ZenML 0.5.5 is jam-packed with new features to take your ML pipelines to the next level. In this blog post we will highlight our three biggest new features: Kubeflow Pipelines, CLI support for our integrations and Standard Interfaces. That’s right, Standard Interfaces are back!
For a detailed look at what’s changed, give our full release notes a glance.
We are super excited about our integration of Kubeflow Pipelines into ZenML. With just a few lines of code you can now spin up your very own local deployment of Kubeflow Pipelines. With Kubeflow Pipelines running on your machine or even in the cloud, you change where to run your code with just a few commands.
With release 0.5.5 we made our integrations into the first class citizens they should be. You can now easily list all integrations and see which integrations are active (by having all their requirements installed).
zenml integration list
But that’s not all! Installing requirements for our integrations is just as easy now. For example, this is how you would
install all requirements to use our
zenml integration get-requirements dash
Standardization is a great way to keep code maintainable, easy to use and scalable across larger teams. With our Standard Interfaces we are making our steps, pipelines and artifacts even more powerful.
pipeline_instance = TrainingPipeline( datasource=PandasDatasource(), splitter=SklearnSplitter(), analyzer=PandasAnalyzer(), preprocesser=SklearnStandardScaler(), trainer=TensorflowBinaryClassifier(), evaluator=SklearnEvaluator() ).with_config('pipeline_config.yaml') pipeline_instance.run()
Using a powerful set of standardized steps like this, it becomes easier than ever to hit the ground running when setting up a new pipeline. Check out how get started with our Standard Interfaces in our docs
Join our Slack to let us know what you think we should build next!