Implement aten.constant_pad_nd.default with constant mode

pytorch / executorch

On-device AI across mobile, embedded and edge for PyTorch

https://pytorch.org/executorch/

Other

1.41k stars 230 forks source link

Implement aten.constant_pad_nd.default with constant mode #3975

Closed huiyujie closed 1 week ago

huiyujie commented 1 week ago

Summary: Implement aten.constant_pad_nd.default, needed in OCR full model.

- func: constant_pad_nd(Tensor self, SymInt[] pad, Scalar value=0) -> Tensor

torch.nn.functional.pad is compiled to this op when mode="constant".

pad has the form: (left, right): pad last one dimension (left, right, top, bottom): pad last 2 dimensions (left, right, top, bottom, front, back): pad last 3 dimensions

Differential Revision: D58441509

pytorch-bot[bot] commented 1 week ago

:link: Helpful Links

:test_tube: See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/3975

:page_facing_up: Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

:white_check_mark: No Failures

As of commit 5f040c1e797d19b59a36e33d30652c9195ce495e with merge base 172574a6be5910a4609e4ed1bef2b6b8475ddb3d (): :green_heart: Looks good so far! There are no failures yet. :green_heart: