torch_geometric.data.TensorAttr

class TensorAttr(group_name: Optional[str] = _FieldStatus.UNSET, attr_name: Optional[str] = _FieldStatus.UNSET, index: Optional[Union[Tensor, ndarray, slice, int]] = _FieldStatus.UNSET)

Bases: CastMixin

Defines the attributes of a FeatureStore tensor. It holds all the parameters necessary to uniquely identify a tensor from the FeatureStore.

Note that the order of the attributes is important; this is the order in which attributes must be provided for indexing calls. FeatureStore implementations can define a different ordering by overriding TensorAttr.__init__().

is_set(key: str) → bool

Whether an attribute is set in TensorAttr.

is_fully_specified() → bool

Whether the TensorAttr has no unset fields.

fully_specify() → TensorAttr

Sets all UNSET fields to None.

update(attr: TensorAttr) → TensorAttr

Updates an TensorAttr with set attributes from another TensorAttr.