hidden_layer_sizes vs hidden_layer_size?

[PapadopoulosDimitrios]

print(imodelsx.KANRegressor(
hidden_layer_sizes=[32,64],
hidden_layer_size=36,
regularize_activation=1.0,
regularize_entropy=1.0
))

trying out this code in a Jupyter notebook I get the following error: AttributeError: 'KANRegressor' object has no attribute 'hidden_layer_size'

Same thing happens if I use hidden_layer_size=32. Any insights?

[csinva]

Hm what is your imodelsx version? Can you maybe try pip install --upgrade imodelsx and then rerunning?

[PapadopoulosDimitrios]

The version is 1.0.7 . I am using google colab so every time I start the environment anew.

The code is basically from the posted example:

X, y = make_regression(n_samples=5000, n_features=5, n_informative=3)

model_KAN = imodelsx.KANRegressor(
hidden_layer_sizes=[32,64],
regularize_activation=1.0,
regularize_entropy=1.0
)

print(model_KAN)

model_KAN.fit(X, y)
y_pred = model_KAN.predict(X)

test_correlation = np.corrcoef(y, y_pred.flatten())[0, 1]
print('Test correlation:', test_correlation)

[Syed-Tangim-Pasha]

I faced the same problem with this library.