load_dataclass_npz

load_dataclass_npz#

Class method.

do_dpc.utils.serialization.load_dataclass_npz(cls, filename)[source]#

Loads a dataclass object from a .npz file, reconstructing NumPy arrays, floats, and integers.

Parameters:

  • cls (Type[T]) – The class type of the dataclass to reconstruct.

  • filename (str) – The .npz file path from which the object should be loaded.

Returns:

The reconstructed dataclass object.

Return type:

T

Raises:

  • FileNotFoundError – If the specified file does not exist.

  • ValueError – If the loaded data contains unsupported types or is incompatible with the dataclass.

Note

  • Converts 0D NumPy arrays (e.g., np.array(0.1)) back to Python scalars (float or int).

  • Only fields that exist in the dataclass definition are reconstructed.

  • If a field is missing in the .npz file, it will raise a TypeError (from dataclass instantiation).

Contents