Michael Schuster

Closing this as we now use standard python type annotations for steps.

Hey @adamwawrzynski, sorry for the super late reply. We've been internally rethinking how we will go about maintaining our examples which also affects this PR: In the future, we will...

@kabinja One little detail for you that will be quite helpful: [This](https://github.com/zenml-io/zenml/blob/d7e58bcd65bfac25a9047e959831a5f0d7c3db9d/src/zenml/config/pipeline_run_configuration.py#L27) is the pydantic class that defines the structure of the yaml configuration file.

I also like the suggestion, seems like a good compromise 👍 RBAC will control which users have permissions to update the stack component configuration but will not affect the `Settings`...

Just one more thing that came to my mind: We should probably also rename the `llm-lora-finetuning` template right?

@wjayesh I see that you modified the workflow that gets dispatched to only create a tag. Wouldn't it be easier to just create a tag in that repo directly, or...

I think it's still weird to do it to be honest, if we ever needed this again we could just copy-paste this from stackoverflow, no need to keep that code...

The underlying problem is not the input validation but rather that passing collections of artifacts (=outputs of other steps) is currently not supported. It's on our list of todos in...

> why don't you use zenml.io.Serializable to wrap custom types Because it doesn't exist?

> > PipelineDefinition.create_run_template > > When `create_run_template` function is called, it is already within the context manager of `temporary_active_stack`: > > https://github.com/zenml-io/zenml/blob/0d059d3a54d7ab86b2e3057319fb75f2e6c25451/src/zenml/cli/pipeline.py#L363 > > I can also add a context...