aten::var implementation and aten::roll complex support #1186
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
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #1186 +/- ##
==========================================
+ Coverage 78.56% 78.63% +0.06%
==========================================
Files 118 118
Lines 15380 15445 +65
Branches 2408 2424 +16
==========================================
+ Hits 12084 12145 +61
- Misses 2900 2902 +2
- Partials 396 398 +2 ☔ View full report in Codecov by Sentry. |
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
justinchuby
reviewed
Nov 27, 2023
justinchuby
added
the
module: torchlib
|
@microsoft-github-policy-service agree |
justinchuby
reviewed
Nov 29, 2023
justinchuby
reviewed
Nov 29, 2023
justinchuby
reviewed
Nov 29, 2023
Comment on lines
6948
to
6954
| for i in range(len(shifts)): # pylint: disable=consider-using-enumerate | ||
| shift = op.Gather(shifts, i, axis=0) | ||
| dim = dims[i] | ||
| self_real = _aten_roll_shift_and_dim_onnx(self_real, shift, dim) | ||
| self_imag = _aten_roll_shift_and_dim_onnx(self_imag, shift, dim) | ||
|
|
||
| result = op.Concat(self_real, self_imag, axis=-1) |
There was a problem hiding this comment.
Since shifts is a tensor, it will not have len defined on it. Iterate on dims?
for i, dim in enumerate(dims):
...
titaiwangms
reviewed
Nov 29, 2023
| ) -> TReal: | ||
| """var(Tensor self, int[1]? dim, bool unbiased=True, bool keepdim=False) -> Tensor""" | ||
|
|
||
| if isinstance(dim, int): |
There was a problem hiding this comment.
dim could be something else? Should we annotate it on function sig?
There was a problem hiding this comment.
I can follow up on this after #1192. I believe this code is just copied from var_mean
justinchuby
reviewed
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
justinchuby
approved these changes
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
justinchuby
reviewed
Nov 30, 2023
|
Will fix the dtype error in a separate PR. |
justinchuby
added a commit
that referenced
this pull request
Nov 30, 2023
Follow up of #1186 to fix the dtype mismatch in Sub in the `var`/`var.correction` implementation.
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 mentioned on #1173, I'm trying to add aten::var and aten::roll (complex support) in order to export one model from PyTorch to ONNX. The model uses fft functions, which requires opset 18 and torch dynamo usage.
fixes #1175 fixes #1174