Softmax label smoothing in gluon

Hi there,

I would really like to use label smoothing in my mxnet gluon training, but I cannot figure out how i should configure the SoftmaxCrossEntropyLoss (SoftmaxCELoss api docs) class to use this.

I can see that label smoothing is included in the symbol api of the class SoftmaxOutput (SoftmaxOutput api docs).

Is it possible to use this with gluon?