-
Notifications
You must be signed in to change notification settings - Fork 74k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
tf.divide handles "name" argument differently #6741
Comments
So what's happening here is that |
@aselle for comment if this is likely to get fixed .... it seems |
I think I have a fix. The problem is that operator overload div cannot pass name, but I do want to use a class' division dispatch to honor the Python2/Python3 modes. I can do that with a dummy class. I have a fix that I'm testing internally. WIll keep this bug updated. |
@ppwwyyxx, master now has the fix, You can cherry pick it if you need it in an older version, otherwise in the mean-time you can use the workaround. Thanks! |
Great! Thanks a lot! |
Similar problem was found in |
I think they are likely to get fixed if they are filed as separate issues (so they get triaged properly). Good to reference this issue so people can refer back to the commit that fixes things |
👍 I was hesitated to open too many similar issues.. |
Imported from GitHub PR openxla/xla#6741 Added a helper function to add suffix to CloneWithNewOperand function. Rework pr openxla/xla#6289 to not change existing API but introduce an override. Copybara import of the project: -- 764cd152008ff78247706fb37009018b54483ea9 by TJ <tjx@nvidia.com>: rework pr openxla/xla#6289 to not change existing API but introduce an override Merging this change closes #6741 PiperOrigin-RevId: 579757627
If installed from binary pip package, provide:
https://ci.tensorflow.org/view/Nightly/job/nightly-matrix-linux-gpu/TF_BUILD_IS_OPT=OPT,TF_BUILD_IS_PIP=PIP,TF_BUILD_PYTHON_VERSION=PYTHON2,label=gpu-linux/lastSuccessfulBuild/artifact/pip_test/whl/tensorflow_gpu-0.12.1-cp27-none-linux_x86_64.whl
python -c "import tensorflow; print(tensorflow.__version__)"
.0.12.head
If possible, provide a minimal reproducible example (We usually don't have time to read hundreds of lines of your code)
Normally one would expect (like almost every other existing function in tf) that with the name='x' option, the output op will be named 'x:0'.
But instead it was named "x/truediv:0".
The text was updated successfully, but these errors were encountered: