Thank you for reaching out with your question about tfplus/kv_variable compatibility with TensorFlow 2 (TF2).
In brief, while TFPlus is developed with support for TF2 (specifically, TensorFlow 2.13.0), it currently utilizes TensorFlow 1 (TF1) compatibility interfaces for certain functionalities, including training. This means our full support for TF2 features, especially the default Eager Execution mode, is still in progress.
For training, TFPlus currently supports CPU-only environments and has been tested on Linux platforms. We are planning to expand this support in future versions.
For detailed information and future updates, please refer to our documentation:
Thank you for reaching out with your question about
tfplus/kv_variable
compatibility with TensorFlow 2 (TF2).In brief, while
TFPlus
is developed with support for TF2 (specifically, TensorFlow 2.13.0), it currently utilizes TensorFlow 1 (TF1) compatibility interfaces for certain functionalities, including training. This means our full support for TF2 features, especially the default Eager Execution mode, is still in progress.For training,
TFPlus
currently supports CPU-only environments and has been tested on Linux platforms. We are planning to expand this support in future versions.For detailed information and future updates, please refer to our documentation:
We appreciate your interest in
TFPlus
and are working towards full TF2 feature support.