concepts.dsl.tensor_value.TensorizedPyObjValues#

class TensorizedPyObjValues[source]#

Bases: object

A value object with nested lists of Python objects. Note that this implies that we can not use list as the Python object type when shape is not specified.

Methods

`clone`([deep_copy])	Recursively clone the values.
`expand`(target_size)
`fast_index`(arguments)
`fast_set_index`(arguments, value)
`item`()
`make_empty`(dtype, shape)
`numel`()
`permute`(indices)	Permute the axes in the nested list of values.
`unsqueeze`(dim)

Attributes

ndim

__getitem__(item)[source]#

__init__(dtype, values, shape=None)[source]#

Parameters:

dtype (PyObjValueType)
values (Any | List[Any] | List[List[Any]] | List[List[List[Any]]])
shape (Tuple[int, ...] | None)

__new__(**kwargs)#

clone(deep_copy=False)[source]#

Recursively clone the values.

Parameters:: deep_copy (bool)

expand(target_size)[source]#

Parameters:: target_size (Tuple[int, ...])

fast_index(arguments)[source]#

Parameters:: arguments (Tuple[int, ...])

fast_set_index(arguments, value)[source]#

Parameters:: arguments (Tuple[int, ...])

item()[source]#

classmethod make_empty(dtype, shape)[source]#

Parameters:

dtype (PyObjValueType)
shape (Iterable[int])

numel()[source]#

Return type:: int

permute(indices)[source]#

Permute the axes in the nested list of values.

Parameters:: indices (Tuple[int, ...])

unsqueeze(dim)[source]#

Parameters:: dim (int)

property ndim#