cursus.steps.configs.config_tsa_training_step

TSA Training Step Configuration with Self-Contained Derivation Logic

This module implements the configuration class for SageMaker TSA (Temporal Self-Attention) Training steps using a self-contained design where derived fields are private with read-only properties. Fields are organized into three tiers: 1. Tier 1: Essential User Inputs - fields that users must explicitly provide 2. Tier 2: System Inputs with Defaults - fields with reasonable defaults that can be overridden 3. Tier 3: Derived Fields - fields calculated from other fields (private with properties)

class TSATrainingConfig(*, author, bucket, role, region, service_name, pipeline_version, model_class='xgboost', current_date=<factory>, framework_version='2.1.0', py_version='py310', source_dir=None, enable_caching=False, use_secure_pypi=False, max_runtime_seconds=172800, project_root_folder, training_entry_point, training_instance_type='ml.p4d.24xlarge', training_instance_count=1, training_volume_size=50, ca_repository_arn='arn:aws:codeartifact:us-west-2:149122183214:repository/amazon/secure-pypi', output_kms_key=None, volume_kms_key=None, skip_hyperparameters_s3_uri=True, use_amp=True, enable_tf32=True, enable_fp8=False, gradient_accumulation_steps=1, max_grad_norm=1.0, checkpoint_freq=20, job_type=None, hyperparameters=None, **extra_data)

Bases: BasePipelineConfig

Configuration specific to the SageMaker TSA Training Step. This version is streamlined to work with specification-driven architecture. Input/output paths are now provided via step specifications and dependencies.

Fields are organized into three tiers: 1. Tier 1: Essential User Inputs - fields that users must explicitly provide 2. Tier 2: System Inputs with Defaults - fields with reasonable defaults that can be overridden 3. Tier 3: Derived Fields - fields calculated from other fields (private with properties)

training_entry_point: str

training_instance_type: str

training_instance_count: int

training_volume_size: int

framework_version: str

py_version: str

ca_repository_arn: str

output_kms_key: str | None

volume_kms_key: str | None

skip_hyperparameters_s3_uri: bool

use_amp: bool

enable_tf32: bool

enable_fp8: bool

gradient_accumulation_steps: int

max_grad_norm: float

checkpoint_freq: int

job_type: str | None

hyperparameters: ModelHyperparameters | None

model_config: ClassVar[ConfigDict] = {'arbitrary_types_allowed': True, 'extra': 'allow', 'protected_namespaces': (), 'validate_assignment': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

property hyperparameter_file: str

Get hyperparameter file path.

property checkpoint_s3_uri: str

Get S3 URI for model checkpoints.

model_dump(**kwargs)

Override model_dump to include derived properties.

initialize_derived_fields()

Initialize all derived fields once after validation.

classmethod validate_job_type(v)

Validate job_type is one of allowed values.

get_environment_variables()

Get environment variables for the TSA training script.

Returns:

Dictionary mapping environment variable names to values

Return type:

Dict[str, str]

get_public_init_fields()

Override get_public_init_fields to include TSA training-specific fields. Gets a dictionary of public fields suitable for initializing a child config. Includes both base fields (from parent) and TSA training-specific fields.

Returns:

Dictionary of field names to values for child initialization

Return type:

Dict[str, Any]

model_post_init(context, /)

This function is meant to behave like a BaseModel method to initialize private attributes.

It takes context as an argument since that’s what pydantic-core passes when calling it.

Parameters:

• self (BaseModel) – The BaseModel instance.

• context (Any) – The context.