Add ops for HF transformers #2217
Open
+34
−8
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Hi @kshitij12345 , @IvanYashchuk could you take a look of the PR |
kshitij12345
approved these changes
Jun 26, 2025
|
Hi @IvanYashchuk @mruberry , could you take a look at the PR |
crcrpar
reviewed
Jul 2, 2025
Co-authored-by: Masaki Kozuki <[email protected]>
IvanYashchuk
approved these changes
Jul 7, 2025
t-vi
reviewed
Jul 7, 2025
| @@ -1177,7 +1179,7 @@ def movedim(a: TensorLike, /, source: int | Sequence[int], destination: int | Se | |||
| return clang.movedim(a, source, destination) | |||
|
|
|||
|
|
|||
| @torchsymbol(torch.nn.functional.pad) | |||
| @torchsymbol(torch.nn.functional.pad, torch._C._nn.pad, id="torch.nn.functional.pad", is_method=False) | |||
There was a problem hiding this comment.
Why are we calling torch._C._nn.pad directly?
Isn't is_method=False the default?
There was a problem hiding this comment.
this is for ThunderFx path, it appears in the dynamo graph as torch._C._nn.pad
There was a problem hiding this comment.
I can remove the is_method=False, keep it here explicit just because I saw a lot of other ops do the same
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As identified in the split reasons investigated in #2169 (https://docs.google.com/spreadsheets/d/1BKK-1T4UqZHaE8x39gIRFpoCuDmMFC3Z3rn_mBWUStA/edit?usp=sharing), this PR addresses several of them to help reduce the number of segments:
Adds support for
torch._C._nn.pad, which is functionally equivalent totorch.nn.functional.pad.(Reference: PyTorch source),
Registers some operators that are aliases of already-registered ones.
Introduces checks for einops operators in the splitter.
(These operators are registered here)
Fixes #2169.