output_padding argument in Conv1DTranspose

[olive380]

output_padding argument in Conv1DTranspose

[sushreebarsa]

@olive380 If you don't require any padding for the output, simply omit the output_padding argument altogether. The default behavior in TensorFlow 2.16 for Conv1DTranspose is to not add any padding, which aligns with your case of using output_padding=0. Also please refer to the Keras documentation for the Conv1DTranspose layer specifically. This might provide additional details and examples regarding the output_padding argument.

Thank you!

[olive380]

@sushreebarsa Thank you for the answer, but your proposal does not solve the issue since the behavior of TensorFlow is totally different if output_padding is omitted or set to zero in all versions of the code up to 2.15. If you are not aware of this, please have a look at the deconv_output_length method defined in /tensorflow/python/keras/utils/conv_utils.py available here. So, please fix the bug in version 2.16 to allow set output_padding=0 in Conv1DTraspose and Conv2DTranspose.

Thank you !

[tilakrayal]

@olive380,
Thank you for the issue. By default Tensorflow v2.16.1 contains keras3.0 version which might be the issue, as you mentioned this feature is available in Tensorflow v2.15 which contains tf-keras. If the same feature needs to be implemented in the keras3.0, could you please raise the request in keras-team/Keras repo for the quick response. Thank you!

[github-actions]

This issue is stale because it has been open for 7 days with no activity. It will be closed if no further activity occurs. Thank you.

[github-actions]

This issue was closed because it has been inactive for 7 days since being marked as stale. Please reopen if you'd like to work on this further.

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No

[olive380]

Thanks, an issue was opened for Keras 3.0 here