sushreebarsa

@adamp87 One workaround could be to quantize the entire model, then fine-tune specific layers with FP32 weights and activation. This approach can be less efficient and might not perfectly address...

@chmendoza You're right, there have been reports already of some TensorFlow API GitHub source code links not working correctly. It's a known issue. Tensorflow has been released at 2.15.0.post1 ([see...

@chmendoza Yes, there are a few other apis are also there which are having the same issue as you mentioned. We are expecting it to be fixed in the next...

@michaelpoluektov Could you please avoid using masks with StridedSlice when strides are all 1. Instead, explicitly use the Slice operator for better performance. Thank you!

@Varfalamei The error message "Error reading protobuf spec. validator error: Padding type for the pooling layer 'PoolingLayerBuilder (MEAN)_57' is not set." indicates that you haven't specified the padding behavior for...

@Caarvaa Verify the exact TensorFlow version required for DeepLabCut 2.2.2. If necessary, install the compatible version using conda install tensorflow== or pip install tensorflow==. Kindly consider using virtual environments that...

@juancresc If you are building tfgo from source, make sure you've configured it for the ARM64 architecture (Apple M1) using the `--config=macos_arm64 `flag. If you are using a pre-built binary,...

@juancresc Could you please confirm if the issue has been resolved? Thank you!

@MoritzMSP Could you consider to calculate shapes and perform tf.concat and tf.transpose outside the loop. Pre-compute Shapes and Operations could help in this case. Thank you!

@MoritzMSP Thank you for the update! @sachinprasadhs Please find the replicated [gist](https://colab.research.google.com/gist/sushreebarsa/094dc8a7cf598f338c1ca7693e2a706d/62861.ipynb) of the error reported. Thank you!