About This Task

When importing through Pipeline Builder, you simply paste the pipeline YAML into the YAML editor. There's no validation of the pipeline.

You can also import using the deepset SDK. It offers more options, such as YAML validation or pipeline input and output configuration. For instructions, see Import a Haystack Pipeline.

Prerequisites

You must have a Haystack pipeline in the YAML format. You can use the following code to serialize your pipeline:

from haystack import Pipeline

pipeline = Pipeline()
with open("/path/to/save/the/pipeline.yaml", "w") as file:
    pipeline.dump(file)

For more information, see Haystack documentation.

Import Your Pipeline

1. Log in to deepset AI Platform and click Pipeline Templates in the navigation.

2. On the Templates page, choose Create empty pipeline.

   

3. Give your pipeline a name and click Create Pipeline. You're redirected to Pipeline Builder.

4. Switch to the YAML view using the code switch.

   

5. Copy your pipeline YAML, and paste it in the code editor.

6. Use the code switch to change to the canvas view and see your pipeline visualization.

What's Next

• To deploy your pipeline in deepset, add the input and output components. For details, see Inputs and Outputs.

• Once you're done updating your pipeline, you can export it to YAML or Python using the Export code button.