Closed Priyanshu-bee closed 1 year ago
Could you please provide a Minimum Working Example (MWE) to reproduce the error? A plus would be the working scenario.
Thank you
from tensorflow.keras import datasets, models import cvnn.layers as complex_layers
model = models.Sequential() model.add(complex_layers.ComplexConv2D(32, (3, 3), activation='cart_relu', input_shape=(32, 32, 3), dtype=np.float32))
running the above is causing the given error, it is working when replacing activation with just 'relu'. As far as I could understand while defining Complex conv layers you have imported activations from tensorflow.keras and then used activations.get(activation) which might be raising an issue as these activations are defined and created by you and importing t_activations might enable us to use these complex activations.
Try updating cvnn version to the latest.
Also, beware that the example cast to np.float32 and not complex.
I used conda to install cvnn, that was giving the error. After doing pip install it works, thanks!!!
getting this error while trying to use complex activation functions with ComplexConv layers. Doesn't give any error while using the complex activations with the complex dense layer....