@Marvin I have a flow that accepts

pydantic.BaseModel

configuration class that can be created from

yaml.safe_load(yaml_file)

. After deploying the flow to prefect server, the deployment can be run by passing a json configuration. How do I define a default value for this JSON when deploying this flow so that the UI provides default value for this object then run from the UI?

To define a default JSON value for a parameter in a Prefect deployment, you can utilize the

prefect.yaml

file. This file allows you to set base settings for deployments, including default parameter values. Here's a general approach to achieve this: 1. Initialize Deployment Configuration: Use the Prefect command line interface to initialize your deployment configuration. This will generate a

prefect.yaml

file in your project directory.

bash
   prefect init

2. Edit the

prefect.yaml

File: In the

prefect.yaml

file, you can define default parameter values under the

parameters

section. This will ensure that when the deployment is run from the UI, the specified default values are used unless overridden. Here's an example structure for the

prefect.yaml

file:

yaml
   deployments:
     - name: my-flow-deployment
       parameters:
         my_config: 
           key1: default_value1
           key2: default_value2

3. Deploy with Defaults: When you deploy your flow using the Prefect CLI, it will use the

prefect.yaml

file to set up the deployment with the specified default parameters. For more detailed information about configuring deployments with YAML, you can refer to the Prefect documentation. This approach will allow you to provide default JSON values for your parameters when running the deployment from the UI.

Is there a way to do this directly from the python flow code? For example, the code below works in prefilling the UI's string argument field

name

with

Marvin

on UI:

@flow
def hello(name: str = "Marvin"):

yeah you can just say like

@flow
def f(foo: FooModel = FooModel())

tried it, doesn't work

can you elaborate on what you tried / what you were expecting / what you saw? it'd be a surprising bug if that doesnt work in general

my python code is like this:

parameters = { 'field_1': 123, .... }

@flow
def f(foo: FooModel = FooModel(**parameters))

I expect the UI to pre-fill field

field_1

to

123

but it's not

hmm on

main

im seeing it work

from pydantic import BaseModel
from prefect import flow


class FooModel(BaseModel):
    name: str
    age: int


@flow
def foo(model: FooModel = FooModel(name="John", age=30)):
    print(model)


if __name__ == "__main__":
    foo.serve()

so if you could share your version and an MRE that'd be appreciated

i see