Release 2.17.0

Release 2.17.0 is identical to release 2.16.0.

Note that with TensorFlow 2.17.0, the TF team dropped support for MacOS/Darwin on x86 (Intel) architectures. Because TFC has never been ported to the new (amd) architecture, we are dropping support for MacOS/Darwin starting with this release.

If you're on Linux and Python 3.9–3.12, install the pre-compiled binary by running:

pip install tensorflow-compression-ops==2.17.0

Release 2.16.0

Release 2.16.0 is identical to release 2.15.0.

If you're on Linux or Darwin (Mac OS) and Python 3.9–3.12, install the pre-compiled binary by running:

pip install tensorflow-compression-ops==2.16.0

Release 2.15.0

Release 2.15.0 is identical to release 2.14.1.

However, beginning with TensorFlow 2.15, TensorFlow Compression binary packages will only contain the low-level (C++) ops, and nearly none of the Python code. This is due to an incompatible change in Keras introduced in TensorFlow 2.15, which would require a complete rewrite of our layer and entropy model classes.

If you're on Linux or Darwin (Mac OS) and Python 3.9–3.11, install the pre-compiled binary by running:

pip install tensorflow-compression-ops==2.15.0

Release 2.14.1

Release 2.14.1 is a maintenance release and has no new features.

It requires TensorFlow 2.14 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.9–3.11, install the pre-compiled binary by running:

pip install tensorflow-compression==2.14.1

Release 2.14.0

Release 2.14.0 allows y4m dataset to read C420 colorspace formats that are tagged with a different chroma offset.

Release 2.14.0 drops Python 3.8 support as TensorFlow 2.14 no longer supports Python 3.8.

It requires TensorFlow 2.14 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.9–3.11, install the pre-compiled binary by running:

pip install tensorflow-compression==2.14.0

Release 2.13.0

Release 2.13.0 is a maintenance release and has no new features.

It requires TensorFlow 2.13 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.8–3.11, install the pre-compiled binary by running:

pip install tensorflow-compression==2.13.0

Release 2.12.0

Release 2.12.0 adds an op for stochastic rounding. It also fixes several issues.

This release requires TensorFlow 2.12 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.8–3.11, install the pre-compiled binary by running:

pip install tensorflow-compression==2.12.0

Release 2.11.0

Release 2.11.0 improves the behavior of estimate_tails, and introduces new run-length coding ops that support both Golomb–Rice and Elias gamma codes. It also fixes several issues.

This release requires TensorFlow 2.11 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.7–3.10, install the pre-compiled binary by running:

pip install tensorflow-compression==2.11.0

Release 2.10.0

Release 2.10.0 improves error handling and other elements of the run_length_gamma_encode/decode ops.

This release requires TensorFlow 2.10 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.7–3.10, install the pre-compiled binary by running:

pip install tensorflow-compression==2.10.0

Release 2.9.2

Release 2.9.2 is a maintenance release that fixes problems with the Linux wheel files on certain systems (issue #148), among other minor improvements.

This release requires TensorFlow 2.9 and TensorFlow Probability 0.15.

If you're on Linux or Darwin (Mac OS) and Python 3.7–3.10, install the pre-compiled binary by running:

pip install tensorflow-compression==2.9.2