save_dict_to_hdf5

save_dict_to_hdf5#

ptyrad.save.save_dict_to_hdf5(d, output_path, none_sentinel='__NONE__', **kwargs)[source]#

Save a nested Python dictionary to an HDF5 file.

Supports common Python, NumPy, and PyTorch types. Non-HDF5-compatible types (e.g., list of tuples, None, etc.) are automatically converted to HDF5-friendly formats.

Note that integer key (e.g. like in optimizer state dict) are coerced to string for HDF5 format.

Parameters:

  • d (Dict[str, Any]) – The nested dictionary to save.

  • output_path (str) – The file path to save the HDF5 output to.

  • none_sentinel (str, optional) – String used to represent None in HDF5. Defaults to “__NONE__”.

  • **kwargs – Additional keyword arguments to pass to h5py.File() or create_dataset(). This can include compression settings like compression=”gzip”, etc.

Returns:

None

Return type:

None

Contents