parameter

parameter

Generic parameter wrapper for Pydantic-based configuration.

Provides Parameter -- a generic dataclass wrapper that integrates with Pydantic v2 core schemas so that configuration values carry metadata (e.g. name) while remaining transparent to serialization and comparison.

Classes:

Name	Description
Parameter	Generic wrapper for a single configuration value.

Parameter(name=None, value=None) dataclass

Bases: Generic[DataT]

Generic wrapper for a single configuration value.

Wraps a primitive or sequence value so it can carry a name, participate in Pydantic validation/serialization, and support rich comparisons against both raw values and other Parameter instances.

Parameters:

Name	Type	Description	Default
name	str | None	Identifier used for logging and lookup (e.g. "holdout").	None
value	DataT | Sequence[DataT] | None	The wrapped configuration value.	None

Example::

>>> param = Parameter[int](name="max_size", value=100)
>>> param.value
100
>>> param == 100
True

Methods:

Name	Description
ser_model	Serialize to the bare value for Pydantic model_dump / model_dump_json.

ser_model()

Serialize to the bare value for Pydantic model_dump / model_dump_json.

Source code in src/nemo_safe_synthesizer/configurator/parameter.py

@model_serializer
def ser_model(self) -> dict[str, DataT] | DataT | Sequence[DataT]:
    """Serialize to the bare value for Pydantic ``model_dump`` / ``model_dump_json``."""
    if hasattr(self, "value"):
        return self.value
    else:
        return self