AdversarialFairnessRegressor - incorrect input size

[parasurama]

Describe the bug

AdversarialFairnessRegressor is throwing an incorrect input size warning.
The input size is (batch_size, n_samples) rather than (batch_size, n_features)

Steps/Code to Reproduce

from fairlearn.adversarial import AdversarialFairnessRegressor
import numpy as np

X = np.random.rand(10000, 10)
y = np.random.rand(10000)
z = np.random.binomial(1, 0.5, 10000)


print(X.shape, y.shape, z.shape)

mitigator = AdversarialFairnessRegressor(backend="torch",
                                         progress_updates=0.5,
                                         batch_size=32)

mitigator.fit(X, y, sensitive_features=z)

Expected Results

Input size should be (batch_size, n_features), not (batch_size, n_samples)

Actual Results

UserWarning: Using a target size (torch.Size([32, 1])) that is different to the input size (torch.Size([32, 10000])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size. 

Screenshots

Versions

• OS: MacOS
• Python version: 3.11.4
• Fairlearn version:
• fairlearn: 0.10.0
  Cython: None
  lightgbm: 4.3.0
  matplotlib: 3.7.1
  numpy: 1.26.4
  pandas: 2.2.2
  pip: 23.2.1
  pytorch: None
  scipy: 1.10.1
  setuptools: 69.5.1
  sklearn: None
  tensorflow: 2.16.1

[parasurama]

Note that AdversarialFairnessClassifier doesn't throw this warning

[adrinjalali]

@SeanMcCarren would you be able to have a look?

[parasurama]

Looks like there is already an open pull request for this issue #1333

[SeanMcCarren]

@SeanMcCarren would you be able to have a look?

Yes, had a look, indeed as @parasurama mentioned it's fixed by #1333