puneet makhija
11/30/2023, 1:44 PMAppPipeline
by using app_pipeline
factory
but how we can register this ?Juan Luis
11/30/2023, 2:08 PMJuan Luis
11/30/2023, 2:09 PMpuneet makhija
11/30/2023, 2:23 PMTakieddine Kadiri
12/01/2023, 1:47 PMpipeline_registry.py
ex:
from kedro.pipeline.modular_pipeline import pipeline
from kedro_boot.pipeline import app_pipeline
your_first_pipeline = pipeline([node(your_function, inputs=["input_dataset", "model"], outputs="output_dataset")])
your_second_pipeline = pipeline([node(your_function, inputs=["input_dataset", "model"], outputs="output_dataset")])
your_first_app_pipeline = app_pipeline(
your_first_pipeline,
name="your_first_pipeline",
inputs="input_dataset", # if you want to inject input dataset to your pipeline
outputs="output_dataset" # to get output from pipeline run
)
your_second_app_pipeline = app_pipeline(
your_second_pipeline,
name="your_second_pipeline",
)
return {"_default_": your_first_app_pipeline + your_second_app_pipeline}
You can flollow the pipeline serving example in the kedro boot examples. And do slightly the same things in your kedro projects.
In the following weeks we'll write a documentation site and plan also to introduce a user friendly interface with an integrated a web server to streamline the pipeline serving process.
let us know if it worked for you