About This Task

You can edit a development pipeline while it remains deployed. You can make the following changes:

• Update query pipeline components, except for the embedding model and its format.
• Update the index a pipeline is using.
• Update the pipeline's name.

You can't modify or add and remove components, or change their order without undeploying the pipeline first.

📘

Production Pipelines

Editing production pipelines is forbidden. You must change the pipeline's service level first.

Prerequisites

• You must be an Admin to perform this task.
• If the changes you want to make require undeploying the pipeline, go to Pipelines, click the more actions button next to the pipeline name, and select Undeploy.
• If you're updating the index the pipeline uses, make sure the index is enabled. For instructions, see Enable an Index.
• To edit pipelines using API endpoints, generate an API Key first.

Edit a Pipeline in Pipeline Builder

This is the default way of editing pipelines. It opens the pipeline in the visual editor. You can use it to replace one component with another or update the parameters quickly.

1. Log in to deepset AI Platform and go to Pipelines.
2. Find the pipeline you want to edit, click the More Actions menu next to it, and choose Edit.
   

ℹ️

Renaming a Pipeline

If you just want to rename a pipeline, you can do it from the More Actions menu, without opening the pipeline in Builder. Click More Actions next to a pipeline and choose Rename.

3. Modify the pipeline and save it.

ℹ️

Update Index

To update an index a pipeline is using, find the Document Store that uses the index you want to change and choose a different index.

3. Deploy the pipeline to use it for search.

Edit a Pipeline with REST API

There are two endpoints you can use to update your pipeline: Update Pipeline and Update Pipeline Yaml.

Update Pipeline YAML

This method lets you quickly upload an updated pipeline YAML to deepset. It's best to change the pipeline configuration and have a pipeline YAML file ready. It uses the Update Pipeline Yaml endpoint.

See this recipe for a step-by-step explanation of the code:

Or copy this code:

curl --request PUT \
     --url https://api.cloud.deepset.ai/api/v1/workspaces/<YOUR_WORKSPACE>/pipelines/<YOUR_PIPELINE_NAME>/yaml \
     --header 'accept: application/json'
     --authorization: Bearer <API_KEY>

import requests

url = "https://api.cloud.deepset.ai/api/v1/workspaces/<WORKSPACE_NAME>/pipelines/<PIPELINE_NAME>"

headers = {
    "accept": "application/json",
    "content-type": "application/json",
    "authorization": "Bearer api_eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiJlNGQyM2MwZS02NTljLTRhNGYtOGUzMC1lODUzYzcxY2E2ZTh8NjI0NTYxMzAxNTQ1ZTYwMDcwYTIyNjRjIiwiZXhwIjoxNzQ4NjQyNDAwLCJhdWQiOlsiaHR0cHM6Ly9hcGkuY2xvdWQuZGVlcHNldC5haSJdfQ.94t9fAQ1Adr5MzM67wEP2FCD62rJrUQePc1y6SLS7_E"
}

response = requests.patch(url, headers=headers)

print(response.text)

Example

This is a script that updates the YAML of a pipeline called RAG in a workspace named default:

import requests # to send HTTP requests
import yaml # to read and parse YAML configuration files

# the endpoint where the pipeline called RAG in a workspace called default will be updated
url = "https://api.cloud.deepset.ai/api/v1/workspaces/default/pipelines/RAG/yaml"

# this loads the queryPipeline.yaml file that contains the updated configuration
with open('PATH/queryPipeline.yaml', 'r') as file:
    query_yaml_data = yaml.safe_load(file)

# convert the parsed YAML to YAML-formatted strings ready to be sent in a request body
query_yaml_string = yaml.dump(query_yaml_data)

# add pipeline definitions
payload = {
    "query_yaml": query_yaml_string
}
# add the request metadata
headers = {
    "accept": "application/json",
    "content-type": "application/json",
    "authorization": "Bearer api_XXXX"
}
# send the request and print the response
response = requests.put(url, json=payload, headers=headers)

print(response.text)

Update Pipeline Settings

Use the Update Pipeline endpoint if you want to change pipeline settings, such as name or service level, without changing the YAML configuration.

Open the recipe for a walkthrough of the request code:

Copy this code and modify it as needed:

curl --request PATCH \
     --url https://api.cloud.deepset.ai/api/v1/<YOUR_WORKSPACE>/pipelines/<YOUR_PIPELINE_NAME> \
     --header 'accept: application/json' \
     --header 'content-type: application/json' \
     --data '
{
  "name": "new_pipeline_name",
  "service_level": "DEVELOPMENT",
  "status": 1
}
'