dataeval.config.set_seed¶

dataeval.config.set_seed(seed, all_generators=False, deterministic=False)¶

Set the seed for use by classes that allow for a random state or seed.

Parameters:¶

seed : int or None¶

The seed to use. When None, clears the DataEval seed and resets all generators (NumPy, PyTorch) and deterministic settings regardless of other parameters.

all_generators : bool, default False¶

Whether to set the seed for all generators, including NumPy and PyTorch.

deterministic : bool, default False¶

Whether to force PyTorch to use deterministic algorithms. When True, calls torch.use_deterministic_algorithms(True), which ensures reproducible results but may reduce performance.