Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Disable LTC and update PyTorch #1472

Merged
merged 2 commits into from
Oct 11, 2022
Merged

Disable LTC and update PyTorch #1472

merged 2 commits into from
Oct 11, 2022

Conversation

ashay
Copy link
Collaborator

@ashay ashay commented Oct 10, 2022

When built using PyTorch's master branch, the LTC code has been failing
to build for a few days. As a result, the PyTorch version referenced by
Torch-MLIR is stalled to the one from October 4th.

This PR contains two patches: the first disables LTC and the second
advances the PyTorch version.

When built using PyTorch's master branch, the LTC code has been failing
to build for a few days.  As a result, the PyTorch version referenced by
Torch-MLIR is stalled to the one from October 4th.

In an effort to advance to PyTorch version, this patch disables LTC, and
a subsequent patch will advance the PyTorch version.
@ashay
Copy link
Collaborator Author

ashay commented Oct 10, 2022

@pashu123 It seems like after updating the PyTorch version, one of your tests fails because the oracle is slightly different from the computed result. Do you have a hunch as to why this might be the case?

> python -m e2e_testing.main --config=eager_mode -v -s --filter=UpSampleNearest2dDynamicFactor_basic

Compiling UpSampleNearest2dDynamicFactor_basic...
Running UpSampleNearest2dDynamicFactor_basic...
FAIL - "UpSampleNearest2dDynamicFactor_basic"

Unexpected outcome summary:

****** Failed tests - 1 tests
    FAIL - "UpSampleNearest2dDynamicFactor_basic"
        @ trace item #0 - call to "forward"
        @ output of call to "forward"
        ERROR: value (Tensor with shape=[2, 3, 4, 9], dtype=torch.float32, min=+0.02233, max=+0.9152, mean=+0.5012) is not close to golden value (Tensor with shape=[2, 3, 4, 9], dtype=torch.float32, min=+0.02233, max=+0.9152, mean=+0.4657)


Summary:
    Failed: 1

@powderluv
Copy link
Collaborator

if required we can xfail to roll it forward with an issue tracker

Also disables the `UpSampleNearest2dDynamicFactor_basic` e2e test, since
the (PyTorch) oracle differs from the computed value for both the
refbackend and the eager_mode backends.
@ashay
Copy link
Collaborator Author

ashay commented Oct 11, 2022

Disabled the test for now, and noted it in this issue: #1473.

@ashay ashay requested a review from powderluv October 11, 2022 02:40
@ashay ashay marked this pull request as ready for review October 11, 2022 02:40
Copy link
Collaborator

@powderluv powderluv left a comment

Choose a reason for hiding this comment

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

@antoniojkim We want to try and get tonights build successful. Happy to enable LTC again once it is fixed.

@powderluv powderluv requested a review from antoniojkim October 11, 2022 02:50
@powderluv
Copy link
Collaborator

@pashu123 please re-enable the tests once it is fixed.

@ashay ashay merged commit aefbf65 into llvm:main Oct 11, 2022
@ashay ashay deleted the ashay/disable-ltc-and-update-pytorch branch October 11, 2022 04:05
@pashu123
Copy link
Member

Thanks, @ashay for filing it as an issue. I will take a look at it.

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.

3 participants