mlx icon indicating copy to clipboard operation
mlx copied to clipboard

fix: conv_general differences between gpu, cpu

Open aturker1 opened this issue 8 months ago • 3 comments

Proposed changes

Fixes inconsistencies in conv_general results between GPU and CPU implementations by properly handling asymmetric padding in the CPU backend. Implements separate padding_lo and padding_hi parameters throughout the convolution code. Addresses issue #2038.

Checklist

Put an x in the boxes that apply.

  • [x] I have read the CONTRIBUTING document
  • [x] I have run pre-commit run --all-files to format my code / installed pre-commit prior to committing changes
  • [ ] I have added tests that prove my fix is effective or that my feature works
  • [x] I have updated the necessary documentation (if needed)

aturker1 avatar Apr 12 '25 14:04 aturker1

@aturker1 the tests are failing on this PR. Are you planning to come back to it?

awni avatar Apr 17 '25 23:04 awni

@awn Ah yes, I am going to fix it thid weekend.

aturker1 avatar Apr 18 '25 05:04 aturker1

Would you mind adding a test for this which exposed the issue in the first place?

awni avatar May 01 '25 18:05 awni