Is there a way in kedro to define or register parameters, like a logger, that all parameters and nodes can get access to, s.a. for the datasets from the catalog file?

Could you rephrase/elaborate more on this question? What is the intent here?

Certainly, in my case, I initialise a weights and biases experiment tracker:

import wandb
run = wandb.init(project="visualize-sklearn")

I then want to pass around this initialised “run” parameter to the nodes without having to initialise it again. Would the simplest method be to define it in as a argument in

conf/base

and add a custom resolver for OmegaConf, or is there a more convenient method?

You have at least 3 options: 1. Initialize the

run

in the first node, then just pass it as an explicit input to every other node that would need to do sth with W&B 2. Initialize the

run

in the hook (e.g.

before_pipeline_run

) and pass it implicitly to every node with

before_node_run

implemented in the same hook class https://docs.kedro.org/en/stable/hooks/introduction.html). 3. Initialize the

run

in the hook (again,

before_pipeline_run

) but do not pass it to nodes and just use

wandb.run

as it’s a global singleton https://docs.wandb.ai/guides/runs#create-a-run There are probably a few other options too