sagemaker.serve.constants

Constants and enums for SageMaker ModelBuilder and serving functionality.

This module defines: - Framework enum for ML framework identification - Supported model servers and local modes - Default serializers and deserializers by framework - Configuration constants for model serving

Example

Using Framework enum:

from sagemaker.serve.constants import Framework, DEFAULT_SERIALIZERS_BY_FRAMEWORK

# Get serializers for PyTorch
serializer, deserializer = DEFAULT_SERIALIZERS_BY_FRAMEWORK[Framework.PYTORCH]

Classes

Framework(value)

Enumeration of supported ML frameworks for ModelBuilder.

class sagemaker.serve.constants.Framework(value)

Bases: Enum

Enumeration of supported ML frameworks for ModelBuilder.

This enum provides standardized framework identifiers used throughout the ModelBuilder ecosystem for: - Framework detection from container images - Serializer/deserializer selection - Model server compatibility

Example

Using framework enum:

if detected_framework == Framework.PYTORCH:
    serializer, deserializer = DEFAULT_SERIALIZERS_BY_FRAMEWORK[Framework.PYTORCH]

CHAINER = 'Chainer'

DJL = 'DJL'

HUGGINGFACE = 'HuggingFace'

LDA = 'LDA'

MXNET = 'MXNet'

NTM = 'NTM'

PYTORCH = 'PyTorch'

SKLEARN = 'SKLearn'

SMD = 'SMD'

SPARKML = 'SparkML'

TENSORFLOW = 'TensorFlow'

XGBOOST = 'XGBoost'