### Install Dependencies for Building Docs

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/CONTRIBUTING.md

Initial setup for building Sphinx documentation. This command installs necessary requirements and the SDK in editable mode.

```shell
# Initial setup, only required for the first run
pip install -r requirements.txt
pip install -e ../
```

--------------------------------

### Setup Environment and Install Dependencies

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/model-customization-examples/recipe_override_evaluator_example.ipynb

Sets up the Python environment by adjusting sys.path and environment variables, and installs necessary packages like omegaconf and pyyaml.

```python
# Setup environment
import sys
import os
sys.path.insert(0, '../../sagemaker-train/src')
sys.path.insert(0, '../../sagemaker-core/src')

# Point botocore at the bundled service model
os.environ['AWS_DATA_PATH'] = os.path.abspath('../../sagemaker-core/sample')

!pip install -q omegaconf pyyaml
```

--------------------------------

### Install Dependencies and Set Environment

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/model-customization-examples/recipe_override_sft_trainer_example.ipynb

Installs necessary libraries and configures environment variables for SageMaker training. This is a prerequisite for running the subsequent code examples.

```python
import sys
import os
sys.path.insert(0, '../../sagemaker-train/src')
sys.path.insert(0, '../../sagemaker-core/src')

os.environ['AWS_DATA_PATH'] = os.path.abspath('../../sagemaker-core/sample')

!pip install -q omegaconf pyyaml
```

--------------------------------

### Install Project

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-train/CONTRIBUTING.md

Install the project locally using pip. Ensure you are in the project's root directory.

```bash
pip install .
```

--------------------------------

### Install SageMaker Python SDK V3 Core Functionality

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/installation.md

Installs only the core components of the SageMaker Python SDK V3, useful for minimal setups.

```bash
pip install sagemaker-core
```

--------------------------------

### Install SageMaker Python SDK from Source

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/README.rst

Install the SageMaker Python SDK directly from its GitHub repository. This involves cloning the repository and then running a pip install command from the root directory.

```bash
git clone https://github.com/aws/sagemaker-python-sdk.git
cd sagemaker-python-sdk
pip install .
```

--------------------------------

### Install All SageMaker Packages

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-mlops/README.md

Install all SageMaker packages together in editable mode for development.

```bash
pip install -e sagemaker-core
pip install -e sagemaker-train
pip install -e sagemaker-serve
pip install -e sagemaker-mlops
```

--------------------------------

### Install SageMaker SDK Helper

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/migration.md

Install the SageMaker SDK helper tool for migration. Use --force-reinstall to ensure the latest version is applied.

```bash
pip install --no-cache-dir https://d3azyja9oqj8z1.cloudfront.net/sagemaker_sdk_helper-0.2.0.tar.gz --force-reinstall
```

--------------------------------

### Verify SageMaker SDK Helper Installation

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/migration.md

Verify the installation of the SageMaker SDK helper by checking its executable path and running the help command.

```bash
which sagemaker-sdk-helper   # Should output the path to the executable
```

```bash
sagemaker-sdk-helper --help   # Test the server runs
```

--------------------------------

### Install Test Libraries

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/README.rst

Installs the necessary libraries for running tests. Use the first command for general use and the second for Zsh users.

```bash
pip install --upgrade .[test]
```

```bash
pip install --upgrade .[test]
```

--------------------------------

### Install Helm 3

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/model_customization/finetuning_hyperpod.md

Installs Helm 3, a prerequisite for the SageMaker HyperPod CLI. This script downloads, makes executable, and installs Helm, then verifies the installation.

```bash
curl -fsSL -o get_helm.sh https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3
chmod 700 get_helm.sh
./get_helm.sh
rm -f ./get_helm.sh
helm version  # Verify installation
```

--------------------------------

### Install Documentation Dependencies with Pip

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/README.rst

Installs documentation dependencies using pip from the requirements.txt file.

```bash
pip install -r doc/requirements.txt
```

--------------------------------

### start

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/index.md

Starts a SageMaker Notebook Instance. This operation can be used to resume a previously stopped instance.

```APIDOC
## start

### Description
Starts a NotebookInstance resource.

### Method
Not specified (SDK method)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Parameters
* **session** (*Session* | *None*) – Boto3 session.
* **region** (*str* | *None*) – Region name.

### Raises
* **botocore.exceptions.ClientError** – This exception is raised for AWS service related errors.
* **ResourceLimitExceeded** – You have exceeded an SageMaker resource limit.

### Return type:
None
```

--------------------------------

### Install Python and Activate Virtual Environment

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/CONTRIBUTING.md

Installs a specific Python version (3.10.14), creates a virtual environment named 'py3.10.14', and activates it.

```bash
pyenv install 3.10.14
pyenv virtualenv 3.10.14 py3.10.14
pyenv activate py3.10.14
```

--------------------------------

### Quick Install SageMaker Python SDK V3

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/installation.md

Installs the latest version of the SageMaker Python SDK V3. This is the most straightforward method for general use.

```bash
pip install sagemaker
```

--------------------------------

### Install SageMaker Python SDK V3 Training Capabilities

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/installation.md

Installs the SageMaker Python SDK V3 components specifically for model training tasks.

```bash
pip install sagemaker-train
```

--------------------------------

### Setup and Configuration

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/model-customization-examples/rlvr_finetuning_example_notebook_v3_prod.ipynb

Imports necessary libraries and configures AWS credentials and region for SageMaker RLVR finetuning.

```python
# Configure AWS credentials and region
#! ada credentials update --provider=isengard --account=<> --role=Admin --profile=default --once
#! aws configure set region us-west-2

from sagemaker.train.rlvr_trainer import RLVRTrainer
from sagemaker.train.configs import InputData
from rich import print as rprint
from rich.pretty import pprint
from sagemaker.core.resources import ModelPackage
from sagemaker.train.common import TrainingType


import boto3
import os
from sagemaker.core.helper.session_helper import Session

# For MLFlow native metrics in Trainer wait, run below line with approriate region
os.environ["SAGEMAKER_MLFLOW_CUSTOM_ENDPOINT"] = "https://mlflow.sagemaker.us-west-2.app.aws"

```

--------------------------------

### Start a SageMaker Notebook Instance

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/index.md

Initiates a SageMaker Notebook Instance. Ensure you have the necessary permissions and that resource limits are not exceeded.

```python
# Assuming 'notebook_instance' is an instance of a SageMaker NotebookInstance class
notebook_instance.start()
```

--------------------------------

### Get All Training Plans

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/api/sagemaker_core.md

Retrieves a list of all TrainingPlan resources, with options for filtering by start time, sorting, and pagination.

```APIDOC
## classmethod get_all

### Description
Get all TrainingPlan resources.

### Method
classmethod

### Parameters
#### Query Parameters
- **start_time_after** (datetime) - Optional - Filter to list only training plans with an actual start time after this date.
- **start_time_before** (datetime) - Optional - Filter to list only training plans with an actual start time before this date.
- **sort_by** (str | PipelineVariable) - Optional - The training plan field to sort the results by (e.g., StartTime, Status).
- **sort_order** (str | PipelineVariable) - Optional - The order to sort the results (Ascending or Descending).
- **filters** (List[TrainingPlanFilter]) - Optional - Additional filters to apply to the list of training plans.
- **session** (boto3.session.Session) - Optional - Boto3 session.
- **region** (str | PipelineVariable) - Optional - Region name.

### Returns
ResourceIterator[[TrainingPlan](#sagemaker.core.resources.TrainingPlan)] - Iterator for listed TrainingPlan resources.

### Raises
- **botocore.exceptions.ClientError** - This exception is raised for AWS service related errors.
      The error message and error code can be parsed from the exception as follows:
      ``
      try:
          # AWS service call here
      except botocore.exceptions.ClientError as e:
          error_message = e.response['Error']['Message']
          error_code = e.response['Error']['Code']
      ``
```

--------------------------------

### Basic SageMaker Session Setup

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/quickstart.md

Import the SDK and create a SageMaker session. This also retrieves the execution role and default S3 bucket.

```python
from sagemaker.core.helper.session_helper import Session, get_execution_role

# Create a SageMaker session
session = Session()
role = get_execution_role()

print(f"Using role: {role}")
print(f"Default bucket: {session.default_bucket()}")
```

--------------------------------

### Get All Training Plans

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/index.md

Retrieves an iterator for all SageMaker TrainingPlan resources. Supports filtering by start time, sorting, and applying additional filters.

```APIDOC
## Get All Training Plans

### Description
Retrieves an iterator for all SageMaker TrainingPlan resources. Supports filtering by start time, sorting, and applying additional filters.

### Method
classmethod

### Signature
get_all(start_time_after=Unassigned(), start_time_before=Unassigned(), sort_by=Unassigned(), sort_order=Unassigned(), filters=Unassigned(), session=None, region=None)

### Parameters
#### Optional Parameters
* **next_token** - A token to continue pagination if more results are available.
* **max_results** - The maximum number of results to return in the response.
* **start_time_after** (datetime | None) - Filter to list only training plans with an actual start time after this date.
* **start_time_before** (datetime | None) - Filter to list only training plans with an actual start time before this date.
* **sort_by** (str | PipelineVariable | None) - The training plan field to sort the results by (e.g., StartTime, Status).
* **sort_order** (str | PipelineVariable | None) - The order to sort the results (Ascending or Descending).
* **filters** (List[TrainingPlanFilter] | None) - Additional filters to apply to the list of training plans.
* **session** (Session | None) - Boto3 session.
* **region** (str | PipelineVariable | None) - Region name.

### Returns
Iterator for listed TrainingPlan resources.

### Raises
* **botocore.exceptions.ClientError** - This exception is raised for AWS service related errors.
```

--------------------------------

### Create ModelBuilder from JumpStart Configuration

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_serve.md

This example shows how to instantiate a ModelBuilder using a JumpStart configuration, specifying compute resources. It's ideal for deploying pre-trained models from SageMaker JumpStart.

```python
>>> from sagemaker.core.jumpstart.configs import JumpStartConfig
>>> from sagemaker.serve.model_builder import ModelBuilder
>>> 
>>> js_config = JumpStartConfig(
...     model_id="huggingface-llm-mistral-7b",
...     model_version="*"
... )
>>> 
>>> from sagemaker.core.training.configs import Compute
>>> 
>>> model_builder = ModelBuilder.from_jumpstart_config(
...     jumpstart_config=js_config,
...     compute=Compute(instance_type="ml.g5.2xlarge", instance_count=1)
... )
>>> 
>>> model = model_builder.build()  # Creates Model resource
>>> endpoint = model_builder.deploy()  # Creates Endpoint resource
>>> result = endpoint.invoke(data=input_data)  # Make predictions
```

--------------------------------

### Initialize SageMaker Session (V3)

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/ml_ops/lineage.md

Sets up a SageMaker session and default bucket using V3 import paths. This is the recommended approach for new projects.

```python
import boto3
from sagemaker.core.helper.session_helper import Session

region = boto3.Session().region_name
sagemaker_session = Session()
default_bucket = sagemaker_session.default_bucket()
```

--------------------------------

### Get All Training Plans with Filters

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_core.md

Retrieves all TrainingPlan resources, allowing filtering by start time, sorting, and additional custom filters. Requires a boto3 session and optionally a region.

```python
sagemaker_core.TrainingPlan.get_all(start_time_after=datetime(2023, 1, 1), sort_by='StartTime', sort_order='Ascending')
```

--------------------------------

### Implement Custom Evaluator in SageMaker

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_train.md

Subclasses of BaseEvaluator must implement the evaluate method to define custom evaluation logic. This example shows how to create a pipeline definition and start an execution.

```python
>>> # In a subclass implementation
>>> class CustomEvaluator(BaseEvaluator):
...     def evaluate(self):
...         # Create pipeline definition
...         pipeline_definition = self._build_pipeline()
...         # Start execution
...         return EvaluationPipelineExecution.start(...)
```

--------------------------------

### Setup SageMaker session and create temporary directories

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/training-examples/local-training-example.ipynb

Initializes a SageMaker session, retrieves the default training image for PyTorch, and creates temporary directories for source code and data. It also sets the Docker platform for Apple Silicon compatibility.

```python
sagemaker_session = Session()
region = sagemaker_session.boto_region_name

# Get the correct ECR image for your region
from sagemaker.core import image_uris
DEFAULT_CPU_IMAGE = image_uris.retrieve(
    framework="pytorch",
    region=region,
    version="2.0.0",
    py_version="py310",
    instance_type="ml.m5.xlarge",
    image_scope="training"
)

# Set Docker platform for Apple Silicon compatibility
import platform
if platform.machine() == 'arm64':
    os.environ['DOCKER_DEFAULT_PLATFORM'] = 'linux/amd64'

# Create temporary directories
temp_dir = tempfile.mkdtemp()
source_dir = os.path.join(temp_dir, "source")
data_dir = os.path.join(temp_dir, "data")
train_dir = os.path.join(data_dir, "train")
test_dir = os.path.join(data_dir, "test")

os.makedirs(source_dir, exist_ok=True)
os.makedirs(train_dir, exist_ok=True)
os.makedirs(test_dir, exist_ok=True)

print(f"Created temporary directories in: {temp_dir}")
```

--------------------------------

### LLMAsJudgeEvaluator with Built-in Metrics and Wait

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_train.md

Initiate an LLM-as-judge evaluation job using built-in metrics and wait for its completion. This example demonstrates a basic setup for evaluating a base model against an evaluator model.

```python
evaluator = LLMAsJudgeEvaluator(
    base_model="llama-3-3-70b-instruct",
    evaluator_model="anthropic.claude-3-5-sonnet-20240620-v1:0",
    dataset="s3://my-bucket/my-dataset.jsonl",
    builtin_metrics=["Correctness", "Helpfulness"],
    s3_output_path="s3://my-bucket/output"
)
execution = evaluator.evaluate()
execution.wait()
```

--------------------------------

### V3 Resource Chaining Example

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/migration.md

Demonstrates V3 resource chaining by training a model and then using its output to build and deploy a model.

```python
# Train a model
model_trainer = ModelTrainer(...)
model_trainer.train()

# Chain training output to model builder
model_builder = ModelBuilder(model=model_trainer)

# Deploy the trained model
endpoint = model_builder.deploy()
```

--------------------------------

### Serve Sphinx Documentation Locally

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/README.rst

Starts a local Python web server to preview the built HTML documentation. Access it via http://localhost:8000.

```bash
cd _build/html
python -m http.server 8000
```

--------------------------------

### Initialize, Build, and Deploy Model with ModelBuilder

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_serve.md

Initialize ModelBuilder with a trained model, then build the SageMaker Model resource and deploy it to an endpoint. Finally, invoke the endpoint to get predictions. This example demonstrates the core workflow for deploying a model.

```python
from sagemaker.serve.model_builder import ModelBuilder
from sagemaker.serve.mode.function_pointers import Mode

# Initialize with a trained model
model_builder = ModelBuilder(
    model=my_pytorch_model,
    role_arn="arn:aws:iam::123456789012:role/SageMakerRole",
    instance_type="ml.m5.xlarge"
)

# Build the model (creates SageMaker Model resource)
model = model_builder.build()

# Deploy to endpoint (creates SageMaker Endpoint resource)
endpoint = model_builder.deploy(endpoint_name="my-endpoint")

# Make predictions
result = endpoint.invoke(data=input_data)
```

--------------------------------

### Alternative Data Mix Configurations

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/model-customization-examples/nova_data_mixing.ipynb

Examples of alternative `DataMixingConfig` setups for different specialization needs: high specialization (90% custom data), balanced mix (50% custom data), and light specialization (30% custom data).

```python
# High specialization: mostly your data
high_specialization = DataMixingConfig(
    customer_data_percent=90.0,
    nova_data_percentages={
        "reasoning": 100.0,
    },
)

# Balanced: equal split with multiple Nova categories
balanced_mix = DataMixingConfig(
    customer_data_percent=50.0,
    nova_data_percentages={
        "code": 40.0,
        "reasoning": 30.0,
        "math": 30.0,
    },
)

# Light specialization: preserve broad capabilities
light_specialization = DataMixingConfig(
    customer_data_percent=30.0,
    nova_data_percentages={
        "code": 25.0,
        "math": 25.0,
        "chat": 25.0,
        "reasoning": 25.0,
    },
)
```

--------------------------------

### Access and Modify Evaluation Hyperparameters

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_train.md

Illustrates how to access, modify, and retrieve evaluation hyperparameters using the `hyperparameters` property of the BenchMarkEvaluator. Includes examples of getting current values, setting new values with validation, converting to a dictionary, and displaying parameter information.

```python
evaluator = BenchMarkEvaluator(...)

# Access current values
print(evaluator.hyperparameters.temperature)

# Modify values (with validation)
evaluator.hyperparameters.temperature = 0.5

# Get as dictionary
params = evaluator.hyperparameters.to_dict()

# Display parameter information
evaluator.hyperparameters.get_info()
evaluator.hyperparameters.get_info('temperature')
```

--------------------------------

### SFT Trainer Nova Testing Setup

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/model-customization-examples/sft_finetuning_example_notebook_pysdk_prod_v3.ipynb

Configure environment variables and initialize an `SFTTrainer` for testing with the Nova model. Ensure to replace placeholder values for bucket paths.

```python
os.environ['SAGEMAKER_REGION'] = 'us-east-1'

# For fine-tuning 
sft_trainer_nova = SFTTrainer(
    #model="test-nova-lite-v2", 
    #model="nova-textgeneration-micro",
    model="nova-textgeneration-lite-v2",
    training_type=TrainingType.LORA, 
    model_package_group="sdk-test-finetuned-models", 
    mlflow_experiment_name="test-nova-finetuned-models-exp", 
    mlflow_run_name="test-nova-finetuned-models-run", 
    training_dataset="arn:aws:sagemaker:us-east-1:<>:hub-content/sdktest/DataSet/sft-nova-test-dataset/0.0.1",
    s3_output_path="s3://<your-bucket>/output/"  # TODO: replace with your S3 output path
)
```

--------------------------------

### start

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/index.md

Start a PipelineExecution resource.

```APIDOC
## start

### Description
Start a PipelineExecution resource.

### Method
Not explicitly defined, assumed to be a Python method call.

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Parameters
* **session** (*Session* | None) – Boto3 session.
* **region** (*str* | None) – Region name.
* **pipeline_name** (*str* | *PipelineVariable*)
* **client_request_token** (*str* | *PipelineVariable*)
* **pipeline_parameters** (*List*[*Parameter*] | None)
* **mlflow_experiment_name** (*str* | *PipelineVariable* | None)

### Raises
* **botocore.exceptions.ClientError** – This exception is raised for AWS service related errors.
* **ConflictException** – There was a conflict when you attempted to modify a SageMaker entity such as an Experiment or Artifact.
* **ResourceLimitExceeded** – You have exceeded an SageMaker resource limit.
* **ResourceNotFound** – Resource being accessed is not found.

### Return type:
None
```

--------------------------------

### SageMaker Entry Point Configuration V2 vs V3

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/training/index.md

Demonstrates the configuration of the entry point script for training jobs in V2 and V3. V3 uses a `SourceCode` object.

```python
entry_point="train.py"
```

```python
SourceCode(entry_script="train.py")
```

--------------------------------

### Install Python Dependencies

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/example_notebooks/sagemaker-core-llama-3-8B.ipynb

Installs or upgrades necessary Python packages including sagemaker-core and huggingface_hub. Use --quiet for silent installation.

```python
%pip install pip --upgrade --quiet
%pip install sagemaker-core huggingface_hub --quiet
```

--------------------------------

### Setup SageMaker Session and Retrieve Training Image

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/training-examples/aws_batch/sm-training-queues_getting_started_with_model_trainer.ipynb

Initializes logging, a SageMaker session, and retrieves the appropriate training image URI for PyTorch.

```python
import logging

logging.basicConfig(
    level=logging.INFO, format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
)
logging.getLogger("botocore.client").setLevel(level=logging.WARN)
logger = logging.getLogger(__name__)

from sagemaker.core.helper.session_helper import Session
from sagemaker.core import image_uris

session = Session()

image_uri = image_uris.retrieve(
    framework="pytorch",
    region=session.boto_session.region_name,
    version="2.5",
    instance_type=INSTANCE_TYPE,
    image_scope="training",
)
```

--------------------------------

### Install MLflow for Path Resolution Fixes

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/ml-ops-examples/v3-mlflow-train-inference-e2e-example.ipynb

Installs a specific version of MLflow to address known issues with model path resolution. A kernel restart is required after installation.

```python
# Install fix for MLflow path resolution issues
%pip install mlflow==3.4.0
```

--------------------------------

### Initialize SageMaker Session and Setup Directories

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/training-examples/distributed-local-training-example.ipynb

Initializes the SageMaker session and creates temporary directories for source code and data. This setup is crucial for organizing training artifacts.

```python
sagemaker_session = Session()
region = sagemaker_session.boto_region_name

DEFAULT_CPU_IMAGE = image_uris.retrieve(
    framework="pytorch",
    region=region,
    version="2.0.0",
    py_version="py310",
    instance_type="ml.m5.xlarge",
    image_scope="training"
)

# Create temporary directories
temp_dir = tempfile.mkdtemp()
source_dir = os.path.join(temp_dir, "source")
data_dir = os.path.join(temp_dir, "data")
train_dir = os.path.join(data_dir, "train")
test_dir = os.path.join(data_dir, "test")

os.makedirs(source_dir, exist_ok=True)
os.makedirs(train_dir, exist_ok=True)
os.makedirs(test_dir, exist_ok=True)

print(f"Created temporary directories in: {temp_dir}")
print("Note: This will use multiple Docker containers locally for distributed training!")
```

--------------------------------

### start

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/api/sagemaker_mlops.md

Starts a Pipeline execution in the Workflow service with optional parameters for customization.

```APIDOC
## start

### Description
Starts a Pipeline execution in the Workflow service. Allows overriding pipeline parameters and specifying execution details.

### Method
Not specified (assumed to be a method call in a Python SDK)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Parameters
- **parameters** (Dict[str, Union[str, bool, int, float]], optional) - Values to override pipeline parameters.
- **execution_display_name** (str, optional) - The display name of the pipeline execution.
- **execution_description** (str, optional) - A description of the execution.
- **parallelism_config** (ParallelismConfiguration, optional) - Parallelism configuration that is applied to each of the executions of the pipeline. It takes precedence over the parallelism configuration of the parent pipeline.
- **selective_execution_config** (SelectiveExecutionConfig, optional) - The configuration for selective step execution.
- **mlflow_experiment_name** (str, optional) - Optional MLflow experiment name to override the experiment name specified in the pipeline’s mlflow_config. If provided, this will override the experiment name for this specific pipeline execution only, without modifying the pipeline definition.
- **pipeline_version_id** (int, optional) - Version ID of the pipeline to start the execution from. If not specified, uses the latest version ID.

### Response
#### Success Response
- **PipelineExecution** - A PipelineExecution instance, if successful.

### Request Example
```python
# Example usage (Python SDK):
parameters = {"param1": "value1"}
sagemaker_pipeline.start(parameters=parameters, execution_display_name="MyExecution")
```

### Response Example
```json
{
  "example": "PipelineExecution(arn='arn:aws:sagemaker:us-east-1:123456789012:pipeline/MyPipeline/execution/MyExecutionId')"
}
```
```

--------------------------------

### Install SageMaker MLOps Package

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-mlops/README.md

Install the sagemaker-mlops package in editable mode for development.

```bash
pip install -e sagemaker-mlops
```

--------------------------------

### Setup SageMaker Session and Create Test Files

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/training-examples/custom-distributed-training-example.ipynb

Initializes the SageMaker session, retrieves the execution role and region, and sets up temporary directories for custom driver and script files. It also retrieves the default PyTorch training image.

```python
sagemaker_session = Session()
role = get_execution_role()
region = sagemaker_session.boto_region_name

DEFAULT_CPU_IMAGE = image_uris.retrieve(
    framework="pytorch",
    region=region,
    version="2.0.0",
    py_version="py310",
    instance_type="ml.m5.xlarge",
    image_scope="training"
)

# Create temporary directories
temp_dir = tempfile.mkdtemp()
custom_drivers_dir = os.path.join(temp_dir, "custom_drivers")
scripts_dir = os.path.join(temp_dir, "scripts")

os.makedirs(custom_drivers_dir, exist_ok=True)
os.makedirs(scripts_dir, exist_ok=True)

print(f"Created temporary directories in: {temp_dir}")
```

--------------------------------

### Install Additional Packages

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/example_notebooks/sagemaker_core_overview.ipynb

Installs or upgrades scikit-learn, pandas, and boto3 to their latest versions.

```bash
# Install additional packages

!pip install -U scikit-learn pandas boto3
```

--------------------------------

### Install SageMaker Python SDK from Source

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-train/README.rst

Clone the repository, navigate to the sagemaker_train directory, and install using pip. This method is useful for development or when needing the latest unreleased changes.

```bash
git clone https://github.com/aws/sagemaker-python-sdk-staging.git
cd sagemaker-python-sdk-staging/sagemaker_train
pip install .
```

--------------------------------

### start

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/index.md

Starts a MonitoringSchedule resource. This operation can be used to initiate a scheduled monitoring job.

```APIDOC
## start

### Description
Starts a MonitoringSchedule resource.

### Method
Not applicable (Python SDK method)

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Parameters
* **session** (*Session* | *None*) – Boto3 session.
* **region** (*str* | *None*) – Region name.

### Raises
* **botocore.exceptions.ClientError** – This exception is raised for AWS service related errors.
* **ResourceNotFound** – Resource being access is not found.

### Return type
None
```

--------------------------------

### Intelligent Defaults Configuration Example

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/example_notebooks/intelligent_defaults_and_logging.ipynb

Example JSON structure for configuring intelligent defaults, including global and resource-specific settings for VPC configuration and training job parameters. Replace placeholder values with your actual resource IDs and ARNs.

```json
{
    "SchemaVesion": "1.0",
    "SageMaker": {
        "PythonSDK": {
            "Resources": {
                "GlobalDefaults": {
                    "vpc_config": {
                        "security_group_ids": [
                            "sg-xxxxxxxxxxxxxxxxx" // Replace with security group id
                        ],
                        "subnets": [
                            "subnet-xxxxxxxxxxxxxxxxx", // Replace with subnet id
                            "subnet-xxxxxxxxxxxxxxxxx" // Replace with subnet id
                        ]
                    }
                    // ...
                },
                "TrainingJob": {
                    "role_arn": "arn:aws:xxxxxxxxxxx:role/xxxxx", // Replace with role arn
                    "output_data_config": {
                        "s3_output_path": "s3://xxxxxxxxxxx", // Replace with S3 URI
                    },
                    // ...
                }
            }
        }
    }
}
```

--------------------------------

### start

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/index.md

Starts a InferenceExperiment resource. This operation initiates the execution of a specified inference experiment.

```APIDOC
## start

### Description
Start a InferenceExperiment resource. This operation initiates the execution of a specified inference experiment.

### Method
Method

### Parameters
#### Path Parameters
None

#### Query Parameters
* **session** (Session | None) - Boto3 session.
* **region** (str | None) - Region name.

### Raises
* **botocore.exceptions.ClientError** - This exception is raised for AWS service related errors.
* **ConflictException** - There was a conflict when you attempted to modify a SageMaker entity such as an Experiment or Artifact.
* **ResourceNotFound** - Resource being access is not found.

### Return type
None
```

--------------------------------

### Setup SageMaker Session and Environment

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/training-examples/hyperparameter-training-example.ipynb

Initializes the SageMaker session, retrieves the execution role and region, defines expected hyperparameters, and sets up a temporary directory for training artifacts. It also retrieves the default training image URI.

```python
sagemaker_session = Session()
role = get_execution_role()
region = sagemaker_session.boto_region_name

# Expected hyperparameters
EXPECTED_HYPERPARAMETERS = {
    "integer": 1,
    "boolean": True,
    "float": 3.14,
    "string": "Hello World",
    "list": [1, 2, 3],
    "dict": {
        "string": "value",
        "integer": 3,
        "float": 3.14,
        "list": [1, 2, 3],
        "dict": {"key": "value"},
        "boolean": True,
    },
}

DEFAULT_CPU_IMAGE = image_uris.retrieve(
    framework="pytorch",
    region=region,
    version="2.0.0",
    py_version="py310",
    instance_type="ml.m5.xlarge",
    image_scope="training"
)

# Create temporary directory
temp_dir = tempfile.mkdtemp()
source_dir = os.path.join(temp_dir, "source")
os.makedirs(source_dir, exist_ok=True)

print(f"Created temporary directory: {temp_dir}")
```

--------------------------------

### Train and Deploy JumpStart Model with ModelBuilder

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/inference/index.md

Train a JumpStart model using ModelTrainer.from_jumpstart_config() and then deploy it via ModelBuilder. Ensure necessary imports are present.

```python
from sagemaker.train.model_trainer import ModelTrainer
from sagemaker.serve.model_builder import ModelBuilder
from sagemaker.core.jumpstart.configs import JumpStartConfig

jumpstart_config = JumpStartConfig(model_id="huggingface-spc-bert-base-cased")

trainer = ModelTrainer.from_jumpstart_config(
    jumpstart_config=jumpstart_config,
    base_job_name="js-training",
    hyperparameters={\"epochs\": 1},
)
trainer.train()

model_builder = ModelBuilder(model=trainer, dependencies={\"auto\": False})
core_model = model_builder.build(model_name="bert-trained")
endpoint = model_builder.deploy(endpoint_name="bert-endpoint")
```

--------------------------------

### Install Additional Python Packages

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/example_notebooks/inference_and_resource_chaining.ipynb

Installs or upgrades scikit-learn, pandas, and boto3 to their latest versions.

```bash
!pip install -U scikit-learn pandas boto3
```

--------------------------------

### SageMaker Migration Tool Usage Examples

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/migration.md

Examples of how to use the SageMaker migration tools for analyzing, transforming, and asking questions about V2 to V3 code migration.

```bash
# Analyze V2 code
Analyze this V2 code for migration:
[paste your V2 code]

# Transform V2 code to V3
Transform this V2 code to V3:
[paste your V2 code]

# Ask migration questions
What is ModelTrainer in V3?
How do I migrate Estimator to V3?

```

--------------------------------

### Import SageMaker SDK and Initialize Session

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/v3-examples/model-customization-examples/rlaif_finetuning_example_notebook_v3_prod.ipynb

Import necessary classes from the SageMaker SDK and initialize the SageMaker session. Configure MLFlow endpoint if needed for native metrics.

```python
#!/usr/bin/env python3

from sagemaker.train.rlaif_trainer import RLAIFTrainer
from sagemaker.train.configs import InputData
from rich import print as rprint
from rich.pretty import pprint
from sagemaker.core.resources import ModelPackage
import os
#os.environ['SAGEMAKER_REGION'] = 'us-east-1'

import boto3
from sagemaker.core.helper.session_helper import Session

# For MLFlow native metrics in Trainer wait, run below line with approriate region
os.environ["SAGEMAKER_MLFLOW_CUSTOM_ENDPOINT"] = "https://mlflow.sagemaker.us-west-2.app.aws"
```

--------------------------------

### Verify HyperPod CLI Installation

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/model_customization/finetuning_hyperpod.md

Checks if the HyperPod CLI is installed correctly by displaying its help message.

```bash
hyperpod --help
```

--------------------------------

### Deploy JumpStart Model with ModelBuilder

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/inference/index.md

This snippet demonstrates deploying pre-trained models from the JumpStart hub using ModelBuilder.from_jumpstart_config(), specifying compute resources and model configuration.

```python
from sagemaker.serve.model_builder import ModelBuilder
from sagemaker.core.jumpstart.configs import JumpStartConfig
from sagemaker.train.configs import Compute

compute = Compute(instance_type="ml.g5.2xlarge")
jumpstart_config = JumpStartConfig(model_id="huggingface-llm-falcon-7b-bf16")

model_builder = ModelBuilder.from_jumpstart_config(
    jumpstart_config=jumpstart_config,
    compute=compute,
)

core_model = model_builder.build(model_name="falcon-model")
endpoint = model_builder.deploy(endpoint_name="falcon-endpoint")

result = endpoint.invoke(
    body=json.dumps({"inputs": "What are falcons?", "parameters": {"max_new_tokens": 32}}),
    content_type="application/json"
)
```

--------------------------------

### Install Pyenv

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/CONTRIBUTING.md

Installs Pyenv, a tool for managing multiple Python versions, using a curl command.

```bash
curl https://pyenv.run | bash
```

--------------------------------

### Initialize SageMaker Session (V2 Legacy)

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/docs/ml_ops/lineage.md

Sets up a SageMaker session and default bucket using V2 import paths. This code is for legacy projects.

```python
import boto3
import sagemaker

region = boto3.Session().region_name
sagemaker_session = sagemaker.session.Session()
default_bucket = sagemaker_session.default_bucket()
```

--------------------------------

### start

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/api/sagemaker_core.md

Starts a SageMaker PipelineExecution resource. This is the entry point for initiating a pipeline run with specified parameters.

```APIDOC
## start

### Description
Start a PipelineExecution resource.

### Method
Not explicitly defined, but implies an SDK method call.

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
None

### Parameters
* **pipeline_name** (str | PipelineVariable) - Required - The name of the pipeline to start.
* **client_request_token** (str | PipelineVariable) - Required - A unique, case-sensitive identifier that you provide to ensure the idempotency of the operation.
* **pipeline_parameters** (List[Parameter] | None) - Optional - A list of parameters to use for the pipeline execution.
* **mlflow_experiment_name** (str | PipelineVariable | None) - Optional - The MLflow experiment name to associate with the pipeline execution.
* **session** (boto3.session.Session | None) - Optional - Boto3 session.
* **region** (str | None) - Optional - Region name.

### Raises
* **botocore.exceptions.ClientError** - This exception is raised for AWS service related errors.
* **ConflictException** - There was a conflict when you attempted to modify a SageMaker entity.
* **ResourceLimitExceeded** - You have exceeded an SageMaker resource limit.
* **ResourceNotFound** - Resource being accessed is not found.
```

--------------------------------

### Create Training Job: Boto3 vs SageMaker Core

Source: https://github.com/aws/sagemaker-python-sdk/blob/master/sagemaker-core/docs/sagemaker_core/index.md

Compares the traditional Boto3 approach with the simplified SageMaker Core approach for creating a training job. SageMaker Core requires fewer parameters due to intelligent defaults.

```python
import boto3

client = boto3.client('sagemaker')
response = client.create_training_job(
    TrainingJobName='my-training-job',
    RoleArn='arn:aws:iam::123456789012:role/SageMakerRole',
    InputDataConfig=[
        {
            'ChannelName': 'training',
            'DataSource': {
                'S3DataSource': {
                    'S3DataType': 'S3Prefix',
                    'S3Uri': 's3://my-bucket/train',
                    'S3DataDistributionType': 'FullyReplicated'
                }
            }
        }
    ],
    # ... many more required parameters
)
```

```python
from sagemaker.core.resources import TrainingJob
from sagemaker.core.shapes import TrainingJobConfig

training_job = TrainingJob.create(
    training_job_name="my-training-job",
    role_arn="arn:aws:iam::123456789012:role/SageMakerRole",
    input_data_config=[
        {
            "channel_name": "training",
            "data_source": "s3://my-bucket/train"
        }
    ]
)
```