Requirements for using custom components in your models

You can define your own transformers, estimators, functions, classes, and tensor operations in models you deploy in IBM Watson Machine Learning as online deployments. This topic lists requirements for using custom components in your models.

Defining and using custom components

To use custom components in your models, you need to package your custom components in a Python distribution package .

See: Packaging Python Projects

Package requirements

• The package type must be: source distribution (distributions of type Wheel and Egg are not supported)

• The package file format must be: .zip

• Any third-party dependencies for your custom components must be installable by pip and must be passed to the install_requires argument of the setup function of the setuptools library

See: Creating a source distribution

Storing your custom package

You must take extra steps when storing your trained model in the Watson Machine Learning repository:

1. Store your custom package in the Watson Machine Learning repository (using the runtimes.store_library function from the Watson Machine Learning Python client, or the store libraries Watson Machine Learning CLI command.)
2. Create a runtime resource object that references your stored custom package, and then store the runtime resource object in the Watson Machine Learning repository (using the runtimes.store function, or the store runtimes command.)
3. When you store your trained model in the Watson Machine Learning repository, reference your stored runtime resource in meta data passed to the store_model function (or the store command.)

Examples

• Using custom components in a scikit-learn model
• Using custom components in a TensorFlow model

Required Watson Machine Learning interface versions

• Command Line Interface (CLI) : 2.1.0

• Watson Machine Learning Python client : 1.0.293

Supported frameworks

For support details, listed by framework, see: Supported frameworks