Skip to content

Conversation

@hugolatendresse
Copy link
Contributor

We used to have an assertion that when using torch.split, the dimension to be split must be an multiple of the split size. This is not required in pytorch: the last tensors will simply be smaller. This PR allows that capability.

@hugolatendresse hugolatendresse marked this pull request as ready for review March 20, 2025 16:57
@hugolatendresse
Copy link
Contributor Author

cc: @MasterJH5574 this is ready for your review

Copy link
Contributor

@MasterJH5574 MasterJH5574 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM!

@MasterJH5574 MasterJH5574 merged commit 67e121d into apache:main Mar 24, 2025
11 checks passed
@hugolatendresse hugolatendresse deleted the split_uneven branch April 19, 2025 15:30
ShiboXing pushed a commit to ShiboXing/tvm that referenced this pull request Aug 10, 2025
We used to have an assertion that when using torch.split,
the dimension to be split must be an multiple of the split size.
This is not required in pytorch: the last tensors will simply be smaller.
This PR allows that capability.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants