Hey,
I had a quick question regarding the TensorFlow 2.20.0 release. The release notes mention that:
tf.lite will be deprecated, in favor of the new repo https://github.com/google-ai-edge/LiteRT.
Does this mean that TensorFlow 2.20.0 already includes support for LiteRT internally, or is LiteRT integration planned for a future TensorFlow release?
If LiteRT is not yet supported under tensorflow/lite, does LiteRT still include all the previous TensorFlow Lite headers? I’m updating some code that used tensorflow/lite, but I can’t find a few headers in the LiteRT repo, that existed in TensorFlow Lite under tensorflow/lite.
Thank you
Hi, @apoorv1009
I apologize for the delay in my response, tf.lite is being replaced by LiteRT. The tf.lite module will be deprecated with development for on-device inference moving to a new, independent repository: LiteRT. The new APIs are available in Kotlin and C++. This code base will decouple from the TensorFlow repository and tf.lite will be removed from future TensorFlow Python packages, so we encourage migration of projects to LiteRT to receive the latest updates. please refer this blog
The TensorFlow 2.20.0 release acts as the definitive point of architectural decoupling. The release notes confirm that the legacy tf.lite module is officially deprecated and its duplicated source code has been removed from the TensorFlow repository.
LiteRT is now a separate, external dependency and this separation is mandatory moving forward. You must install LiteRT as an independent package. For Python this involves using the ai-edge-litert PIP package and for Android/C++ it involves configuring dependencies to point to the new LiteRT Maven packages and also please refer this Migrate to LiteRT from TensorFlow Lite documentation
Please refer this Tools for building with C and C++ section from this offcial documentation to build your apps, I’ve answered your follow up question in the Github issue comment
Please let us know if you have any further questions. Thank you for your cooperation and patience.