torch.nn.utils.rnn 的源代码

```html

import warnings
from typing import Iterable, List, NamedTuple, Tuple, Union

import torch
from torch import Tensor
from ... import _VF
from ..._jit_internal import Optional


__all__ = ['PackedSequence', 'invert_permutation', 'pack_padded_sequence', 'pad_packed_sequence', 'pad_sequence',
           'unpad_sequence', 'pack_sequence', 'unpack_sequence']


class PackedSequence_(NamedTuple):
    data: torch.Tensor
    batch_sizes: torch.Tensor
    sorted_indices: Optional[torch.Tensor]
    unsorted_indices: Optional[torch.Tensor]


def bind(optional, fn):
    if optional is None:
        return None
    return fn(optional)


class PackedSequence(PackedSequence_):
    r"""Holds the data and list of :attr:`batch_sizes` of a packed sequence.

    All RNN modules accept packed sequences as inputs.

    Note:
        Instances of this class should never be created manually. They are meant
        to be instantiated by functions like :func:`pack_padded_sequence`.

        Batch sizes represent the number elements at each sequence step in
        the batch, not the varying sequence lengths passed to
        :func:`pack_padded_sequence`.  For instance, given data ``abc`` and ``x``
        the :class:`PackedSequence` would contain data ``axbc`` with
        ``batch_sizes=[2,1,1]``.

    Attributes:
        data (Tensor): Tensor containing packed sequence
        batch_sizes (Tensor): Tensor of integers holding
            information about the batch size at each sequence step
        sorted_indices (Tensor, optional): Tensor of integers holding how this
            :class:`PackedSequence` is constructed from sequences.
        unsorted_indices (Tensor, optional): Tensor of integers holding how this
            to recover the original sequences with correct order.

    .. note::
        :attr:`data` can be on arbitrary device and of arbitrary dtype.
        :attr:`sorted_indices` and :attr:`unsorted_indices` must be ``torch.int64``
        tensors on the same device as :attr:`data`.

        However, :attr:`batch_sizes` should always be a CPU ``torch.int64`` tensor.

        This invariant is maintained throughout :class:`PackedSequence` class,
        and all functions that construct a :class:`PackedSequence` in PyTorch
        (i.e., they only pass in tensors conforming to this constraint).

    """

    def __new__(cls, data, batch_sizes=None, sorted_indices=None, unsorted_indices=None):
        return super().__new__(
            cls,
            *_packed_sequence_init_args(data, batch_sizes, sorted_indices,
                                        unsorted_indices))

    # NOTE [ device and dtype of a PackedSequence ]
    #
    # See the note above in doc string (starting with ":attr:`data` can be on
    # arbitrary device...").
    def pin_memory(self):
        # Why not convert `batch_sizes`?
        # See NOTE [ device and dtype of a PackedSequence ]
        return type(self)(self.data.pin_memory(), self.batch_sizes,
                          bind(self.sorted_indices, lambda t: t.pin_memory()),
                          bind(self.unsorted_indices, lambda t: t.pin_memory()))

    def cuda(self, *args, **kwargs):
        # Tests to see if 'cuda' should be added to kwargs
        ex = torch.tensor((), dtype=self.data.dtype, device=self.data.device).to(*args, **kwargs)
        if ex.is_cuda:
            return self.to(*args, **kwargs)
        return self.to(*args, device='cuda', **kwargs)

    def cpu(self, *args, **kwargs):

        ex = torch.tensor((), dtype=self.data.dtype, device=self.data.device).to(*args, **kwargs)
        if ex.device.type == 'cpu':
            return self.to(*args, **kwargs)
        return self.to(*args, device='cpu', **kwargs)

    def double(self):
        return self.to(dtype=torch.double)

    def <span class