Version pipelines
Understanding how and when the version of a pipeline is incremented.
This is an older version of the ZenML documentation. To read and view the latest version please visit this up-to-date URL.
Version pipelines
You might have noticed that when you run a pipeline in ZenML with the same name, but with different steps, it creates a new version of the pipeline. Consider our example pipeline:
Running this the first time will create a single run
for version 1
of the pipeline called first_pipeline
.
Running it again (python run.py
) will create yet another run
for version 1
of the pipeline called first_pipeline
. So now the same pipeline has two runs. You can also verify this in the dashboard.
However, now let's change the pipeline configuration itself. You can do this by modifying the step connections within the @pipeline
function or by replacing a concrete step with another one. For example, let's create an alternative step called digits_data_loader
which loads a different dataset.
This will now create a single run
for version 2
of the pipeline called first_pipeline
.
Last updated