Here you find the C++ installation tar.gz files as mentioned in our guide.
Please see this page on how to install TensorFlow Lite C++.
For convenience reasons, we also placed the Python3 wheels.
Note, the provided wheels are not suitable for C++ programming, only the tar files are.
Both C++ and Python support NEON and XNNPACK.
Select your Python3 version depending on your 64-bit operating system.
| Python3 | Operating system | TF 2.10.0 | TF 2.9.1 | TF 2.8.0 | TF 2.7.0 |
|---|---|---|---|---|---|
| Python 3.6 | Jetson Nano 4.6 | ✔️ | |||
| Python 3.7 | Raspberry Pi Buster | ✔️ | ✔️ | ✔️ | ✔️ |
| Python 3.8 | Ubuntu 20.04 | ✔️ | ✔️ | ✔️ | ✔️ |
| Python 3.9 | Raspberry Pi Bullseye | ✔️ | ✔️ | ✔️ | ✔️ |
| Python 3.10 | Ubuntu 22.04 | ✔️ | ✔️ | ✔️ | ✔️ |
Find TensorFlow Lite with other frameworks and deep-learning examples on our SD-image
The TensorFlow Lite versions support NEON and XNNPack.
Please notice the provided wheels are not suitable for C++ programming!
Only for the Python3 installation of TensorFlow Lite on your machine.
TensorFlow Lite, intended for mobile devices, does not support CUDA.
You will have the regular aarch64 installation on your Nano.