TypeError: UnitScaler() takes no arguments

[ajaynr]

Trying to run the Minimum Example, I am thrown this error. I am working with v1.2.1 installed using pip with python v3.9.

Detailed error message:

>>> from ConfigSpace import ConfigurationSpace
>>> cs = ConfigurationSpace({
...     "myfloat": (0.1, 1.5),                # Uniform Float
...     "myint": (2, 10),                     # Uniform Integer
...     "species": ["mouse", "cat", "dog"],   # Categorical
... })
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "$HOME/.conda/envs/gpt/lib/python3.9/site-packages/ConfigSpace/configuration_space.py", line 205, in __init__
    hyperparameters = list(_parse_hyperparameters_from_dict(space))
  File "$HOME/.conda/envs/gpt/lib/python3.9/site-packages/ConfigSpace/configuration_space.py", line 103, in _parse_hyperparameters_from_dict
    yield UniformFloatHyperparameter(name, lower, upper)
  File "$HOME/.conda/envs/gpt/lib/python3.9/site-packages/ConfigSpace/hyperparameters/uniform_float.py", line 98, in __init__
    scaler = UnitScaler(f64(self.lower), f64(self.upper), log=log, dtype=f64)
TypeError: UnitScaler() takes no arguments

This is my first time using ConfigSpace, so I apologize in advance in case there is any naive mistake from my end!
Any help to resolve this would be appreciated. Thanks!

[eddiebergman]

I'm not sure why you're getting this error, I tried to reproduce it but it all works as intended:

~/code/repro » cat repro.py                                                                                      eddiebergman@mllap12-2
from ConfigSpace import ConfigurationSpace

cs = ConfigurationSpace(
    name="myspace",
    space={
        "a": (0.1, 1.5),  # UniformFloat
        "b": (2, 10),  # UniformInt
        "c": ["mouse", "cat", "dog"],  # Categorical
    },
)

configs = cs.sample_configuration(2)
print(configs)

(repro) --------------------------------------------------------------------------------------------------------------------------------
~/code/repro » python repro.py                                                                                   eddiebergman@mllap12-2
[Configuration(values={
  'a': 1.4472818725845,
  'b': 7,
  'c': np.str_('dog'),
}), Configuration(values={
  'a': 0.1622273472625,
  'b': 3,
  'c': np.str_('dog'),
})]

(repro) --------------------------------------------------------------------------------------------------------------------------------
~/code/repro » uv pip list                                                                                       eddiebergman@mllap12-2
Package           Version
----------------- -------
configspace       1.2.1
more-itertools    10.6.0
numpy             2.0.2
pyparsing         3.2.1
scipy             1.13.1
typing-extensions 4.12.2

[eddiebergman]

Can you provide me with your OS as well?

Judging by the $HOME in the traceback, I'm assuming it's either Linux or Mac, both of which I've tested personally.

---

Can you run this in your environment for me?

python -c "from ConfigSpace.hyperparameters.hp_components import UnitScaler; print(UnitScaler(0, 1, dtype='blah'))"

You can also run it in a standalone file if you prefer.

If that doesn't work then:

• Create a clean new environment with nothing in it.
• Install ConfigSpace into it.
• Try again.

If the error persists, then I really have no clue why this would be happening.

---

You could also try:

git clone git@github.com:automl/ConfigSpace.git
pip install -e ConfigSpace
python -c "from ConfigSpace.hyperparameters.hp_components import UnitScaler; print(UnitScaler(0, 1, dtype='blah'))"

We would need more information to go on because that's a very odd error.

[ajaynr]

Hi eddie,

You are right about the OS, it was indeed Linux.

Although the error still persists in my environment, creating a clean new environment with nothing in it solved it for me. I now need to figure out what exactly was triggering the issue in my previous environment.

Thank you for the response and suggestions!

[LGro]

I see the same, reproducible on Ubuntu 20.04 with:

conda create -n cs-debug python=3.9.7
conda activate cs-debug
pip install ConfigSpace==1.2.1
python -c "from ConfigSpace.hyperparameters.hp_components import UnitScaler; print(UnitScaler(0, 1, dtype='blah'))"

fails with TypeError: UnitScaler() takes no arguments

EDIT: Works with Python 3.12.9