Global configuration for DataEval runtime settings.
This Pydantic model backs the global configuration state and provides
validation on assignment. Users typically interact with the functional
API (get_*, set_*, use_*) rather than this class directly.
-
device
Default PyTorch device for computations.
- Type:
torch.device or None, default None
-
batch_size
Default batch size for data processing.
- Type:
int or None, default None
-
max_processes
Maximum number of worker processes for parallel tasks.
- Type:
int or None, default None
-
seed
Random seed for reproducibility.
- Type:
int or None, default None
-
classmethod validate_batch_size(v)
Validate that batch_size is greater than zero.
-
classmethod validate_max_processes(v)
Validate that max_processes is not zero.