tensorflow-onnx icon indicating copy to clipboard operation
tensorflow-onnx copied to clipboard

Published 20 hours ago • verified publisher icon onnx

Fix channel num for depthwise conv3d

Open ShuaiShao93 opened this issue 2 years ago • 7 comments

ShuaiShao93 avatar Jul 19 '23 01:07 ShuaiShao93

Thanks for your contribution.

Could you please:

  1. Open a new issue to describe this problem and link it to this PR?
  2. Add a test for this change?

fatcat-z avatar Jul 28 '23 12:07 fatcat-z

Thanks for your contribution.

Could you please:

  1. Open a new issue to describe this problem and link it to this PR?
  2. Add a test for this change?

Sure. I'm not verify familiar with the codebase, can you suggest where to add the unit tests?

ShuaiShao93 avatar Jul 28 '23 16:07 ShuaiShao93

Thanks for your contribution. Could you please:

  1. Open a new issue to describe this problem and link it to this PR?
  2. Add a test for this change?

Sure. I'm not verify familiar with the codebase, can you suggest where to add the unit tests?

You can add it here.

fatcat-z avatar Jul 29 '23 02:07 fatcat-z

Thanks for your contribution. Could you please:

  1. Open a new issue to describe this problem and link it to this PR?
  2. Add a test for this change?

Sure. I'm not verify familiar with the codebase, can you suggest where to add the unit tests?

You can add it here.

Thanks! I added a test but it failed with wrong results. Can this be a bug with the onnx runtime?

ShuaiShao93 avatar Aug 07 '23 19:08 ShuaiShao93

Thanks for your contribution. Could you please:

  1. Open a new issue to describe this problem and link it to this PR?
  2. Add a test for this change?

Sure. I'm not verify familiar with the codebase, can you suggest where to add the unit tests?

You can add it here.

Thanks! I added a test but it failed with wrong results. Can this be a bug with the onnx runtime?

What do you mean by "wrong results"? The result of ORT is different than tf? Is it possible for you to export the onnx graph of this test to check if it is correct?

fatcat-z avatar Aug 21 '23 02:08 fatcat-z

Thanks for your contribution. Could you please:

  1. Open a new issue to describe this problem and link it to this PR?
  2. Add a test for this change?

Sure. I'm not verify familiar with the codebase, can you suggest where to add the unit tests?

You can add it here.

Thanks! I added a test but it failed with wrong results. Can this be a bug with the onnx runtime?

What do you mean by "wrong results"? The result of ORT is different than tf? Is it possible for you to export the onnx graph of this test to check if it is correct?

Ah it doesn't work for all the TF versions. After adding tf version constraints, it works now

ShuaiShao93 avatar Oct 12 '23 22:10 ShuaiShao93

Also filed issue: https://github.com/onnx/tensorflow-onnx/issues/2255

ShuaiShao93 avatar Oct 12 '23 22:10 ShuaiShao93