Custom Gradient Computation not working in TF 2.14 #62053
Labels
comp:ops
OPs related issues
regression issue
To spot regression issues in latest version
stat:awaiting tensorflower
Status - Awaiting response from tensorflower
TF2.14
For issues related to Tensorflow 2.14.x
type:bug
Bug
Issue type
Bug
Have you reproduced the bug with TensorFlow Nightly?
No
Source
binary
TensorFlow version
2.14.0
Custom code
Yes
OS platform and distribution
Linux
Mobile device
No response
Python version
3.10.12
Bazel version
No response
GCC/compiler version
No response
CUDA/cuDNN version
No response
GPU model and memory
No response
Current behavior?
I am using W&B's Keras callback
WandbCallback
. This callback has a feature to log gradients of each layer at every step. This feature works fine till TF 2.13.0 but is erroring out in TF 2.14.0.This piece of code works fine in Tf 2.13.0 but errors out in TF 2.14.0:
I investigated further and was able to narrow it down to the gradient logging logic which again works fine for 2.13.0 but not for 2.14.0.
I think this has to do with the breaking changes with
tf.Tensor
.The piece of code below is the gradient logging logic which errors out in the latest version.
Standalone code to reproduce the issue
Relevant log output
The text was updated successfully, but these errors were encountered: