I want to train a WGAN where the conv layers in the critic are only allowed to have non-negative weights (for a technical reason). The biases, nonetheless, can take both +/- values. There is no constraint on the generator weights. I did a toy experiment on MNIST and observed that the performance is significantly worse than a regular WGAN. Any insights are welcome. Can you suggest some architectural modifications so that the nonnegativity constraint doesn’t severely impair the model capacity?