Can't deepcopy or load a NeuralProphet instance (combination of save & load methods throws an exception)

[almostintuitive]

Prerequisites

• Put an X between the brackets on this line if you have done all of the following:
  • Reproduced the problem in a new virtualenv with only neuralprophet installed, directly from github:

    git clone <copied link from github>
    cd neural_prophet
    pip install .

  • Checked the Answered Questions on the Github Disscussion board: https://github.com/ourownstory/neural_prophet/discussions
    If you have the same question but the Answer does not solve your issue, please continue the conversation there.
  • Checked that your issue isn't already filed: https://github.com/ourownstory/neural_prophet/issues
    If you have the same issue but there is a twist to your situation, please add an explanation there.
  • Considered whether your bug might actually be solveable by getting a question answered:
    • Please post a package use question
    • Please post a forecasting best practice question
    • Please post an idea or feedback

Describe the bug

deepcopy() throws an exception, a manual implementation also fails with the following error message:
AttributeError: 'NeuralProphet' object has no attribute 'early_stopping'

To Reproduce

Steps to reproduce the behavior:

1. Either call deepcopy() on a fitted model, results in RuntimeError: Only Tensors created explicitly by the user (graph leaves) support the deepcopy protocol at the moment

OR

2. Run the following code:

from io import BytesIO
from neuralprophet.utils import load, save
buff = BytesIO()
save(self.model, buff)
buff.seek(0)
model = load(buff)

Results in : AttributeError: 'NeuralProphet' object has no attribute 'early_stopping'

Expected behavior

Able to deepcopy a NeuralProphet instance, one way or another.

What actually happens

When deepcopy is called, we get an exception:
RuntimeError: Only Tensors created explicitly by the user (graph leaves) support the deepcopy protocol at the moment

When we save & load the object, we get the exception:
AttributeError: 'NeuralProphet' object has no attribute 'early_stopping'

Screenshots

If applicable, add screenshots and console printouts to help explain your problem.

Environement (please complete the following information):

Python 3.9.15, in standalone venv with no other packages
NeuralProphet 0.5.2, installed from PYPI with pip install neuralprophet

Additional context

We want to integrate NeuralProphet into our time series cross-validation library called fold. We have written a wrapper that fails during a mandatory call to deepcopy() that we have to enable paralellism and safety.
dream-faster/fold-models#17

A github Action log failing with one of the exceptions can be found here:
https://github.com/dream-faster/fold-models/actions/runs/4356218584/jobs/7613915633

[leoniewgnr]

Hi @almostintuitive, thanks for reporting this!
Are you currently working yourself on this or do you need our help?
Because we are planning to close all open issues soon, so just want to check in what's the status on this.
Let me know if you need help:)

[leoniewgnr]

@almostintuitive isn't this tested by the pytest def test_save_load() in https://github.com/ourownstory/neural_prophet/blob/main/tests/test_utils.py?

[leoniewgnr]

closed due to inactivity

[almostintuitive]

@leoniewgnr I believe this is happening when we're trying to deepcopy OR save/load an unfitted model, before fitting begins. we do this so we don't end up mutating a shared object (between threads, for example).
would it be useful if I contributed a (failing) unit test demonstrating this?