AMCT (TensorFlow, Ascend)

OS for the online inference environment

For details, see Supported OSs.

Download the required software release from the Ubuntu official website and install it. Run the following command to query the installation:

cat /etc/*release && uname -m

Quantization on the NPU or CPU is supported. Quantization on the GPU is not supported.

Python

Python3.7.x, Python3.8.x, Python3.9.x

Python 3.9.2 is recommended.

For details, see Appendix A: Reference > Dependencies in CANN Software Installation.

To set up the Python 3.9.2 environment, do as follows:

• For Ubuntu OS, see Python 3.9.2 Installation on Ubuntu.
• For EulerOS, see Python3.9.2 Installation on EulerOS.

This document uses Python 3.9.2 as an example. The environment variables and installation commands are subject to the actual Python version.

CPU version of TensorFlow

2.6.5 or 1.15.0

For details, see Installing Dependencies.

The fake-quantized model cannot run on the GPU. Therefore, you only need to install the CPU version.

Mapping between TensorFlow and Python versions:

• Python 3.9.2 is recommended for TensorFlow 2.6.5.
• Only Python 3.7.5 can be used for TensorFlow 1.15.0.

Pillow

6.0.0 or later

For details, see Installing Dependencies.

-