noether.data.datasets.cfd.drivaernet.zarr_dataset¶
Zarr-backed DrivAerNet(++) dataset with chunk-based subsampling.
Reads from a converted Zarr store (local path or fsspec URL such as
oci://bucket@namespace/zarr_store). The store is self-contained: the split files
({train,test,val}_design_ids.txt) and blacklists (blacklist.txt,
blacklist2.txt) live next to manifest.json and are read through fsspec, so the
same split/blacklist semantics as DrivAerNetDataset apply.
Attributes¶
Classes¶
Config for the Zarr-backed DrivAerNet dataset ( |
|
DrivAerNet(++) dataset reading from a converted Zarr store. |
Module Contents¶
- noether.data.datasets.cfd.drivaernet.zarr_dataset.logger¶
- class noether.data.datasets.cfd.drivaernet.zarr_dataset.ZarrDrivAerNetDatasetConfig(/, **data)¶
Bases:
noether.data.datasets.cfd.zarr_aero_dataset.ZarrAeroDatasetConfigConfig for the Zarr-backed DrivAerNet dataset (
root= converted Zarr store).Create a new model by parsing and validating input data from keyword arguments.
Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.
self is explicitly positional-only to allow self as a field name.
- Parameters:
data (Any)
- class noether.data.datasets.cfd.drivaernet.zarr_dataset.ZarrDrivAerNetDataset(dataset_config)¶
Bases:
noether.data.datasets.cfd.zarr_aero_dataset.ZarrAeroDatasetDrivAerNet(++) dataset reading from a converted Zarr store.
Mirrors
DrivAerNetDataset’s split handling (split id files, blacklists, category filtering and"train[0:100]"subset notation) while subsampling via chunked Zarr reads. When the config’snum_*counts are set, the pipeline’s anchor/geometry sampling becomes inert (seegeometry_position_from_dataset).- Parameters:
dataset_config (ZarrDrivAerNetDatasetConfig) – DrivAerNet Zarr config;
rootis the Zarr store root (local or fsspec URL) and thenum_*/sampling_seed/read_concurrencyfields drive subsampling.
- design_ids¶