Fixes #126949
torch.tensor_split documents its second parameter as indices_or_sections (matching NumPy), but the underlying overloaded functions use different keyword argument names:
torch.tensor_split(x, indices_or_sections=3) # TypeError
torch.tensor_split(x, sections=3) # works
torch.tensor_split(x, indices=(1, 6)) # worksThis adds a note to the docstring documenting the actual keyword argument names (sections for int, indices for list/tuple) so users can use them correctly.
Pull Request resolved: #182075
Approved by: https://github.com/zou3519
SOCIAL SHARE CARD GENERATOR