This repository provided a pre-release of TensorFlow and TensorFlow Addons tailored for macOS 11+ with native hardware acceleration via Apple’s ML Compute, supporting both Apple Silicon (M1) and Intel Macs. It shipped ready-made Python 3.8 wheels and install scripts so developers could quickly get an accelerated stack running without building from source. As TensorFlow added a Metal PluggableDevice path, the project directed users toward using Apple’s tensorflow-metal to get GPU acceleration on Mac directly through Metal. The README and releases emphasized this was a macOS-optimized distribution, not an upstream fork of TensorFlow, and it cataloged common installation notes and issues encountered by early Apple Silicon users. In practice, the repo served as a bridge during the transition to native Mac acceleration, making modern TensorFlow viable on Apple hardware while upstream support matured.
Features
- macOS 11+ builds of TensorFlow and Addons with ML Compute acceleration
- Support for Apple Silicon and Intel Macs in one distribution
- Prebuilt Python wheels and one-step install scripts
- Pathway to Metal GPU acceleration via tensorflow-metal
- Practical troubleshooting via issues for early M1 adoption
- Transitional bridge until upstream Mac acceleration stabilized
Project Samples
