Skip to main content
Version: 4.0

Developing Third-party Algorithm Model

You can upload third-party algorithm models to X-BD, and manage, run and call the model with supOS object data on the platform.

Developing Algorithm Model​

A model package should include a folder named backend containing a model algorithm file runtime.py and a dependency file requirements.text, and a model description file manifest.json.

runtime.py​

  • Only 3 entry arguments
    • inputData: 2-dimensional array, corresponds to input variables.
    • dataStruct: Defined data structure.
    • config: Custom model configuration.
  • Must contain a whole process function defined as algorithmrun

  • Only one return argument: outputdata.

info

If there are other referenced projects or model files, you can either reference in runtime.py or put them together with runtime.py in the same folder backend.

requirements.txt​

For the following dependencies, the requirements file content should be: cvxopt==1.3.0 pipdeptree==2.2.1

manifest.json​

ParameterDescription
modelName
  • Must be unique.
  • Cannot end with a number.
  • Cannot exceed 50 in length.
  • Update with the same name.
  • Required.
descriptionCannot exceed 500 in length.
devLanguagePython
versionMust be unique. Required.
versionFeaturesFeatures of the current version. Cannot exceed 500 in length.
runEnvPython. Required.
runEnvVersionPython version: 3.7.1. Required.
runtimeRunClassruntime.py. Required.
inputStructsInput data structure.
[{"name":"temperature","dataType":"Double","description":" temperature"},
{"name":"","dataType":"Double","description":" pressure"}]
outputStructsOutput data structure.
[{"name":"heat","dataType":"Double","description":" heat"}]
configCustom configuration.

Uploading Algorithm Model​

  1. Log in to supOS, and then select Data Intelligence > Data Science > Model Management.
  2. Click Import model, select the model file from local and then click OK.