r""" Args: inp_args: list of input datas. inp_dict: dict of named input datas. Returns: a dict {

Set up thread-local state and return a contextmanager for managing it. This function is a convenience wrapper. It defines a flag, environment

get logical index; valonly must be false

! * \brief whether var value could be statically inferred */

r"""Returns the value of this :class:`~.Tensor` as a standard Python :class:`numbers.Number`. This only works for tensors with one element. Fo

! * \brief name of this data type */

share the data to the tensor Args: data: the data will shared to the tensor, it should be a numpy.ndarra

r""" Returns an iterable for the :class:`~.Tensor` of the module. :param recursive: If ``True``, returns all :class:`~.Tensor` within

r"""Returns a new :class:`~.Tensor`, detached from the current graph.

r"""Gets default computing node. It returns the value set by :func:`~.set_default_device`.

Check if finite number. \param arg number to check \retval true if neither infinity nor NaN \retval false else template<typename T> typename enable<bo

! * \brief get a symbolvar whose value is tensor shape * * \param inp used to determine the computing graph, which can be any symbolvar * bel

r"""Get the positional arguments of the operation corresponding to this Expr.

r"""C++ graph version of :func:`~.set_execution_strategy`. Used to inplacely modify dumped graph's fast-run strategy. Args: dest_vars

copy the data to the tensor Args: data: the data to copy to tensor, it should be list, numpy.ndarraya or

get the physical locator that created this comp node

r"""Set the VarNode use symbolic shape or not, return the last status. Please set to True and must recover after dump if want to change the input