Documentation

Operator Development Dependencies

Before developing an operator, ensure that the following dependencies have been installed in the Python environment where MindStudio IDE runs:

If you run the following commands as a non-root user, add --user to the end of each installation command, for example, pip3 install xlrd==1.2.0 --user.
Table 1 Python dependencies required for MindStudio IDE operator development

Dependency

Required Version

Application Scenario

Reference Installation Method (Python 3 in the x86 Architecture Is Used as an Example)

xlrd

1.2.0

Operator project creation

 
pip3 install xlrd==1.2.0

gnureadline

-

TIK development

 
pip3 install gnureadline

mindspore

-

MindSpore operator development

For details about the installation method, see the MindSpore Installation Guide.

absl-py

-

UT

 
pip3 install absl-py

coverage

-

 
pip3 install coverage

jinja2

3.1.2

 
pip3 install jinja2

onnx

-

ST

 
pip3 install onnx

tensorflow

-

 
pip3 install tensorflow

csv

-

 
pip3 install python-csv

google

-

 
pip3 install google
  • The dependencies may fail to be installed in an AArch64 OS. In this case, use the source code to build and install the dependencies or refer to the official installation guides of corresponding dependencies.
  • TensorFlow 1.15.0 is recommended.
Parent topic: Operator Development