Thibaut Goetghebuer-Planchon
Thibaut Goetghebuer-Planchon
This is solved by https://github.com/tensorflow/tensorflow/commit/a3c37bce4f04befa1b15edf5cce108887a36aeb5 now. Closing.
Thank you very much for your answers. I did my tests with the nightly build. Note that if we lower the ADD_N to a series of ADD, it would be...
Hi @abattery @karimnosseir I was wondering if there was any eventual news on the issue? Should we aim to lower the `tf.math.add_n` operator to a series of add similar to...
For now we can just replace `tf.math.add_n` with `tf.keras.layers.Add` as temporary workaround but we're mainly interested by resolving the incoherence between the two operators to make it easier to support...
Hi, yes `tf.keras.layers.Add` and `tf.math.add_n` still behave differently and can produce different quantized results.
Hi, The issue is still present in the latest 2.13 version of TF.
This is solved by https://github.com/tensorflow/tensorflow/commit/200be965419f2548bfbdb16a826d2547841479ec now. Closing.
Sorry for the long delay. I created a new release with a CMake minimum version of 3.10.
Thanks for the report. It seems to be an issue with g++ 9 as I couldn't reproduce the issue with g++-11 to g++14 (and clang and MSVC in the CI...
Thanks for the report. Is it built with C++17? As 1.4 dropped support for C++11 and moved to C++17. From the errors about the missing `std::launder`, `std::clamp` and other errors,...