-
Notifications
You must be signed in to change notification settings - Fork 516
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
Resolve remaining LTC CI failures #1110
Resolve remaining LTC CI failures #1110
Conversation
dfe71ab
to
7613155
Compare
e7af2b0
to
b88264b
Compare
431da1f
to
18f9e35
Compare
Based on the timing and errors, I think this is the root cause: pytorch/pytorch@2bfae07 We should still be good to merge though, since we saw source build succeed earlier prior to this PyTorch update |
Yes, it's failing due to upstream PyTorch changes. #1121 |
54b3c5e
to
b149d13
Compare
5b44d86
to
6802f89
Compare
Looks like the latest failures are due to some changes on the side of Torch-MLIR -- fix coming shortly |
b77b50d
to
6cae84b
Compare
b149d13
to
89e14e8
Compare
6cae84b
to
4401da9
Compare
Ok to land as long as CI is green |
Thanks, can we get this PR stamped as approved then? I'll land this in tonight once CI passes |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🚀
* Replace CHECK_EQ with TORCH_CHECK_EQ * Check value of TORCH_MLIR_USE_INSTALLED_PYTORCH during LTC build * Update LTC XFAIL with NewZerosModule ops * Explicitly blacklist _like ops * Automatically blacklist new_/_like ops * Prune away unused Python dependencies from LTC * Add flag to disable LTC * Autogen dummy _REFERENCE_LAZY_BACKEND library when LTC is disabled * Implement compute_shape_var * Removed Var tests from XFAIL Set * XFAIL tests using _local_scalar_dense or index.Tensor * Add StdDim tests to XFAIL set * Autogen aten::cat
Support RandomNormalLike ONNX operation.
This PR addresses several issues currently faced by CI:
TORCH_MLIR_USE_INSTALLED_PYTORCH=OFF
due to LTC not checking this variable. The build scripts have been updated such that LTC uses the appropriate PyTorch buildnew_
or end with_like
in response to Make all of new_/_like factory functions composite explicit autograd pytorch/pytorch#82238var
op, which is used by a recently added testa. To avoid import issues from the LTC pybind no longer generating, this PR introduces a script to autogen a dummy Python file that allows for a successful import (however throws an error if it's actually used)
cc: @antoniojkim @ke1337