training job container. Example: "python my_script.py".
             If not specified, entry_script must be provided.
+        ignore_patterns: (Optional[List[str]]) :
+            The ignore patterns to ignore specific files/folders when uploading to S3. If not specified,
+            default to: ['.env', '.git', '__pycache__', '.DS_Store', '.cache', '.ipynb_checkpoints'].
     """
 
     source_dir: Optional[str] = None
     requirements: Optional[str] = None
     entry_script: Optional[str] = None
     command: Optional[str] = None
+    ignore_patterns: Optional[List[str]] = [
+        ".env",
+        ".git",
+        "__pycache__",
+        ".DS_Store",
+        ".cache",
+        ".ipynb_checkpoints",
+    ]
 
 
 class Compute(shapes.ResourceConfig):

src/sagemaker/modules/train/model_trainer.py

@@ -119,7 +119,8 @@ class ModelTrainer(BaseModel):
         from sagemaker.modules.train import ModelTrainer
         from sagemaker.modules.configs import SourceCode, Compute, InputData
 
-        source_code = SourceCode(source_dir="source", entry_script="train.py")
+        ignore_patterns = ['.env', '.git', '__pycache__', '.DS_Store', 'data']
+        source_code = SourceCode(source_dir="source", entry_script="train.py", ignore_patterns=ignore_patterns)
         training_image = "123456789012.dkr.ecr.us-west-2.amazonaws.com/my-training-image"
         model_trainer = ModelTrainer(
             training_image=training_image,
@@ -654,6 +655,7 @@ def train(
                     channel_name=SM_CODE,
                     data_source=self.source_code.source_dir,
                     key_prefix=input_data_key_prefix,
+                    ignore_patterns=self.source_code.ignore_patterns,
                 )
                 final_input_data_config.append(source_code_channel)
 
@@ -675,6 +677,7 @@ def train(
                 channel_name=SM_DRIVERS,
                 data_source=tmp_dir.name,
                 key_prefix=input_data_key_prefix,
+                ignore_patterns=self.source_code.ignore_patterns,
             )
             final_input_data_config.append(sm_drivers_channel)
 
@@ -755,7 +758,11 @@ def train(
             local_container.train(wait)
 
     def create_input_data_channel(
-        self, channel_name: str, data_source: DataSourceType, key_prefix: Optional[str] = None
+        self,
+        channel_name: str,
+        data_source: DataSourceType,
+        key_prefix: Optional[str] = None,
+        ignore_patterns: Optional[List[str]] = None,
     ) -> Channel:
         """Create an input data channel for the training job.
 
@@ -771,6 +778,10 @@ def create_input_data_channel(
 
                 If specified, local data will be uploaded to:
                 ``s3://<default_bucket_path>/<key_prefix>/<channel_name>/``
+            ignore_patterns: (Optional[List[str]]) :
+                The ignore patterns to ignore specific files/folders when uploading to S3.
+                If not specified, default to: ['.env', '.git', '__pycache__', '.DS_Store',
+                '.cache', '.ipynb_checkpoints'].
         """
         channel = None
         if isinstance(data_source, str):
@@ -810,11 +821,28 @@ def create_input_data_channel(
                     )
                     if self.sagemaker_session.default_bucket_prefix:
                         key_prefix = f"{self.sagemaker_session.default_bucket_prefix}/{key_prefix}"
-                    s3_uri = self.sagemaker_session.upload_data(
-                        path=data_source,
-                        bucket=self.sagemaker_session.default_bucket(),
-                        key_prefix=key_prefix,
-                    )
+                    if ignore_patterns and _is_valid_path(data_source, path_type="Directory"):
+                        tmp_dir = TemporaryDirectory()
+                        copied_path = os.path.join(
+                            tmp_dir.name, os.path.basename(os.path.normpath(data_source))
+                        )
+                        shutil.copytree(
+                            data_source,
+                            copied_path,
+                            dirs_exist_ok=True,
+                            ignore=shutil.ignore_patterns(*ignore_patterns),
+                        )
+                        s3_uri = self.sagemaker_session.upload_data(
+                            path=copied_path,
+                            bucket=self.sagemaker_session.default_bucket(),
+                            key_prefix=key_prefix,
+                        )
+                    else:
+                        s3_uri = self.sagemaker_session.upload_data(
+                            path=data_source,
+                            bucket=self.sagemaker_session.default_bucket(),
+                            key_prefix=key_prefix,
+                        )
                     channel = Channel(
                         channel_name=channel_name,
                         data_source=DataSource(
@@ -861,7 +889,9 @@ def _get_input_data_config(
                 channels.append(input_data)
             elif isinstance(input_data, InputData):
                 channel = self.create_input_data_channel(
-                    input_data.channel_name, input_data.data_source, key_prefix=key_prefix
+                    input_data.channel_name,
+                    input_data.data_source,
+                    key_prefix=key_prefix,
                 )
                 channels.append(channel)
             else:

src/sagemaker/serve/utils/conda_in_process.yml

@@ -8,7 +8,7 @@ dependencies:
     - fastapi>=0.111.0
     - nest-asyncio
     - pip>=23.0.1
-    - attrs>=23.1.0,<24
+    - attrs>=24,<26
     - boto3>=1.34.142,<2.0
     - cloudpickle==2.2.1
     - google-pasta

src/sagemaker/session.py

@@ -7509,7 +7509,7 @@ def get_model_package_args(
     if source_uri is not None:
         model_package_args["source_uri"] = source_uri
     if model_life_cycle is not None:
-        model_package_args["model_life_cycle"] = model_life_cycle
+        model_package_args["model_life_cycle"] = model_life_cycle._to_request_dict()
     if model_card is not None:
         original_req = model_card._create_request_args()
         if original_req.get("ModelCardName") is not None:

src/sagemaker/workflow/utilities.py

@@ -21,7 +21,15 @@
 import hashlib
 from urllib.parse import unquote, urlparse
 from contextlib import contextmanager
-from _hashlib import HASH as Hash
+
+try:
+    # _hashlib is an internal python module, and is not present in
+    # statically linked interpreters.
+    from _hashlib import HASH as Hash
+except ImportError:
+    import typing
+
+    Hash = typing.Any
 
 from sagemaker.utils import base_from_name
 from sagemaker.workflow.parameters import Parameter