frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

Documentation Fix: torch.tensor.scatter_ docs (#157929)

Browse Source 
updated torch.tensor.scatter_ docs to reflect proper broadcasting behavior

Fixes #157419

Pull Request resolved: https://github.com/pytorch/pytorch/pull/157929
Approved by: https://github.com/albanD
This commit is contained in:
morrison-turnansky
2025-07-10 19:22:49 +00:00
committed by PyTorch MergeBot
parent da4e7c77a1
commit 8c5b070d1f
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
torch/_tensor_docs.py
View File

@ -4394,11 +4394,12 @@ For a 3-D tensor, :attr:`self` is updated as::
This is the reverse operation of the manner described in :meth:`~Tensor.gather`.
:attr:`self`, :attr:`index` and :attr:`src` (if it is a Tensor) should all have
the same number of dimensions. It is also required that
It is also required that
``index.size(d) <= src.size(d)`` for all dimensions ``d``, and that
``index.size(d) <= self.size(d)`` for all dimensions ``d != dim``.
Note that ``index`` and ``src`` do not broadcast.
Note that ``input`` and ``index`` do not broadcast against each other for NPUs,
so when running on NPUs, :attr:`input` and :attr:`index` must have the same number of dimensions.
Standard broadcasting occurs in all other cases.
Moreover, as for :meth:`~Tensor.gather`, the values of :attr:`index` must be
between ``0`` and ``self.size(dim) - 1`` inclusive.
@ -4525,6 +4526,8 @@ For a 3-D tensor, :attr:`self` is updated as::
dimensions. It is also required that ``index.size(d) <= src.size(d)`` for all
dimensions ``d``, and that ``index.size(d) <= self.size(d)`` for all dimensions
``d != dim``. Note that ``index`` and ``src`` do not broadcast.
When :attr:`index` is empty, we always return the original tensor
without further error checking.
Note:
{forward_reproducibility_note}