AWS sagemaker documentation change

Service: sagemaker · 2026-06-28 · Documentation low

File: sagemaker/latest/dg/incremental-training.md

Summary

Added SageMaker Python SDK v3 code examples alongside existing v2 examples for incremental training workflow

Security assessment

The changes only add new SDK version examples without modifying security configurations or mentioning security vulnerabilities

Diff

diff --git a/sagemaker/latest/dg/incremental-training.md b/sagemaker/latest/dg/incremental-training.md
index fc60109cb..a98a23eaf 100644
--- a//sagemaker/latest/dg/incremental-training.md
+++ b//sagemaker/latest/dg/incremental-training.md
@@ -146,0 +147,21 @@ Get an AWS Identity and Access Management (IAM) role that grants required permis
+SageMaker Python SDK v3
+    
+    
+    
+    import sagemaker
+    from sagemaker.core.helper.session_helper import Session, get_execution_role
+    from sagemaker.train import ModelTrainer
+    from sagemaker.train.configs import Compute
+    
+    role = get_execution_role()
+    print(role)
+    
+    sess = Session()
+    
+    bucket=sess.default_bucket()
+    print(bucket)
+    prefix = 'ic-incr-training'
+
+SageMaker Python SDK v2 (Legacy)
+    
+    
@@ -222,0 +244,51 @@ Create an estimator object and train the first model using the training and vali
+SageMaker Python SDK v3
+    
+    
+    
+    from sagemaker.train.configs import InputData, StoppingCondition
+    from sagemaker.core.shapes import OutputDataConfig, Channel, DataSource, S3DataSource
+    
+    # Train the base ModelTrainer
+    s3_output_location = 's3://{}/{}/output'.format(bucket, prefix)
+    ic = ModelTrainer(training_image=training_image,
+                                       role=role,
+                                       compute=Compute(
+                                           instance_count=1,
+                                           instance_type='ml.p2.xlarge',
+                                           volume_size_in_gb=50,
+                                       ),
+                                       stopping_condition=StoppingCondition(max_runtime_in_seconds=360000),
+                                       training_input_mode='File',
+                                       output_data_config=OutputDataConfig(s3_output_path=s3_output_location),
+                                       hyperparameters=hyperparams)
+    
+    train_data = Channel(
+        channel_name="train",
+        data_source=DataSource(
+            s3_data_source=S3DataSource(
+                s3_data_type='S3Prefix',
+                s3_uri=s3train,
+                s3_data_distribution_type='FullyReplicated',
+            )
+        ),
+        content_type='application/x-recordio',
+    )
+    validation_data = Channel(
+        channel_name="validation",
+        data_source=DataSource(
+            s3_data_source=S3DataSource(
+                s3_data_type='S3Prefix',
+                s3_uri=s3validation,
+                s3_data_distribution_type='FullyReplicated',
+            )
+        ),
+        content_type='application/x-recordio',
+    )
+    
+    data_channels = {'train': train_data, 'validation': validation_data}
+    
+    ic.train(input_data_config=[train_data, validation_data])
+
+SageMaker Python SDK v2 (Legacy)
+    
+    
@@ -247,0 +320,25 @@ To use the model to incrementally train another model, create a new estimator ob
+SageMaker Python SDK v3
+    
+    
+    
+    # Given the base ModelTrainer, create a new one for incremental training
+    incr_ic = ModelTrainer(training_image=training_image,
+                                            role=role,
+                                            compute=Compute(
+                                                instance_count=1,
+                                                instance_type='ml.p2.xlarge',
+                                                volume_size_in_gb=50,
+                                            ),
+                                            stopping_condition=StoppingCondition(max_runtime_in_seconds=360000),
+                                            training_input_mode='File',
+                                            output_data_config=OutputDataConfig(s3_output_path=s3_output_location),
+                                            hyperparameters=hyperparams,
+                                            # Pass the previous model artifacts as an input data channel
+                                            input_data_config=[
+                                                InputData(channel_name="model", data_source=ic.model_data),
+                                            ])
+    incr_ic.train(input_data_config=[train_data, validation_data])
+
+SageMaker Python SDK v2 (Legacy)
+    
+