[Mlir-commits] [mlir] [mlir][linalg] Use inferConvolutionDims for generic convolution downscaling (PR #180586)

[Han-Chung Wang]

https://github.com/hanhanW commented:

Overall looks good to me, just some nits about implementation and comment.

I agree that we can remove the second RUN now, because the implementation is using abstarctions; we have both named and generic ops form. The future contribution that touches these code should improve the test structure if they break something. This aligns with my mental model of testing.

https://github.com/llvm/llvm-project/pull/180586