CNN forward function , AutoTuning the number of layers

Question

class ConvolutionalNetwork(nn.Module):
    def __init__(self, in_features, trial):
        # we optimize the number of layers, hidden units and dropout ratio in each