Optional tag like keepdim may have been removed in several funcs

[ILCSFNO]

📚 The doc issue

Seen from #146156 and its PR pull#146485, maybe some other funcs also have optional tags removed:

What I met is torch.any(), but suggest fix this by change the description of keepdim and any other similar arguments, for that there may be more funcs that have similar problem.

Note that there may have some funcs that has keepdim or any other similar arguments not optional. So change the description of them may have risk!

pytorch/torch/_torch_docs.py

Lines 852 to 880 in d94cc0e

	.. function:: any(input, dim, keepdim=False, *, out=None) -> Tensor
	:noindex:
	For each row of :attr:`input` in the given dimension :attr:`dim`,
	returns `True` if any element in the row evaluate to `True` and `False` otherwise.
	{keepdim_details}
	Args:
	{input}
	{dim}
	{keepdim}
	Keyword args:
	{out}
	Example::
	>>> a = torch.randn(4, 2) < 0
	>>> a
	tensor([[ True, True],
	[False, True],
	[ True, True],
	[False, False]])
	>>> torch.any(a, 1)
	tensor([ True, True, True, False])
	>>> torch.any(a, 0)
	tensor([True, True])
	""".format(**multi_dim_common),

While keepdim is showed here:

pytorch/torch/_torch_docs.py

Lines 55 to 61 in d94cc0e

	multi_dim_common = merge_dicts(
	reduceops_common_args,
	parse_kwargs(
	"""
	dim (int or tuple of ints): the dimension or dimensions to reduce.
	"""
	),

pytorch/torch/_torch_docs.py

Lines 43 to 52 in d94cc0e

	reduceops_common_args = merge_dicts(
	common_args,
	parse_kwargs(
	"""
	dtype (:class:`torch.dtype`, optional): the desired data type of returned tensor.
	If specified, the input tensor is casted to :attr:`dtype` before the operation
	is performed. This is useful for preventing data type overflows. Default: None.
	keepdim (bool): whether the output tensor has :attr:`dim` retained or not.
	"""
	),

Some similar arguments may include dim, etc.

Suggest to solve it together with #146156

Suggest a potential alternative/fix

I suggest any one below:

• Change the description of keepdim and any other similar arguments from origin def (have risk)
• Change the description of keepdim and any other similar arguments one by one (long time to fix)

Thanks for noting!

cc @svekars @sekyondaMeta @AlannaBurke @albanD

[soulitzer]

Thanks for the report, we'd accept PRs to fix missing optional tags
Although it also looks like we aren't very good at tagging with optional in general, e.g. for dim as well.