Charles Volzka

Any update on this one or is investigation ongoing?

Any ETA on when this one will be resolved?

I tried the fix. It appears to work for CPU compiled models but when I tried for NNPA, I still got the incorrect output shape (final output dim was `-1`...

Yes it has. Thanks for reminding me!

I tried a build using https://github.com/onnx/onnx-mlir/pull/1777/commits/67fc796fde89c9739c03a32e3243102a13a6f781 From the original issue, the following models compiled successful: ``` yolov3-10 yolov3-12 candy-9 mosaic-9 pointilism-9 rain-princess-9 udnie-9 ``` The following still failed: ``` bvlcalexnet-6...

Any updates on the remaining two leaks?

@negiyas I did a test run last night and confirmed the models now compile. Thanks!

@AlexandreEichenberger > I don't understand this. Maybe you can just say "Run on NNPA as a combination of Relu, multiply, and sub" Would be a different comment. That would be...

@liqunfu This has a milestone for 1.16. Is it still valid and should it be considered a blocker to cutting a 1.16 branch?

I see this has a 1.16 Milestone but hasn't had an update since December. Should this be considered a blocker to cutting a 1.16 release branch?